B

Backpropagation of Error Barriers to Organizational

Algorithm Learning
▶ Connectionist Theories of Learning DENIS FISCHBACHER-SMITH, MOIRA FISCHBACHER-SMITH
Department of Management, University of Glasgow
Business School, Glasgow, Scotland, UK

Backward Inhibitory Synonyms

Conditioning
▶ Pain-relief Learning
Balance Theory
Balance theory is a motivational theory of attitude
change proposed by Fritz Heider (1958) which concep-
tualizes the consistency motive as a drive toward psy-
chological balance. Heider proposes that “sentiment”
or liking relationships are balanced if the affect valence
in a system multiplies out to a positive result.
References
Heider, F. (1958). The psychology of interpersonal relations. New York:
Wiley.
Bands
▶ Instrumental Learning in Music Education
Banking Education
▶ Reproductive Learning
N. Seel (ed.), Encyclopedia of the Sciences of Learning, DOI 10.1007/978-1-4419-1428-6,
# Springer Science+Business Media, LLC 2012
Crisis incubation
Definition
Barriers to organizational learning are seen as those
systems and behaviors that prevent or inhibit organi-
zations from adapting to the main decision-making
challenges that they face. They can also arise from the
processes of identifying and adopting new behaviors
and practices in light of successes and failures. These
barriers can occur at multiple levels within and between
organizations, and involve both individual and group
processes and behaviors.
Theoretical Background
The manner in which organizations fail to learn lessons
from both internal and external “sources” has been an
issue that lies at the core of a number of debates within
“management” theory and practice (Argyris 1990;
Senge 1990; Toft and Reynolds 1994). General theories
of management have tended to focus on the positive
role of learning as a means of enhancing organizational
performance, and have advocated learning across
organizational settings. However, there are inevitable
problems in simply taking a successful approach from
one organization/context and applying it elsewhere,
without grounding that learning within the cultural,
spatial, and temporal settings in which that learning is
believed both to have taken place and subsequently to
apply. Simplistic approaches to learning – termed
“template management” – are significant in shaping
and limiting an organization’s abilities to learn and
408 B Barriers to Organizational Learning
may generate considerable barriers to appropriately
defining, framing, and addressing the problems from
which learning is required. In the same way that learn-
ing is associated with improved performance, failure to
learn is associated with diminished performance.
Research has identified a series of barriers to learn-
ing (Smith and Elliott 2007) that appear to have valid-
ity in explaining the processes by which organizational
failures occur and are “incubated.” Several key barriers
to organizational learning have been outlined within
the literature (for a summary see, Fischbacher-Smith
and Fischbacher-Smith 2009; Smith and Elliott 2007).
A common barrier is generated by the core beliefs values
and assumptions held by decision-makers about the
organization and its operating environment and core
processes. These beliefs, values, and assumptions serve
to shape both the awareness and acknowledgment of
issues facing the organization along with the
approaches that are taken toward addressing them. In
relation to corporate social responsibility (CSR), for
example, a central argument is that the “ethical”
approach taken by the organization (or possibly, the
lack of an appropriate ethical stance) will frame the
decisions of managers in relation to how they tackle
considerations around risk and uncertainty. Organiza-
tional culture, and in particular dominant cultural
norms within the organization, is an important ele-
ment in this framing process and is linked to the
power of technical and expert elites within the deci-
sion-making process. Learning becomes inhibited
when the apparent lessons of an adverse event seem to
run counter to the dominant paradigmatic view of key
members of the organization. Denial of the potential
for adverse events – “it couldn’t happen here” – rests on
the often unspoken, but closely held misconception
about the values held by organizational members and
those of the wider society and, can serve to be a key
barrier to learning.
A related process concerns the manner in which the
organization communicates the extent of its uncer-
tainty surrounding the decisions that it takes. The
failure to deal with information difficulties and the
likelihood of such difficulties helping to produce inef-
fective communication, are shaped by, and reinforce the
cultural dynamics that prevail within the organization
and can further exacerbate the potential for failure.
One obvious problem in this regard is the manner in
which an organization deals with risk – particularly for
rare (often termed, extreme) events. Given their
low frequency of occurrence and potential for serious
consequences, extreme events create a set of problems
for decision-makers who have no historical precedent
with which to ground their sense making around the
problem. Such events also inhibit the ability of the
organization to communicate the nature of the issue
and this may further escalate the problem. Organiza-
tions that constrain communication (both in terms of
volume and content) may also fail to pick up on early
warnings and near-miss events and this becomes
compounded by the temporal constraints that are
often imposed upon decision making. By narrowing
these temporal boundaries, organizations will discount
the potential for long-term impacts associated with their
decisions. An incremental approach to decision making
may ensure that the organization is incapable of dealing
with both emergence and interaction effects that increas-
ingly arise in complex, multi-locational organizational
settings. The greater the interactions between elements of
the organization, the greater the potential that exists for
the generation of emergent properties associated
with such interactions. It is this emergence that can
serve to bypass organizational controls and constrain
the effectiveness of learning from near-miss events.
The assumptions that are made around the nature
and extent of such systems’ perturbations will
determine the controls that are put in place to deal
with fluctuations in organizational performance. By
constraining the “worst-case scenario” for such events,
organizations will inhibit the abilities of decision-
makers to test the limits of their assumptions associ-
ated with systems’ performance. When the operating
environment moves from a typically ordered form into
a complex one, these constraints will inhibit the deci-
sion performance of management through processes of
cognitive narrowing and they will become increasingly
unable to recognize and absorb the full range of salient
information, and will unduly narrow their focus onto
constrained, and deceptively manageable, tasks. Such
narrowing will be enhanced further as the task
demands associated with such environmental shifts
increase. All too often, organizations fail to learn
from such experiences by reconstructing the event
narrative and blame projection after a crisis, doing
so in a way that sees the failure as a function of
the performance of an individual rather than the
“system.”

Such limitations derive, in part, from the difficulties
of foresight and hindsight. Within the literature on
crisis management and its related research areas, for
example, failure to learn has been seen as a key element
in the process of crisis incubation for many of the
reasons highlighted above. Turner (Turner 1978) also
highlighted the potential for failure within organiza-
tions due to a failure of foresight, whereby managers
were unable to anticipate and prevent failures arising
from their actions and the core processes of the orga-
nization. Subsequent research extended Turner’s initial
research to focus on the failures of hindsight (Toft and
Reynolds 1994) where lessons were not learned due to
the difficulties of post hoc interpretations of events and
processes, or in other words, challenges associated with
the often rapid, constrained, and context-specific pro-
cesses of sense making that take place during crises.
A further, and final, barrier to learning concerns the
abilities that managers and management teams have to
cope with environmental shifts and the manner in
which they “make sense” of these changing dynamics
while the organization itself may also be undergoing
a process of change. An inability to develop the adap-
tive capabilities (often termed “dynamic capabilities”)
to cope with such phase transitions can also have
a profound impact on the processes of learning and
may, in turn, be undermined by the wider barriers to
learning that exist within the organization. At the core
of this discussion is the manner in which organizations
are able to deal with uncertainty and the limitations to
predictive validity within decision-making processes,
and to then manage their responses accordingly.
Important Scientific Research and
Open Questions
It can be argued that the learning process is central to
organizational performance and to mitigating poten-
tial crisis incubation. Essential to embedding learning
within organizations is ensuring that cultural values
and behaviors endorse learning and reflection and the
avoidance of blame, as well as ensuring that decision-
making processes and the behavior of the organization
as a system are understood and evaluated. Also impor-
tant are the interfaces between elements of the organi-
zation and indeed between organizations. While this
section has tackled learning within organizations, the
networked structure of business behooves organiza-
tions to consider learning across organizations through
Basal Ganglia B 409
both formal and informal processes. Also of signifi-
cance are the dynamic capabilities within – and
between – organizations to identify and respond to B
weak signals and to cope with the uncertainties and
pressures around decision making within fluid envi-
ronments. It is within these territories that the barriers
to learning arise, and therefore, need to be identified
and removed if learning and improved organizational
performance is to be enabled.
A key area of research around the development of
barriers to learning concerns the nature of the process
around environmental shifts. As the operating environ-
ment becomes more complex (moving from an ordered
through a complex into a chaotic state) then the impact
of such phase transitions on the abilities of organizations
to learn remains a key area for further research.
Cross-References
▶ Absorptive Capacity and Organizational Learning
▶ Acquiring Organizational Learning Norms
▶ Action Learning (and Organizational Development)
▶ Adaptive Behavior
▶ Adaptive Learning
▶ Organizational Change and Learning
▶ Technological Learning in Organizations
References
Argyris, C. (1990). Overcoming organizational defences. Facilitating
organizational learning. Englewood Cliffs: Prentice Hall.
Fischbacher-Smith, D., & Fischbacher-Smith, M. (2009). We may
remember but what did we learn? Dealing with errors, crimes
and misdemeanours around adverse events in healthcare. Finan-
cial Accountability and Management, 25(4), 451–474.
Senge, P. (1990). The fifth discipline: The art and practice of the
learning organisation. London: Century.
Smith, D., & Elliott, D. (2007). Moving beyond denial: Exploring the
barriers to learning from crisis. Management Learning, 38(5),
519–538.
Toft, B., & Reynolds, S. (1994). Learning from disasters. London:
Butterworth.
Turner, B. A. (1978). Man-made disasters. London: Wykeham.
Basal Ganglia
The basal ganglia are a collection of subcortical nuclei
at the base of the forebrain that are interconnected
with cortex, the thalamus, and the brainstem. These
410 B Basal Ganglia Learning
interconnections form cortico-basal ganglia loops can
be anatomically and functionally differentiated. As a
result, the basal ganglia play a critical role in affective,
cognitive, and motor functioning.
The basal ganglia is a region of the brain located
underneath the frontal parts of the brain, adjacent
to the cerebral cortex and the thalamus. It consists
of the striatum, pallidum, substantia nigra, and the
subthalamic nucleus. Collectively, it plays a fundamen-
tal role in task performance, including the selection of
actions in response to stimuli, task switching, and
motor control, which are fundamental aspects of pro-
cedural knowledge.
Basal Ganglia Learning
JULIEN VITAY, FRED H. HAMKER
Department of Artificial Intelligence, Faculty of
Computer Science, Chemnitz University of
Technology, Chemnitz, Germany
Synonyms
Biological groundings of reinforcement learning
Definition
The basal ganglia (BG) are a set of nuclei located in the
basal forebrain, receiving inputs mostly from the cere-
bral cortex (especially the frontal lobe) and projecting
on various motor centers, as well as back to the cortex
through the thalamus, forming a closed-loop. The
main input station is the striatum, which can be ana-
tomically divided into three parts: the nucleus
accumbens (NAcc), the caudate nucleus (CN), and
the putamen (PUT). Striatal neurons collect cortical
activity and inhibit the tonically active neurons (high
activity at rest) in the output nuclei of the BG: the
substantia nigra pars reticulata (SNr) and the internal
segment of the globus pallidus (GPi). These output
structures further inhibit some motor centers and tha-
lamic nuclei. This double inhibition allows to selec-
tively open some recurrent loops between the
thalamus and the cortex, increasing the signal-to-
noise ratio in the cortex and triggering movements or
cognitive functions. Other nuclei in the BG, such as the
subthalamic nucleus (STN) and the external part of the
globus pallidus (GPe), create functionally different
pathways to allow for a more complex role of BG in
adapting behavior. The main characteristic of the BG is
its dense innervation by dopaminergic (DA) cells in the
substantia nigra pars compacta (SNc) and ventral teg-
mental area (VTA), whose firing is related to reward
delivery and prediction. DA can modulate the activa-
tion and learning of most cells in the BG, placing it as
a core structure in reinforcement learning processes.
Theoretical Background
The BG are classically divided into three main func-
tional domains that are engaged in different types of
learning, even though finer subdivisions can be made
(Alexander et al. 1986): the limbic domain, which is
involved in ▶ Pavlovian conditioning and evaluation of
motivational valence; the associative domain, involved
in ▶ goal-directed learning, action–outcome associa-
tion and ▶ working memory; the sensorimotor
domain, involved in selection of action, stimulus–
response associations and ▶ habitual control. These
domains are anatomically segregated, with regard to
the cortical areas and thalamic nuclei that are targeted,
but also to the region of the striatum and other BG
nuclei that are involved, what creates independent and
parallel cortico-basal-ganglia-thalamic loops.
The limbic loop interacts with cortical areas
involved in emotional processing, motivation, and
goal-directed behaviors: The orbitofrontal (OFC), ven-
tromedial prefrontal (vmPFC), and anterior cingulate
(ACC) cortices. Its striatal part, NAcc, integrates also
information coming from the hippocampus (episodi-
cal memory, spatial navigation) and the amygdala
(emotional evaluation, novelty detection). It learns
the emotional valence of stimuli, whether rewarding
or painful, in the context of the organism’s state (value
of food relative to satiety/hunger), location, or plans. It
helps OFC and vmPFC to decide the motivational
value of an outcome, i.e., how interesting it is to try
to obtain it with respect to the previous experiences,
forming the basic mechanism of goal-directed learning.
It is particularly involved in Pavlovian conditioning,
especially in predicting the reward/punishment associ-
ated to the CS. A key experiment is the observed pat-
tern of activation of dopaminergic cells in VTA during
a Pavlovian conditioning task (Schultz et al. 1997).

At the beginning of learning, these cells respond at
the onset of the US, but progressively respond at the
onset of the CS, while the US-related activation van-
ishes, reflecting the learned association. Furthermore,
when the US is omitted, these cells show a pause in
firing that influence learning in other BG domains.
This behavior was compared to the reward-prediction
error signal of the ▶ temporal difference (TD) algo-
rithm in reinforcement learning, and the limbic loop is
often compared to the critic in actor/critic architec-
tures. Furthermore, it is a critical region in learning
action–outcome associations in ▶ instrumental condi-
tioning, as shown by lesions studies. Contrary to Pav-
lovian conditioning, this learning is sensible to the
devaluation of the value of the outcome and to the
contingency between the action and the obtaining
of the predicted outcome: The associative loop tracks
the consequences of actions to guide goal-directed
learning.
The associative loop, comprising mostly the dorsolat-
eral prefrontal cortex (dlPFC), is involved in executive
functions like working memory (WM), planning or set-
shifting, as well as action–outcome associations. WM, the
ability to maintain and manipulate sensory information
over short delays, is principally supported by sustained
activation of cells in dlPFC and CN, the associative loop
learning to selectively maintain and manipulate informa-
tion in WM through the disinhibition of the corres-
ponding thalamocortical loops. This loop also rapidly
learns task-relevant rules through reinforcement learning
to guide processing in dlPFC (Pasupathy and Miller
2005). Part of this loop is furthermore involved in visual
categorization and visual memory retrieval with its
connections with the inferotemporal cortex that
processes high-level visual features.
The motor loop, with premotor (PM), motor (M),
and somatosensory (S) cortices, is involved in motor
control, action selection, stimulus–responses associa-
tions, and habit formation. The degeneration of its
corresponding dopaminergic cells is responsible for
the Parkinson’s disease. PUT is somatotopically orga-
nized, with adjacent regions being selective for different
effectors or body parts. It is active both in the prepara-
tion and execution of movements and is thought to
favorize the most adequate movement (in terms of
predicted reward) through the disinhibition of the
corresponding thalamocortical loop or directly of
Basal Ganglia Learning B 411
motor centers like the superior colliculus for saccades
(Hikosaka et al. 2000). It principally learns rewarded
stimulus–response associations, which are not sensitive B
to outcome devaluation and changes in action-
outcome contingencies after learning, as shown by
lesion studies. This implies an involvement in habit
formation rather than in goal-directed learning. It is
a key structure in sequence learning, thanks to its
looped inner architecture, by associating a known
movement by its follower in the sequence and by con-
trolling the timing of the switch.
Important Scientific Research and
Open Questions
Although these three loops have been thought to
behave independent, the question of their interaction
during learning is still not clear. The Pavlovian-to-
Instrumental transfer, for example, shows that Pavlov-
ian conditioning influences instrumental responses
learned in a similar context. Indeed, the three loops
are not totally segregated. They interact through
overlapping cortical projections (for example, dlPFC
also projects on NAcc and PUT, although less inten-
sively) and divergent thalamocortical projections, but
also through the ascending spiral pattern of connectiv-
ity between the striatum and the dopaminergic areas
SNc/VTA (Haber et al. 2000): Each domain of the
striatum forms closed reciprocal connections with
a region of SNc/VTA, but also projects to the adjacent
dopaminergic area. This anatomical arrangement cre-
ates a hierarchy of information flow from the limbic via
the associative to the motor loop that has functional
implications still to be investigated.
Another open issue is the exact role of dopamine
firing in behavior. The classical analogy with the error
signal of the TD algorithm faces several problems:
DA cells fire stereotypically with a very short latency to
reward-predicting events, but also to novel or salient
stimuli, that is inconsistent. An interesting alternative
hypothesis is that DA cells only signal unpredicted behav-
iorally relevant events through subcortical processes in
order to direct attention and cognitive resources toward
this event and to promote the reselection of components
of behavior and context that immediately precede it
(Redgrave and Gurney 2006).
More generally, the BG can be seen as a central
selection device that favorizes cortical representations
412 B Base Domain
which have been associated to emotionally relevant
events in the past experience. It is able to learn these
associations faster than the cortex and acts as
a supervising teacher for the cortex, where they can be
stored on a longer timescale. The respective contribu-
tions to learning of the BG and the cortex are still
unclear.
Cross-References
▶ Emotional Learning
▶ Habit Learning in Animals
▶ Operant Learning
▶ Pavlovian Conditioning
▶ Reinforcement Learning
▶ Rule Learning
▶ Working Memory and Information Processing
References
Alexander, G. E., DeLong, M. R., & Strick, P. L. (1986). Parallel
organization of functionally segregated circuits linking the
basal ganglia and cortex. Annual Review of Neuroscience, 9,
357–381.
Haber, S. N., Fudge, J. L., & McFarland, N. R. (2000). Striatoni-
grostriatal pathways in primates form an ascending spiral from
the shell to the dorsolateral striatum. The Journal of Neuroscience,
20(6), 2369–2382.
Hikosaka, O., Takikawa, Y., & Kawagoe, R. (2000). Role of the basal
ganglia in the control of purposive saccadic eye movements.
Physiological Reviews, 80(3), 953–978.
Pasupathy, A., & Miller, E. K. (2005). Different time courses of
learning-related activity in the prefrontal cortex and striatum.
Nature, 433(7028), 873–876.
Redgrave, P., & Gurney, K. (2006). The short-latency dopamine
signal: a role in discovering novel actions? Nature Reviews Neu-
roscience, 7(12), 967–975.
Schultz, W., Dayan, P., & Montague, P. R. (1997). A neural substrate
of prediction and reward. Science, 275(5306), 1593–1599.
Base Domain
Familiar situation, context or problem; the original
learning context.
Basic Units
▶ Human Feature Learning
Basis Behaviors for Robot
Learning
▶ Motor Schemas in Robot Learning
Bateson and Anthropology of
Learning
▶ Bateson, Gregory (1904–1980): Anthropology of
Learning
Bateson, Gregory (1904–1980):
Anthropology of Learning
WOLFRAM LUTTERER
Department of Education, University of Freiburg,
Freiburg, Germany
Synonyms
Anthropological aspects of learning; Bateson and
anthropology of learning; Cultural aspects of learning;
Gregory Batesons’s anthropology of learning; Learning
and anthropology
Definition
Human learning is not only dependent on its close
social settings (family, peers, school, etc.), it is also
largely influenced by the particular culture, and it is
likewise enabled as also restricted by human biology.
The anthropology of learning searches for the variety
and homogeneity of human culture, and especially of
human learning. So the research undertaken is largely
comparative.
The Anglo-American anthropologist Gregory
Bateson (1904–1980) undertakes a quite formal
approach to clarify the structure and the scope of the
learning process. He delivers a theoretical frame for
research. This is be done by a differentiation of five
levels or types of learning. Learning of a higher logical
type is defined as “change in the process” of the lower
level. So, for example, “Learning II is change in the
 Bateson, Gregory (1904–1980): Anthropology of Learning B 413

process of Learning I ” (Bateson, Steps to an Ecology of character, something to be done sometimes in psy-
Mind, 1972/2000, p. 293). All in all, Bateson differenti- chotherapy or in religious conversion. So people
ates from Learning 0 (“zero learning”) up to Learning could learn “to form more readily” habits, or “to B
IV. These levels clarify the complexity and the structure change habits acquired by Learning II” (p. 303). In
in the process of learning. the 1970s, this level of learning gave reason to some
speculations about change in consciousness.
● Learning 0 describes a reaction to something, but
● Learning IV is change in the process of Learning III.
without learning anything “new” (in the common
The combination of phylogenesis with ontogenesis
sense view of learning). This Learning 0 may at first
would be of this type. But according to Bateson this
seem trivial, but it is needed for a complete descrip-
type of learning does probably not occur by any
tion of all five levels, and also in correspondence to
living organism. The limit of the human learning
other levels. The result of Learning 0 is indeed
process is at Level III.
a change in behavior, but in a highly stereotyped
way, as a simple reaction to something: Maybe the This listing should also demonstrate that the
alarm clock rings and you know it is time to get up. “higher” levels of learning are not the “better” ones.
● Learning I is more or less equal with our common They are different types, and they relate to each other.
sense view of learning. It is formally defined as With these levels of learning, several points in the
“change in specificity of response by correction of anthropology of learning can be clarified.
errors of choice within a set of alternatives.” So,
1. It is shown that there is an upper limit of the learn-
the precondition of this type of learning is the
ing process. Learning 0 to Learning II are part of our
ability for the perception of error and then to
everyday life, but the acquisision of Learning II
change behavior. The whole range of conventional
already tends to slow down by further learning, by
psychological learning theory is equivalent to
life time. Learning III is difficult and seldom, Learn-
Learning I (operant and classical conditioning,
ing IV impossible. So the “higher” the level, the
rote learning, etc.). So I am learning to repair my
more fundamental and the more restricted by
bike, to dance, or to speak another language.
nature and by former learning experiences.
● Learning II or Deuterolearning (see cross-refer-
It is demonstrated that learning (in sense of
ence) is “a corrective change in the set of alterna-
Learning I) is always in context. This context is of
tives from which choice is made, or it is a change in
significance for the interpretation of events. It is
how the sequence of experience is punctuated.” This
double sided: context created by others, and also
type of learning is in context to Learning I. The way
by oneself, the own expectations, the own world-
in which learning occurs has an impact to the way
view as a result of Learning II.
we see things, what we anticipate, what we experi-
2. Different cultures as also individual worldviews
ence as “normal” and therefore, also to the way we
create different contexts, and therefore different
behave. Our worldview is at least partly influenced
understandings of “reality” and of “truth.”
by learning processes and this kind of influence is
3. The process of learning is shown in his complexity.
on the level II of our learning process. Learning II is
It is to be assumed that Learning II occurs synchro-
also reached by some mammals (explicitly shown
nous to Learning I and to Learning 0. In addition to
by dolphins, see Bateson 2000, p. 394 ff). Habits as
Bateson, it can be assumed that Learning 0 is essen-
“fatalism”, “pride,” or “dependency” are results of
tial for the validation of Learning II, or our world-
learning processes of this type. Bateson understands
view. Successful learning leads to Learning 0 in
that much of this Learning II dates from early
repeated contexts.
infancy and it is unconscious. Another result of
4. It is formally shown how habits develop and, there-
Learning II is that I am learning to learn. Then
fore, how they are a matter of possible change.
I am starting to learn faster.
● Learning III is change in Learning II and therefore The complexity, the variability, and the flexibility of
something more difficult and rare even for human learning contexts is formally represented by
beings. It stands for a profound reorganization of a multilevel view of the learning process.
414 B Bateson, Gregory (1904–1980): Anthropology of Learning
Theoretical Background
Bateson’s learning theory was developed over
three decades, beginning with a first version,
distinguishing between proto- and deuterolearning
(later: Learning I and Learning II), in 1942. The final
version of his theory, consisting in the five levels of
learning, was mainly developed in the 1960s, finished
1971, and published 1972.
There are three main fields of research within
Bateson’s own writings, which are creating
a theoretical background. First, his anthropological
research in New Guinea and Bali, performed partly
together with his at that time wife Margaret Mead
and resulting in the first draft of the theory (period
from 1927 to 1942, see Bateson and Mead 1942).
Second, his research on human communication and
of pathologies in communication, resulting in his com-
munication theory, 1951, and double-bind theory,
1956 (1948–1962). Third, research with dolphins and
octopuses, showing the complexity of communication
and of learning by animals (1959–1972).
Bateson was an active member of the famous Josiah
Macy conferences on cybernetics between 1946 and
1953. His work is strongly influenced by cybernetics,
informations theory, and game theory. The impact of
cybernetics in relation to learning theory is especially to
be seen in a wide-range consideration of the circularity
of interaction. But Batesons view of cybernetics differs
from the steering and control-orientated view by
Norbert Wiener and others. His reflexive understand-
ing of cybernetics partly anticipates the second-order
cybernetics of Heinz von Foerster.
A strong impact on the construction of the learning
theory itself has Bertrand Russell’s theory of
logical types, developed in the Principia Mathematica
(1910–1913) to distinguish between elements and
classes. So, as Bateson points out, for example, the
name of a thing is not the thing itself: you can sit on
a chair but not on the word “chair.” In the final version
of Bateson’s theory of learning, type theory is central
for the definition.
Important Scientific Research and
Open Questions
The full impact of cybernetics was not exhausted by
Bateson. So, his theory remains largely hierarchical
(not systemic) and strongly influenced by Bertrand
Russell’s theory of logical types. There is a whole
section within the learning theory dedicated to
type theory.
Furthermore, Bateson’s application of type theory
seems to be a bit idiosyncratic: On a first view, it may
seem to be quite clear to distinguish the “higher” level
as of a higher type than the other, but on a second view
it remains unexplained how Learning 0 relates to Learn-
ing II, which is done directly and not via Learning I:
Learning experiences are either Learning 0 or Learning
I, but they are always touching to Learning II. Further-
more it is not really clear how the temporal aspect of
the several levels changes. Learning 0 to Learning II
surely are to be viewed as synchronous to each other,
but Learning II has also diachronic aspects, still further
Learning III.
Further research strengthens the impact of Learning
0 : The multiplicy of Learning 0 experiences is essential
for the affirmation of worldviews. There are also argu-
ments, which suggest that some aspects of Learning III
are less difficult to achieve than proposed by Bateson
(cf. Lutterer 2010).
Although Bateson’s contribution to learning theory
seems to stand somehow peripheral in the scientific
discourse, it is also in the background of several
established concepts (see entry to Deuterolearning).
This applies also for his contribution to the anthropol-
ogy of learning. It seems to be the result of the failure in
the constitution of a sustainable cybernetically orien-
tated scientific discourse. Bateson’s whole theory
remains as another kind of scientific language (cf.
Bateson and Bateson 1987), which is apparently often
not compatible to the scientific mainstream.
There are also other anthropological approaches to
learning, but they remain largely comparative. An over-
view including further discussion gives Wolcott (1982).
The anthropological approach to learning shows the
diversity of human culture: the more the differences,
the less the biological (or genetical) constraints in the
formation of character, behavior, and of society.
Bateson himself shows this paradigmatically on his
research on the Iatmul in New Guinea (Bateson
1958), and of the people of Bali (Bateson and Mead
1942; Bateson 2000, p. 107 ff).
Cross-References
▶ Anthropology of Learning and Cognition
▶ Deuterolearning
▶ Informal Learning

References
Bateson, G. (1958). Naven: A survey of the problems suggested by
a composite picture of the culture of a New Guinea tribe drawn
from three points of view (2nd ed.). Stanford: Stanford University
Press.
Bateson, G. (2000). Steps to an ecology of mind. Chicago: University of
Chicago Press [Orig 1972].
Bateson, G., & Bateson, M. C. (1987). Angels fear. Towards an episte-
mology of the sacred. New York: Bantam.
Bateson, G., & Mead, M. (1942). Balinese character: A photographic
analysis. New York: New York Academy of Sciences.
Lutterer, W. (2010). Der Prozess des Lernens: Eine Synthese der
Lerntheorien von Jean Piaget und Gregory Bateson. Velbrück:
Weilerswist.
Wolcott, H. F. (1982). The anthropology of learning. Anthropology &
Education Quarterly, 13(2), 83–108.
Bayesian Learning
BRADLEY C. LOVE1, MATT JONES2
1
Department of Psychology, The University of Texas at
Austin, Austin, TX, USA
2
University of Colorado, Boulder, CO, USA
Synonyms
Bayesian model; Normative; Probabilistic approaches;
Rational
Theoretical Background
Bayesian methods have undergone tremendous pro-
gress in recent years, due largely to mathematical
advances in probability and estimation theory (Chater
et al. 2006). These advances have allowed theorists to
express and derive predictions from far more sophisti-
cated models than previously possible. These models
have generated a good deal of excitement for at least
two reasons. First, they offer a new interpretation of the
goals of cognitive systems, in terms of inductive prob-
abilistic inference, which has revived attempts at ratio-
nal explanation of human behavior (Oaksford and
Chater 2007). Second, Bayesian models may have the
potential to explain some of the most complex aspects
of human cognition, such as language acquisition or
reasoning under uncertainty, where structured infor-
mation and incomplete knowledge combine in a way
that has defied previous approaches (e.g., Kemp and
Tenenbaum 2008).
Bayesian Learning B 415
Constructing a Bayesian model involves two steps.
The first step is to specify the set of possibilities for the
state of the world, which is referred to as the hypothesis B
space. Each hypothesis can be thought of as
a prediction by the subject about what future sensory
information will be encountered. However, the term
hypothesis should not be confused with its more tradi-
tional usage in psychology, connoting explicit testing of
rules or other symbolically represented propositions.
In the context of Bayesian modeling, hypotheses need
have nothing to do with explicit reasoning, and indeed
the Bayesian framework makes no commitment what-
soever on this issue. For example, in Bayesian models of
visual processing, hypotheses can correspond to
extremely low-level information, such as the presence
of elementary visual features (contours, etc.) at various
locations in the visual field (Geisler et al. 2001). There is
also no commitment regarding where the hypotheses
come from. Hypotheses could represent innate biases
or knowledge, or they could have been learned previ-
ously by the individual. Thus, the framework has no
position on nativist–empiricist debates. Furthermore,
hypotheses representing very different types of infor-
mation (e.g., a contour in a particular location,
whether or not the image reminds you of your mother,
whether the image is symmetrical, whether it spells
a particular word, etc.) are all lumped together in
a common hypothesis space and treated equally by
the model. Thus, there is no distinction between dif-
ferent types of representations or knowledge systems
within the brain. In general, a hypothesis is nothing
more than a probability distribution. This distribution,
referred to as the likelihood function, simply specifies
how likely each possible pattern of observations is
according to the hypothesis in question.
The second step in constructing a Bayesian model is
to specify how strongly the subject believes in each
hypothesis before observing data. This initial belief is
expressed as a probability distribution over the hypoth-
esis space, and is referred to as the prior. The prior can
be thought of as an initial bias in favor of some hypoth-
eses over others, in that it contributes extra “votes” (as
elaborated below) that are independent of any actual
data. This decisional bias allows the model’s predic-
tions to be shifted in arbitrary directions regardless
of the data. As we discuss below, the prior can be
a strong point of the model if it is derived indepen-
dently, from empirical statistics of real environments.
416 B Bayesian Learning
However, more commonly, the prior is chosen ad hoc,
providing substantial unconstrained flexibility to
models that are advocated as rational and assump-
tion-free.
Together, the hypotheses and the prior fully deter-
mine a Bayesian model. The model’s goal is to decide
how strongly to believe in each hypothesis after data
have been observed. This final belief is again expressed
as a probability distribution over the hypothesis space
and is referred to as the posterior. The statistical identity
known as Bayes’ Rule is used to combine the prior with
the observed data to compute the posterior. Bayes’ Rule
can be expressed in many ways, but here we explain
how it can be viewed as a simple vote-counting model.
Specifically, Bayesian inference is equivalent to tracking
evidence for each hypothesis, or votes for how strongly
to believe in each hypothesis. The prior provides the
initial evidence counts, Eprior, which are essentially
made-up votes that give some hypotheses a head start
over others, before observing any actual data. When
data are observed, each observation adds to the existing
evidence according to how consistent it is with each
hypothesis. The evidence contributed for a hypothesis
that predicted the observation will be greater than the
evidence for a hypothesis under which the observation
was unlikely. The evidence contributed by the ith
observation, Edatai ; is simply added to the existing
evidence to update each hypothesis’ count. Therefore
the final evidence, Eposterior, is nothing more than a sum
of the votes from all of the observations, plus the initial
votes from the prior. (Formally, Eposterior equals the
logarithm of the posterior distribution, Eprior is the
logarithm of the prior, and Edata(H) is the logarithm
of the likelihood of the data under hypothesis H. The
model’s prediction for the probability that hypothesis
H is correct, after data have been observed, is propor-
tional to exp[Eposterior(H)]).
X Bayesian modeling become more complex than the
Eposterior ðHÞ ¼ Eprior ðHÞ þ Edatai ðHÞ ð1Þ
i
This sum is computed for every hypothesis, H, in
the hypothesis space. The vote totals determine how
strongly the model believes in each hypothesis in the
end. Thus, any Bayesian model can be viewed as track-
ing evidence for each hypothesis, with initial evidence
coming from the prior and additional evidence coming
from each new observation. At its core, this is all there is
to Bayesian modeling.
To illustrate these two steps and how inference
proceeds in a Bayesian model, consider the problem of
determining whether a fan entering a football stadium is
rooting for the University of Southern California (USC)
Trojans or the University of Texas (UT) Longhorns based
on three simple questions: (1) Do you live by the ocean?
(2) Do you own a cowboy hat? (3) Do you like Mexican
food? The first step is to specify the space of possibilities
(i.e., hypothesis space). In this case, the hypothesis space
consists of two possibilities: being a fan of either
USC or UT. Both of these hypotheses entail
probabilities for the data we could observe, for example,
PðoceanjUSCÞ ¼ :8 and PðoceanjUTÞ ¼ :3. Once these
probabilities are given, the two hypotheses are fully
specified. The second step is to specify the prior. In
many applications, there is no principled way of doing
this, but in this example, the prior corresponds to the
probability that a randomly selected person will be
a USC or a UT fan, that is, one’s best guess as to the
overall proportion of USC and UT fans in attendance.
With the model now specified, inference proceeds
by starting with the prior and accumulating evidence as
new data are observed. For example, if the football
game is being played in Los Angeles, one might expect
that most people are USC fans, and hence the prior
would provide an initial evidence count in favor of
USC. If our target person responded that he lives near
the ocean, this observation would add further evi-
dence for USC. The magnitudes of these evidence
values will depend on the specific numbers assumed
for the prior and for the likelihood function for each
hypothesis, but all that the model does is take the
evidence values and add them up. Each new observa-
tion adds to the balance of evidence among the
hypotheses, strengthening those that predicted it rel-
ative to those under which it was unlikely.
There are several ways in which real applications of
simple example above. However, these all have to do
with the complexity of the hypothesis space rather than
the Bayesian framework itself. For example, many
models have a hierarchical structure, in which hypoth-
eses are essentially grouped into higher-level
overhypotheses. Overhypotheses are generally more
abstract and require more observations to discriminate
among; thus hierarchical models are useful for model-
ing learning or change over developmental timescales
(e.g., Kemp et al. 2007). However, each overhypothesis

is just a weighted sum of elementary hypotheses, and
inference among overhypotheses comes down to
exactly the same vote-counting scheme as described
above. As a second example, many models assume
special mathematical functions for the prior, such as
conjugate priors, that simplify the computations
involved in updating evidence. However, such assump-
tions are generally made solely for the convenience of
the modeler, rather than for any psychological reason
related to the likely initial bias of a human subject.
Finally, for models with especially complex hypothesis
spaces, computing exact predictions often becomes
computationally intractable. In these cases, sophisti-
cated approximation schemes are used, such as Mar-
kov-chain Monte Carlo (MCMC) or particle filtering
(i.e., sequential Monte Carlo). These algorithms yield
good estimates of the model’s true predictions while
requiring far less computational effort. However, once
again they are used for the convenience of the modeler
and usually are not meant as proposals for how human
subjects might solve the same computational problems.
To summarize: Hypotheses are probability distribu-
tions and have no necessary connection to explicit rea-
soning. The model’s predictions depend on the initial
biases on the hypotheses (i.e., the prior). The heart of
Bayesian inference – combining the prior with observed
data to reach a final prediction – is formally equivalent
to a simple vote-counting scheme. Learning and one-off
decision-making both follow this scheme, and are iden-
tical except for timescale and specificity of hypotheses.
Most of the elaborate mathematics that often arises in
Bayesian models comes from the complexity of their
hypothesis sets or the tricks used to derive tractable
predictions, which generally have little to do with the
psychological claims of the researchers. Bayesian infer-
ence itself, aside from its assumption of optimality and
close relation to vote-counting models, does not make
psychological claims in recards to representational for-
mat, encoding, retrieval, attention, etc. However, the
flexibility and power of the Bayesian framework has
allowed researchers to model complex learning and
decision-making behaviors that have proven intractable
or unwieldly under other formulations.
Important Scientific Research and
Open Questions
The restriction to computational-level accounts (cf.
Marr 1982) severely limits contact with process-level
Bayesian Learning B 417
theory and data. Rational approaches attempt to
explain why cognition produces the patterns of behav-
ior that it does, but they offer no insight into how B
cognition is carried out. Second, in general, there are
multiple rational theories of any given task,
corresponding to different assumptions about the envi-
ronment and the learner’s goals. Consequently, there is
insufficient acknowledgement of these assumptions
and their critical roles in determining model predic-
tions. It is extremely rare to find a comparison among
alternative Bayesian models of the same task to deter-
mine which is most consistent with empirical data.
Likewise, there is little recognition when the critical
assumptions of a Bayesian model logically overlap
closely with those of other theories. These challenges
are currently being addressed by members of the Bayes-
ian community. The end goal is to integrate Bayesian
approaches with what we know about the mental pro-
cesses that support learning and decision making
(Jones and Love 2011).
Cross-References
▶ Concept Learning
▶ Human Causal Learning
▶ Learning Algorithms
▶ Mathematical Models/Theories of Learning
▶ Metatheories of Learning
▶ Normative Reasoning and Learning
References
Chater, N., Tenenbaum, J. B., & Yuille, A. (2006). Probabilistic models
of cognition: Conceptual foundations. Trends in Cognitive
Sciences, 10(7), 287–291.
Geisler, W. S., Perry, J. S., Super, B. J., & Gallogly, D. P. (2001). Edge
co-occurrence in natural images predicts contour grouping per-
formance. Vision Research, 41, 711–724.
Jones, M., & Love, B. C. (in press, 2011). Bayesian fundamentalism or
enlightenment? On the explanatory status and theoretical con-
tributions of bayesian models of cognition. Behavioral and Brain
Sciences.
Kemp, C., & Tenenbaum, J. B. (2008). The discovery of structural
form. Proceedings of the National Academy of Sciences, 105,
10687–10692.
Kemp, C., Perfors, A., & Tenenbaum, J. B. (2007). Learning
overhypotheses with hierarchical Bayesian models. Developmen-
tal Science, 10, 307–321.
Marr, D. (1982). Vision. San Francisco: W.H. Freeman.
Oaksford, M., & Chater, N. (2007). Bayesian rationality: The proba-
bilistic approach to human reasoning. Oxford: Oxford University
Press.
418 B Bayesian Model
Bayesian Model
▶ Bayesian Learning
Bee Learning and
Communication
MANDYAM V. SRINIVASAN
Queensland Brain Institute, School of Information
Technology and Electrical Engineering, University of
Queensland, St. Lucia, QLD, Australia
Australian Research Council Centre of Excellence in
Vision Science, University of Queensland, St. Lucia,
QLD, Australia
Synonyms
Associative learning in bees; Honeybee learning
Definition
Research over several decades has revealed that honey-
bees are probably the most highly evolved insect in
terms of the capacity of individuals to learn colors,
shapes, and smells, as well as relatively abstract associ-
ations and complex navigational routes. These crea-
tures have also evolved sophisticated methods of
communicating information to one another through
complex, ritualized behaviors. This entry summarizes
our current knowledge about these capacities, and
highlights some of the unanswered questions.
Theoretical Background
The lifestyle of a honeybee requires the ability to learn
and recognize the visual appearance, smell, and location
of nectar and pollen bearing flowers, so that it can return
to the food source quickly, repeatedly, and unerringly. It
also requires an individual bee to be able to communi-
cate information on the location of the food source to its
nest mates so that they, too, can find the source and
forage from it to build up the colony’s nectar and pollen
reserves. Here we outline some of the well-known learn-
ing capacities of honeybees.
Color learning: Honeybees possess excellent tri-
chromatic color vision, involving three classes of
photoreceptors that are sensitive to the ultraviolet,
blue, and green regions of the light spectrum. Bees
can readily be trained, by food reward, to learn colors
and to discriminate between them (von Frisch 1993).
As in humans, bees are able to perceive the inherent
color (the reflectance spectrum) of an object largely
independently of the spectrum of the light that illumi-
nates it, a phenomenon known as “color constancy”
(Srinivasan 2009). This enables a blue flower to be
perceived as having more or less the same blue hue
regardless of whether it is viewed under the midday
sun – when the spectrum of the illumination is approx-
imately “white” – or in the evening, when the illumi-
nation is more reddish. Color is perhaps the strongest
visual cue that bees use to distinguish between flowers.
Pattern learning: Bees can be trained to distinguish
between different patterns – such as horizontal stripes
versus vertical stripes, star-shaped patterns versus pat-
terns composed of concentric rings, objects of different
sizes, and so on (Wehner 1981; Horridge 2009). Obvi-
ously, this is also of great importance in recognizing
food-bearing flowers. In learning to distinguish
between patterns, bees often display what appears to
be an “eidetic” type of memory in which the patterns
are learned as a photographic image (Wehner 1981).
However, bees can also “generalize” pattern learning to
distinguish between different categories of patterns –
such as the class of all horizontally oriented patterns
versus the class of all vertically oriented patterns, the
class of symmetrical patterns versus the class of asym-
metrical patterns, and so on (Srinivasan 2009). These
faculties are put to good use in distinguishing between
different flowers, as well as between different classes of
flowers, nests, or navigational landmarks. Bees can also
be trained to detect camouflaged patters that are nor-
mally invisible to them (Srinivasan 2009).
Olfactory learning: Bees can learn a scent if they are
trained to feed at a sugar water feeder that is laced with
the scent. They can be trained to distinguish between
two different scents by associating one of them with
a sugar reward (von Frisch 1993). Rather like the Pav-
lovian dog, bees can be conditioned to extend their
proboscis in anticipation of food when they experience
a particular scent. This so-called proboscis extension
reflex (PER) is being used in many laboratories to
investigate odor discrimination at the behavioral and
the molecular levels, and to tease out the mechanisms
of olfactory learning.
Navigational learning in the laboratory: The ability
to learn routes through mazes has been investigated

extensively in several vertebrates, notably rats, mice,
pigeons, and fish. However, until recently, relatively
few studies have explored maze learning in inverte-
brates. We now know that bees can be trained to nego-
tiate complex labyrinths in several different ways
(Srinivasan 2009). For example, they can learn to nav-
igate a labyrinth if a visual label is provided to identify
the correct exit in each chamber. Bees can also learn
a maze route if they are given symbolic guidance cues:
for example, turn left if the chamber is blue, and right if
it is yellow. And they can also learn to conquer laby-
rinths that carry no guidance cues whatsoever –
although their performance in such mazes is not as
good as in cued mazes. Bees can be trained to fly
through unmarked mazes if the correct route involves
a pattern of turns that is repetitive or predictable: for
example, always turn left, or always turn right, or turn
alternately left and right in successive chambers.
Learning complex associations: Mazes have been use-
ful not only for studying the learning of complex
routes, but also for probing the ability of bees to form
complex associations. One example of such an appli-
cation is the so-called Delayed Match to Sample
(DMTS) task. Here the bee is shown a “sample” stim-
ulus at the entrance to a chamber. When it enters the
chamber, it is confronted with two stimuli, only one of
which matches sample. The bee has to learn to choose
the matching sample, by reward. Bees learn to perform
this DMTS task well, regardless of whether the stimuli
are presented in the color domain, in the pattern
domain, or in the olfactory domain (Srini ARE). Fur-
thermore, bees that have been trained to match olfac-
tory stimuli are able to match visual stimuli, without
having to be trained specifically on the visual stimuli.
Bees can therefore learn the concept of “matching” in
a general way and apply it across stimulus modalities
(Srinivasan 2009). Bees can also learn the concept
of a “non-match”: that is, they can learn to choose the
stimulus that does not match the sample stimulus
(Srinivasan 2009). A task of greater complexity is the
“Delayed Symbolic Match to Sample” (DSMTS) task.
Here the bee has to use the identity of a sample stim-
ulus (which can be A or B) to choose between two other
comparison stimuli (C and D) that are presented
simultaneously in a subsequent chamber. The bee has
to learn to choose stimulus C if the sample is A, and
stimulus D if the sample is B. In other words, the bee
has to learn to associate A with C, and B with D. A and
Bee Learning and Communication B 419
B could be, for example, two different colors (blue and
yellow) and C and D two different patterns (horizontal
and vertical stripes). In this task, the matching is sym- B
bolic rather than literal. Bees are able to learn this task
well, too (Srinivasan 2009). Until recently, only verte-
brates – humans, monkeys, and pigeons – were known
to be able to learn DMTS and DSMTS tasks. In the
honeybee, these capacities are likely to enhance perfor-
mance in finding a target or navigating reliably to it.
Learning what to do where and when: Bees can learn
tasks that require a flexible, context-dependent
response. For example, an individual bee can be trained
to find food at one particular location when it enters
one enclosed space, and at a different location when
it enters a different enclosed space. As another
example, bees, trained to distinguish between two
different scents, say A and B, can learn to choose
A over B at a particular time of day (e.g., in the morn-
ing) but to choose B over A at a different time of day
(e.g., in the afternoon). The ability to learn to make
flexible choices that vary with changes in space and
time is invaluable to a flower forager, because the avail-
ability of nectar-bearing flowers in the environment
can often vary with location as well as time of day
(Srinivasan 2009).
Navigation in natural outdoor environments: Navi-
gation in flying insects has been studied most intensely
in the honeybee. The reason for this probably arises
from the famous “waggle dance,” which a so-called
scout bee performs after returning home from an
attractive food source to advertise to its nest mates
the distance and direction of the goal. The dance is
performed on the vertical surface of the honeycomb.
The bee moves in a series of alternating left- and right-
hand loops, each pair of loops shaped roughly like
a figure of eight. At the end of each loop, the bee enters
a so-called waggle phase in which she waves her abdo-
men rapidly from side to side. The angle between
the axis of the waggle and the vertical direction repre-
sents the angle between the sun and the direction in
which a bee should fly in order to find the goal. The
duration of the waggle phase is proportional to the
distance of the food source from the hive. If the food
source is a very short distance away from the hive
(within 50 m), the dance consists of a series of curved
loops, with no waggle phase. This type of dance, termed
the “round dance,” conveys no direction information
but signifies that the goal is very close to the hive.
420 B Bee Learning and Communication
The dancing bee therefore uses a highly symbolic
“language” to communicate the location of the food
source relative to the hive (von Frisch 1993). A bee that
has discovered a new food source typically requires
about five visits to learn its location well, after
which it commences to dance and recruit other bees
to the location – but only if it finds the food to be
sufficiently nutritious and worthy of the energy that is
expended to make the journey to it. Bees estimate the
distance they have flown to a food source by sensing the
extent to which the image of the world has moved in
the eyes during the journey (Srinivasan 2011). Learning
the location of a food source, and the route to it,
requires combining information on the moment-to-
moment directions of flight (as determined from the
celestial compass) with information on the distance
traveled along each of these directions during the jour-
ney. This process is known as “path integration.” Route
learning and goal recognition are made more robust by
acquiring visual “snapshots” at various reference points
along the way (which may contain salient landmarks)
and at the destination.
Counting: Bees are able to pinpoint the location of
a food source not only by measuring its distance from
the nest as described above, but also by learning to
count the number of salient landmarks that are
encountered en route. Bees can count sequentially up
to a maximum of four in this way. Bees can also learn
to assess the relative numerosity of objects viewed
simultaneously, again up to a maximum of four
(Srinivasan 2009).
Signaling schemas: We have already seen how the
dance is used to signal the location of a food source.
Other behavioral forms of communication are,
for example, (a) “begging”signals, that a potential
recruit directs at a dancing bee to request a taste of
the nectar that the dancer has just brought in; (b) the
“tremble” dance, in which a bee returning with a nectar
load signals that she is waiting for a nest mate to collect
the nectar from her; (c) “stop”signals, where a bee head-
butts a dancing bee and/or emits a brief buzzing tone to
stop it from dancing, either to stop recruitment because
there are already too many foragers at the site
being advertised, or because the site harbors danger
from a waiting predator, such as a spider (Srinivasan
2010). In addition, honeybees use a number of
pheromones to communicate information via the
olfactory sense. This is a separate topic, too extensive
discuss here.
Important Scientific Research and
Open Questions
Although some of the capacities outlined above would
appear to be more ecologically relevant than others,
these studies have served to explore the boundaries of
the visual, perceptual, cognitive, and navigational
faculties of honeybees. They raise a number of ques-
tions, particularly in relation to the underlying neural
mechanisms. Some of the challenging, and as yet unan-
swered questions are:
● What are the neural mechanisms that underpin
the learning of colors? Although there is data on
the color-induced responses in the photoreceptors
and in neurons at various levels of the visual path-
way, we have scant knowledge about how colors are
ultimately represented in the brain, and where their
memories are stored.
● What are the neural mechanisms that underlie
pattern recognition? What are the features of the
image that are extracted by the visual system, and
how they are assembled to represent an object in
a compact and reliable way so that it can be recog-
nized when it is encountered again?
● What is the nature of the neural circuitry that
computes, registers, and stores how far a bee has
flown? And how is this information “read out” and
translated into a waggle dance?
● At the neural level, how does the brain combine
information on the distance and direction of flight
to perform the process of “path integration” that
informs the insect about where it is located in
relation to its nest?
● How are landmarks encountered en route learned,
recognized, and incorporated into the bee’s neural
representation of the route that it learns to
a particular food source?
● How can some of these feats, accomplished by a
creature possessing a 1 mg brain and fewer than
a million neurons, be translated into novel, biolog-
ically inspired algorithms for learning machines
and miniature, intelligent flying robots?
 Behavior Modification, Behavior Therapy, Applied Behavior Analysis and Learning
Cross-References
▶ Abstract Concept Learning in Animals
▶ Accounting and Arithmetic Competence in Animals
▶ Animal Learning and Intelligence
▶ Individual Learning
▶ Intelligent Communication in Social Animals
▶ Memory for “What”?, “Where”?, and “When”?
Information in Animals
▶ Learning in Honeybees: Associative Processes
▶ Place Learning and Spatial Navigation
References
Horridge, G. A. (2009). What does the honeybee see? And how do we
know? Canberra: ANU E Press, The Australian National
University.
Srinivasan, M. V. (2009). Honeybees as a model for vision, perception
and ‘cognition’. Annual Review of Entomology, 55, 267–284.
Srinivasan, M. V. (2010). Honeybee communication: A signal for
danger. Current Biology, 20, R366–R367.
Srinivasan, M. V. (2011). Honeybees as a model for the study of
visually guided flight, navigation, and biologically inspired
robotics. Physiological Reviews, 91, 389–411.
von Frisch, K. (1993). The dance language and orientation of bees.
Cambridge, MA: Harvard University Press.
Wehner, R. (1981). Spatial vision in arthropods. In H. Autrum (Ed.),
Handbook of sensory physiology. Berlin: Springer.
Beginner Learning
▶ Novice Learning
Behavior Change
▶ Learning from Counseling
Behavior Family Therapy
▶ Cognitive-Behavioral Family Therapy
B 421
Behavior Modification,
Behavior Therapy, Applied B
Behavior Analysis and Learning
JOSEPH J. PEAR1, GARRY L. MARTIN2
1
Department of Psychology, University of Manitoba,
Winnipeg, MB, Canada
2
St. Paul’s College University of Manitoba,
Winnipeg, MB, Canada
Synonyms
Behavioral engineering; Conditioning applications;
Conditioning therapies; Learning-based therapies; Pav-
lovian and operant technologies
Definition
Behavior modification (BM), behavior therapy (BT),
and applied behavior analysis (ABA) all refer to evi-
dence-based applications of learning theory to the
solution of practical problems. These terms all came
into prominence in the 1950s and 1960s, and their
meanings have evolved since their first introduction.
Theoretical Background
The roots of BM, BT, and ABA lie in the discovery of
conditioning or the development of new reflexes
through the pairing of stimuli, as identified indepen-
dently by Ivan P. Pavlov (1849–1936) and Vladimir M.
Bechterev (1857–1927). Pavlov and his students also
discovered counterconditioning, whereby an aversive
stimulus such as electric shock can be made less aver-
sive or even positive by pairing it with another stimulus
(e.g., food) that elicits a positive reaction. Pavlov and
Bechterev believed that conditioning or the formation
of what Bechterev termed association reflexes accounts
for all learning, including the learning of maladaptive
behavior (Pear 2007).
In what has been termed the behavioral revolution,
John B. Watson (1878–1958) adopted Pavlov’s termi-
nology, or at least a close approximation to it (e.g.,
“conditional reflex” is a more accurate translation of
the term that is usually translated from Russian as
“conditioned reflex”) although not his methodology.
Using methodology more similar to Bechterev’s – i.e.,
422 B Behavior Modification, Behavior Therapy, Applied Behavior Analysis and Learning
focusing on motor rather than salivary and other glan-
dular responses – Watson and his graduate student
Rosalie Rayner (1899–1935) demonstrated the condi-
tioning of fear by startling an infant while he observed
a white rat. Mary Cover Jones (1896–1987), a protégé
of Watson, demonstrated that a fear response in a child
could be eliminated through a counterconditioning
procedure in which the feared object (a white rabbit)
was introduced gradually while the child was engaging
in an incompatible behavior (eating). The following
generation of behaviorists focused mainly on animal
research in a field known as learning or learning theory,
where “learning” was generally thought of as another
name for conditioning (Pear 2001). From an applied
point of view, much theoretical effort was devoted to
demonstrating the compatibility of learning theory
with Freudian psychoanalysis. However, a clear break
between Freudian and learning theory occurred in the
writings and work of Hans J. Eysenck (1916–1997).
Through activities that included editing books on BT
and founding the journal Behaviour Research and Ther-
apy, Eysenck brought together a number of researchers,
such as Joseph Wolpe (1915–1997), who were
extending the work of Pavlov, Bechterev, Watson,
Rayner, and Mary Cover Jones, and members of the
next generation of learning theorists, such as Clark L.
Hull (1884–1952), to the treatment of clinical prob-
lems. Eysenck and others also provided strong empir-
ical arguments for the greater efficacy of BT relative to
other psychotherapeutic techniques (Kazdin 1979).
While BT stemmed from traditional learning the-
ory, BM has a broader base. It began as an extension of
the methodology of experimental psychology to clini-
cal problems without regard to specific theoretical
positions. For example, in an edited book in which
the term “behavior modification” was first used prom-
inently, R. I. Watson (1962) credited Carl Rogers
(1902–1987) with having “launched the research
approach in behavioral modification through psycho-
therapy” (p. 21). Rogers’ humanistic theory was at odds
with traditional learning theory.
The origin of ABA can be traced to B. F. Skinner’s
book The Behavior of Organisms: An Experimental
Analysis (Skinner 1938). Skinner (1904–1990),
a strong advocate of a conditioning/learning approach
to behavior, was heavily influenced by both Pavlov and
Watson. However, Skinner stressed the importance of
distinguishing between two types of conditioning:
Pavlovian, which works by the pairing of two stimuli,
and operant, which works by pairing a response with
a subsequent reinforcing stimulus, resulting in the
increased rate of the response. The emphasis of Skinner
and his followers on studying behavior in individuals,
on focusing on rate or probability of responding, and
eschewing theories that attempt to account for behav-
ior on the basis of nonbehavioral (e.g., mental) pro-
cesses led to the founding of the Journal of the
Experimental Analysis of Behavior. Skinner and many
of his followers stressed the application of the behavior
principles they were elucidating. This emphasis
included all areas in which human and animal behavior
occurs. The exponential growth of applied behavior
analytic studies led to the founding of the Journal of
Applied Behavior Analysis. Thus, behavior analysis sub-
sumes both ABA and the experimental analysis of
behavior (or basic behavior analysis). Behavior analysis
has grown into a controlled profession with the rise of
certifying bodies such as the Behavior Analysis Certifi-
cation Board.
In their systematic textbook, Martin and Pear
(2011) recommend that the term behavior modification
be used to refer to all applications (including those
under the rubric of ABA) of learning principles to
practical problems and that the term behavior therapy
be restricted to applications of learning principles to
clinical problems. BM according to these authors is
broader than both BT and ABA, and includes them.
Important Scientific Research and
Open Questions
The current and potential applications of BM are vast.
The numerous combinations and permutations of pro-
cedures within BTand ABA that can be applied give rise
to two overarching scientific tasks: (a) empirical deter-
mination of which procedure is most effective for each
type of application; and (b) the search for a unifying
theory that will indicate which type of procedure is
most effective for each type of application. Regarding
the first overarching task, each new type of problem
that behavior modifiers wish to address will require
extensive empirical study to determine the most effec-
tive procedures for that problem. Although more
ambitious, the second task, if its goal is achieved, will
allow the most effective procedures to be applied to
new problems. In addition, the second task will provide
a convenient rubric or mnemonic to facilitate the
 Behavior Modification, Behavior Therapy, Applied Behavior Analysis and Learning
training of new behavior modifiers and for established
behavior modifiers to consult when confronted with
a problem they may not have dealt with previously. At
this point, we are in the very early stages of embarking
on these two tasks, so it is not clear to what extent they
can or will be successfully completed.
At about the time that BT was being developed,
Albert Ellis (1913–2007) and Aaron Beck (1921-)
were independently developing an approach called cog-
nitive therapy. Both of these investigators focused on
debilitating or maladaptive thoughts in the etiology of
mental disorders, and proposed eliminating those
thoughts as the main goal of treatment. Ellis and Beck
had three things in common with behavior therapists:
(a) rejection of Freudian psychoanalysis with its focus
on the past events such as childhood memories and
peripheral phenomena such as dreams, (b) homework
assignments for dealing with maladaptive thoughts and
behavior, and (c) a focus on empirical outcomes as the
most important evidence for therapeutic success.
Under the influence of cognitive therapists and the
so-called cognitive revolution that was occurring
against behaviorism, many psychotherapists adopted
a blend of BT and cognitive therapy called cognitive
behavior therapy (CBT; or sometimes cognitive behavior
modification, CBM). There is a long-standing tradition
within conditioning theory, going as far back as Pavlov
and Bechterev, that speech (whether overt or covert) is
both behavior and stimuli for subsequent behavior.
Following this tradition, the issue arises as to whether
the cognitive component of CBT is (a) simply an addi-
tional behavioral procedure (as argued by, e.g., Martin
and Pear 2011) or a qualitatively different kind of
procedure; and (b) whether the cognitive component
of CBT provides any added therapeutic value. The first
issue, depending on how one views it, is either
a theoretical, philosophical, or semantic problem that
has yet to be resolved. Regarding the second issue,
however, the evidence so far indicates that the behav-
ioral component of CBT is the effective ingredient; the
cognitive component appears to provide little or no
additional therapeutic effectiveness.
If CBT constitutes a second wave within BT, it has
been argued that a third wave has emerged. This third
wave adds components such as experiential acceptance,
mindfulness (e.g., meditation), dialectical thinking,
interpersonal relationship building, and spirituality.
Clearly, the inclusion of these factors adds complexities
B 423
beyond those of straightforward cognition. Teasing all of
these factors out for component analyses – i.e., deter-
mining which of these new factors constitute effective B
therapeutic ingredients and for which specific psycho-
logical problems – will be a decades-long task. In addi-
tion, the question arises as to how or whether this third
wave will maintain contact with the basic conditioning or
learning theory literature that was so pivotal to the initial
development of BT. Of the third wave therapies, only
advocates of one – acceptance and commitment therapy
(ACT) – explicitly claim adherence to basic learning
theory findings and methodology.
ABA also has been applied to clinical problems,
although much less frequently than BT has. Typically
ABA starts by identifying, for each client, behaviors
that need to be decreased (problem behaviors) or
behaviors that need to be increased (desirable behav-
iors). Initial baselines are taken on those behaviors
(often but not necessarily singly), and procedures (typ-
ically based on current learning theory, functional ana-
lyses of conditions maintaining the problem behaviors,
and successful interventions in previous cases) to
decrease the problem behaviors and increase the desir-
able behaviors are implemented. Various methodolog-
ical designs – e.g., reversal designs, multiple-baseline
designs – are used to determine whether the interven-
tion was effective to the desired extent, and corrective
procedures are taken if it was not. ABA is not incom-
patible with the use of BT procedures, since the inter-
vention component of ABA can come from various
sources. Thus, ABA should be among the procedures
that are considered and compared to BT procedures in
the treatment of clinical problems.
As mentioned previously, ABA has been used with
a variety of nonclinical behaviors. These include aca-
demic and other classroom behavior, drug-dependent
behavior, medical compliance, sports and other skill-
related behaviors, parenting, community problems,
behavior in institutional settings, and rehabilitation.
Although generally successful wherever it has been
tried, ABA’s most widely recognized success to date
has been in the fields of autism and developmental
disabilities. Various government bodies have desig-
nated ABA as the only proven effective treatment
for these conditions. One factor hampering research
on ABA in other fields is that other approaches are
demonstrating some effectiveness, so the need to
consider new approaches is not seen as urgent as in
424 B Behavior State
the fields of autism and developmental disabilities.
Advances in genetics and neuroimaging provide addi-
tional areas to investigate. Research questions to ask
with respect to these new technologies concern
whether, for example, different components of BT and
ABA would be more effective with some individuals
than with others on the basis of their genetic or neuro-
logical constituents. In addition, the effects of different
BT and ABA procedures on various neural correlates
may provide theoretical insights regarding the pro-
cesses involved in these procedures.
Cross-References
▶ Behavioral Approaches to Instruction
▶ Conditioning
▶ Formal Learning Theory
▶ Operant Behavior
▶ Operant Learning
▶ Pavlovian Conditioning
▶ Physiological Homeostasis and Learning
▶ Psychology of Learning (Overview)
▶ Social Learning Theory
References
Kazdin, A. E. (1979). Fictions, factions, and functions of behavior
therapy. Behavior Therapy, 10, 629–654.
Martin, G., & Pear, J. (2011). Behavior modification: What it is and how
to do it (9th ed.). Upper Saddle River, NJ: Pearson Prentice Hall.
Pear, J. J. (2001). The science of learning. Philadelphia, PA: Psychology
Press.
Pear, J. J. (2007). A historical and contemporary look at psychological
systems. Mahwah, NJ: Lawrence Erlbaum.
Skinner, B. F. (1938). The behavior of organisms: An experimental
analysis. Cambridge, MA: B.F. Skinner Foundation.
Watson, R. I. (1962). The experimental tradition and clinical psy-
chology. In A. J. Bachrach (Ed.), Experimental foundations of
clinical psychology (pp. 3–25). New York: Basic Books.
Behavior State
▶ Alertness and Learning of Individuals with PIMD
Behavior Styles
▶ Adult Learning Styles
Behavior Systems and Learning
WILLIAM TIMBERLAKE
Department of Psychological and Brain Sciences,
Indiana University, Bloomington, IN, USA
Definitions
A Behavior System is an evolved, motivational system
expressed through a sequence of search modes (strate-
gies) and sensory-motor modules relating environmen-
tal stimuli and adaptive actions. As an example, Fig. 1
illustrates the Predatory Subsystem of the Feeding
System of a rat (rattus norvegicus), evolved by descent
and modified in individuals through chance, develop-
ment, and experience to serve the functions of
searching for, acquiring and ingesting food for pur-
poses of survival and reproduction. The system is orga-
nized by motivational modes (strategies), including:
general search, focal search, and handling-consuming.
Each mode is expressed through perceptual-motor mod-
ules (combinations of stimulus filters and action pat-
terns), selected by evolution and tuned by development
and learning to facilitate the discovery, capture, han-
dling, and ingestion of food, and often adapted to serve
differentially the functional modes of general search,
focal search, and handling-consuming.
1. Basic function(s): Searching for, capturing, and
ingesting food for survival and reproduction
2. ▶ Motivational modes (strategies): Functional
groupings of sensory-motor modules creating
sequences of general search, focal search, and han-
dling-consuming behaviors, often involving returning
and repeating portions of the sequence as
appropriate)
3. ▶ Sensory-motor modules: Combinations of
stimulus filters and action patterns selected by
evolution, but often tuned by learning to facilitate
food discovery, handling, and ingestion
▶ Learning is a stable, usually adaptive, change in the
expression of sensory-motor modules, motivational
modes, and regulatory functions, based on:
1. Repeated stimulus presentations: producing habit-
uation, sensitization, stimulus integration
2. Pavlovian conditioning procedures: repeated
pairings of a predictive stimulus with reward,
 Behavior Systems and Learning B 425

SUBSYSTEM MODE MODULE ACTION

TRAVEL LOCOMOTE B
SCAN
GENERAL SOCIALIZE CRAWL OVER
SEARCH SNIFF

INVESTIGATE NOSE
PAW
CHASE TRACK
CUT OFF
LIFE IN WAIT IMMOBILE

PREDATORY FOCAL POUNCE

SEARCH CAPTURE GRAB
BITE
TEST GNAW
HOLD
INGEST CHEW
SWALLOW
HANDLE/
CONSUME REJECT SPIT OUT
WIPE OFF
HOARD CARRY

Behavior Systems and Learning. Fig. 1 The predatory subsystem of the feeding behavior system of Norway rats consists
of three search modes: General, focal, and handling/consuming, which control stimulus-response modules related to
foraging actions. Bouts of the expression of foraging modes and modules generally start at the top of the diagram and run
downward, moving back toward the top as the food is lost, rejected, or consumed. Experiments presenting stimuli
predicting food typically engage behavior appropriate to the physical resemblance and typical and actual temporal
proximity of the cue to food

typically altering skeletal and autonomic responses
to the stimulus; for example, repeated presentations
of a brief tone to a hungry dog – each tone followed
by meat powder in its mouth – soon produces
reliable environmental orienting search behavior
and salivation during the tone.
3. (Discriminative) operant conditioning: repeatedly
following a lever press with reward (in the context
of a tone) typically increases the frequency and rate
of lever pressing in the presence of the tone.
Theoretical Background
The development of general learning paradigms and
procedures has resulted in general mathematical and
neurophysiological models of the learning process,
improved techniques of animal training, and clarified
the use of rewards in child rearing and education. What
is missing is the integration of the learning models with
evolution-based predispositions and constraints
pointed to by ethologists and other biologically ori-
ented researchers concerned with how learning
produces adaptive behavior. A behavior system
approach clarifies and suggests applications and limits
on the traditional strengthening and prediction-based
models dominating the study of learning. Specifically,
a behavior system approach focuses on stimulus sensi-
tivities, response forms, and processing and timing
characteristics influenced by the evolutionary ecology
of a species’ niche, and the specific experiences of an
animal.
The dominant approach to the study of laboratory
learning has focused on general, response-, or stimu-
lus-reinforcer, models in which learning is attributed to
the frequency and predictability of the pairing of stim-
uli and/or responses with a reinforcer (e.g., food, water,
safety, social contact, and mating). For example, in the
case of food reward, the well-studied Norway rat is
assumed to learn and perform a required response,
such as maze running or lever pressing, because these
responses differentially predict the receipt of food, and
often because the more rapidly and accurately the rat
performs the response, the faster the food arrives.
426 B Behavior Systems and Learning
A behavior system approach focuses on relating the
stimulus sensitivities, response forms, and timing char-
acteristics of learning to the evolution and individual
development of the members of a species. Specific
examples of species’ related learning include: retention
of components of adult male song by subadult male
songbirds (guiding later production) and by females
(guiding recognition of suitable mates); rapid imprint-
ing in young precocial birds related to following
a proximate moving stimulus; and the ability of mam-
malian omnivores to acquire one-trial avoidance of the
taste of any novel food followed by gastric distress
within 24 h.
The following sections provide specific examples of
how a feeding behavior system influences learning in
rats by clarifying how system components contribute to
learning in the common laboratory tasks of maze learn-
ing, autoshaping, and omission training.
The Role of Perceptual-Motor
Modules in Maze Learning in Rats:
Mazes as Trails
Although rats rapidly acquire efficient maze behavior
when food is presented in the goal box, two important
data sources question the sufficiency of this account.
The first is Calhoun’s (USPHS) monograph
documenting how a colony of wild rats rapidly estab-
lishes burrows and trails connecting the burrows with
foraging locations on the colony’s periphery. It is
important to note that the establishment of trails and
burrows is not limited to wild rats. A 1998 film by
Manuel Berdoy of Oxford (“The Laboratory Rat,
a Natural History”) documents the release of a sizeable
group of laboratory-born and reared rats on a farm in
Oxfordshire, England. Within a few days, these
multigenerational offspring of lab rats, like their wild
cousins in Calhoun’s work, began to establish burrows
and trails, interconnecting them with sources of food
and water – traversing them rapidly and showing
robust predator avoidance.
Of course, it might be argued that such bravura
performances were actually rapidly learned based on
consuming food after a successful run through the
trails. However, in the laboratory where feeding time
is well controlled, rats acquire efficient locomotion in
unbaited standard maze apparatus, including straight
alleys, radial arm mazes, and complex multi-choice
mazes, all in the absence of food reward, food
deprivation, systematic removal from a “goal,” and
with maze arms flat on the floor (see Timberlake
2002). Lest the reader be concerned that the apparatus
somehow imposed the behavior, it should be reported
that Kangaroo “rats” (an unrelated genus with a niche
in high desert areas of California) completely ignored
floor arms, edges, and walls in favor of “looping” out
from and back to a home base area (Timberlake 2002).
In short, although psychologists models have long
credited food deprivation and reward with producing
maze learning and traversal in Norway rats, it seems
apparent that Norway rats are evolutionarily selected to
learn and follow paths related to landmarks, whisker
contact of edges, and repeated exposures.
Pavlovian Procedures: Perceptual-
Motor Modules in Autoshaping
and Omission
A second data source questioning a general reinforce-
ment-learning model concerns the effects of applying
autoshaping and omission contingencies to the com-
mon instrumental responses of lever pressing in rats,
and pecking in pigeons directed at circular plastic
“keys.” It is widely assumed that experimenters picked
the responses of lever manipulation by rats and pecking
by pigeons because they are easy to measure and conve-
nient to teach by using food reinforcers to shape succes-
sive approximations to pressing or pecking. However, it
became clear in the early 1970s that this was not a full
account. In a phenomenon called “autoshaping,” food-
trained pigeons quickly began to peck a response key lit
several seconds before food was made available; similarly
rats presented with a lever inserted shortly before food
arrived, began to tug, bite, and manipulate the lever. In
both cases, food was presented on a fixed-time schedule,
independent of the animal’s behavior.
Although these data argue that manipulating a lever
or pecking a “key” have species-related aspects,
their development can still be comfortably explained
by assuming that occasional pecks or presses were
reinforced by accidental response contiguities with
food. However, when experimenters added to the food
schedule an “omission” component that canceled
food presentation if the animal contacted the lever or
key during the preceding 15 s interval, it became evident
that the accidental food proximity explanation of why
lever pressing or keypecking occurred was incorrect.
Under an omission contingency, food is prevented

from following recent presses (or pecks), so no acciden-
tal response shaping can occur. Nevertheless, responding
still occurs and persists. The conclusion best supported
is that contacting a moving lever in rats or a lighted key
in pigeons is based on species-based perceptual-motor
modules that underlie the evolution development, and
expression of their foraging behavior.
Combining the rat data showing rapid learning to
traverse a maze in the absence of food deprivation or
obvious reward, with the above data showing emer-
gence and persistence of lever or key contact under an
omission contingency (preventing a positive relation
between contact and food), argues for two conclusions:
(1) Scientists building experimental apparatus and
designing learning procedures for laboratory rats and
pigeons proved to be perceptive judges of the nature of
the stimulus-response organization (perceptual-motor
modules) underlying foraging in these species.
(2) These perceptual-motor modules are sufficiently
integrated that they are engaged without food, and
persist even when food is omitted when the animal
expresses them.
It is worth noting that the perceptual-motor mod-
ules constituting a rat foraging system are not restricted
to locomotion and food manipulation responses. For
example, a social investigation module can be engaged
in both juvenile and adult rats simply by presenting a
(constrained) adult rat as a predictor of food
(Timberlake 1990). A subject rat of any age will
approach and socially engage an adult rat (though not
a juvenile) rat predicting food, as well as an adult rat
randomly predicting food. These outcomes are predict-
able from work by B. G. Galef Jr. on the role of social
communication in the feeding system of rats. That
male hamsters will not approach adult male hamsters
predicting food (male hamsters establish individual
feeding territories) clarifies the unique social foraging
system of rats (Timberlake 2001).
Contributions of Motivational Modes
to Laboratory and Ecological Learning
As noted in Fig. 1, modes are motivational substates
that control modules as a function of their proximity to
reward. Predictive stimuli more distant temporally or
physically will support modules related to a general
search mode. For example, Timberlake (2001) reported
that rats under long fixed time (FT) food schedules
frequently ran in an available wheel while waiting for
Behavior Systems and Learning B 427
food to arrive. In the absence of the wheel, they more
often searched at a distance from the food tray and
drank excessive water. On the other hand, predictive B
stimuli temporally or physically more proximate to
food entrained focal search behaviors, such as tugging
on a protruding lever, or digging and nosing around
the food tray. Finally, when food was highly proximate,
behaviors related to a handling-consuming mode
appeared, such as prey capture, or digging/nosing vig-
orously in the food tray.
Silva and Timberlake (2005) summarized a number
of studies using a probe technique to document differ-
ent search modes for food. For example, if a lever is
inserted in the middle of the FT, the animal approaches
and interacts with the lever for a time before moving
toward the food tray. On the other hand, if the same
lever is inserted close to the end of the interval, the rat
more often moved directly to the food tray. Similarly,
a far-to-near sequence of two levers predicting food is
tracked in that order, but a near-to-far predictive
sequence is not tracked. Instead, the animal goes from
the near lever directly to the feeder.
Important Scientific Research and
Open Questions
Not only do behavior system modes, and modules
engaged by predictive cues or intervals vary with the
species, but also with the motivational system (e.g.,
feeding vs. mating), but they also vary with the species,
and sometimes with changes in the availability of
a resource. In all cases, the reinforcer is less
a “strengthener” of stimulus-response connections,
than a modifier of attention, a means of changing
relative attractiveness of behaviors and locations, and
a determinant of available repertoire. Further, species’
differences in reactions to the same predictive stimulus
can vary considerably based on how characteristics of
the stimulus fit with the set of available modules.
For example, Timberlake and Washburn (see
Timberlake 1989), basing their procedures on previous
work demonstrating predatory modules of rats
expressed toward a rolling ball bearing predicting
food, tested the reactions of seven rodent species to
the same stimulus predicting food. They predicted that
a species’ attention to and interaction with moving
bearing in the laboratory would be related to the
amount of prey found in their stomachs in the wild,
as well as to how species members dealt with live
428 B Behavior Systems and Learning
crickets placed in their home cages. Supporting these
predictions, behavior to the ball bearing varied directly
with the predatory behaviors of a species. Grasshopper
mice (O. leucogaster), and rats (R. norvegicus) attacked
the bearing; the Peromyscus species (P. leucopous and
P. californicus) ran alongside the bearing, biting at it;
while the herbivorous species, the cotton rat
(S. hispidus) “froze” as the bearing rolled by. Because
all these responses constitute misbehavior (to get food
each subject needed only to go to the food tray when
the bearing appeared), the data strongly support the
view that misbehavior reflects a locally inappropriate
engagement of species-typical components of a feeding
system.
For additional examples of how species and system
behavior systems interact with learning procedures, see
symposium papers listed under Timberlake and
Fanselow (1994) in the Bibliography, including work
by Fanselow on the rat defensive system, by Domjan
(and Akins) on the mating system of quail, and by
Hogan on chick development.
In short, the evidence reviewed and referenced
above supports the view that reliable laboratory learn-
ing phenomena in rats and pigeons stem from the
interaction of perceptual-motor modules and search
modes in the relevant behavior systems of a species
with the apparatus and tasks selected by experimenters.
This conclusion reflects an important, two-sided,
irony.
In the case of the laboratory experimentalists, there
is irony in the contrast between their focus on abstract
general models of response strengthening and cue
validity and their ability to design ecological learning
tasks that clearly engage the evolutionary behavior
systems of their subjects. It is remarkable that neither
keypecking by pigeons, nor maze performance or lever
manipulation by rats requires operant training using
response-contingent reward. Maze performance is pro-
duced by repeated exposure to the apparatus, and the
acquisition of lever manipulation by rats and
keypecking by pigeons requires only that movement
of the lever or illumination of the key signals the immi-
nent arrival of food. Rat manipulations of the lever and
pigeons pecking the key simply emerges and continues,
even if the response is under an omission contingency
and, thus, is never followed by a food reward.
The other side of the irony relates to the belief of
some biologists that laboratory rats and apparatus are
unnatural to begin with, and that psychologists make
matters worse by running (sleepy) rats during the day,
when (as a nocturnal species) they should be asleep. It
appears, though (as noted above), that the adaptive
behavior of laboratory rats released on a farm highly
resembles that of wild rats, providing good evidence of
their basic behavioral “reality” as rats. In addition,
psychologists clearly figured out how to circumvent
the issue of testing sleepy rats during the day by dis-
covering that rats organize their daily circadian activity
rhythm around the absolute time they are presented
with a significant amount of food (regardless of time
of day).
Finally, both behavioral ecologists and behavior
system researchers appear to agree that: (1) ecologically
selected and constrained learning can be a major con-
tributor to vertebrate behavior; and (2) a behavior sys-
tem approach shows important potential for
integrating the study of learning in the laboratory and
field into an analytic, robust, and ecologically mean-
ingful research program.
Cross-References
▶ Biological and Evolutionary Constraints of Learning
▶ Ecology of Learning
▶ Erotic/Sexual Learning
▶ Fear Conditioning in Animals and Humans
▶ Learning and Instinct
▶ Operant Learning
▶ Pavlovian Conditioning
▶ Place Learning and Spatial Navigation
▶ Selective Associations
▶ Spatial Learning
References
Silva, K. M., & Timberlake, W. (2005). A behavior systems view of
partially reinforced responding during an interfood clock. Learn-
ing & Behavior, 33, 99–110.
Timberlake, W. (1990). Natural learning in laboratory paradigms. In
D. A. Dewsbury (Ed.), Contemporary issues in comparative psy-
chology (pp. 31–54). Sunderland: Sinauer Associates.
Timberlake, W., & Fanselow, M. S. (1994). Symposium on behavior
systems: Learning, neurophysiology, and development.
Psychonomic Bulletin & Review, 1, 403–456, includes:
Timberlake, W. Behavior systems, associationism, and Pavlovian
conditioning. (405–420); Domjan, M. Formulation of a behavior
system for sexual conditioning. (421–428); Fanselow, M. Neural
organization of the defensive behavior system responsible for
fear. (429–438); Hogan, J. A. Structure and development of

behavior systems. (439–450); Shettleworth, S. Commentary:
What are behavior systems and what use are they? (451–456).
Timberlake, W. (2001). Motivational modes in behavior systems. In
R. R. Mowrer & S. B. Klein (Eds.), Handbook of contemporary
learning theories (pp. 155–209). Mahwah, NJ: Erlbaum.
Timberlake, W. (2002). Niche-related learning in laboratory para-
digms: The case of maze behavior in laboratory rats. Behavioural
Brain Research, 134, 355–374.
Behavior Therapy
▶ A Tripartite Learning Conceptualization of
Psychotherapy
Behavioral Approaches to
Education/Teaching
▶ Behavioral Approaches to Instruction
Behavioral Approaches to
Instruction
JOSEPH J. PEAR
Department of Psychology, University of Manitoba,
Winnipeg, MB, Canada
Synonyms
Behavioral approaches to education/teaching; Empiri-
cally supported methods of instruction; Evidence-
based teaching; Technology of teaching
Definition
Behavioral approaches to instruction have the following
properties: (1) the learning that is to occur is defined
behaviorally – i.e., in terms of observable, measurable
behavior; (2) the procedures used to produce the desired
learning are defined in terms of clearly defined opera-
tions; and (3) the procedures are evidence-based.
Theoretical Background
The theoretical roots of behavioral approaches to
instruction lie in the functionalist school of psychology
Behavioral Approaches to Instruction B 429
(Pear 2007). The mind, according to functionalists, was
an evolved entity whose existence depended on its
contribution to evolutionary survival. Three function- B
alists of special importance to the development of
behavioral approaches to instruction are William
James, John Dewey, and Edward L. Thorndike.
For William James (1842–1910) animals, including
humans, begin life with a set of instincts that are grad-
ually molded into habits. Beginning with the very
young child, the educator’s job is to modify these
habits. At some point the well-educated person learns
to control and modify his or her own habits. However,
James believed that education was an art to which the
psychology of his day could offer little.
John Dewey (1859–1952) actively applied James’s
functionalist approach (which James called pragmatism
and Dewey called instrumentalism) to education.
Dewey emphasized that learning is best viewed as an
active process in which thinking, doing, and problem
solving are closely intertwined. In short, he believed
that individuals learn best through active engagement
with their environments.
Edward L. Thorndike (1874–1949) was strongly
influenced by both James (under whom he received
his masters degree) and Dewey. Working with animals,
Thorndike demonstrated that three laws explain much
of what might appear to be the result of higher-order
faculties. The most important of these laws is the law of
effect which states that if an animal in a given situation
makes a response that is closely followed by an event
called a satisfier (e.g., food), a bond between that situ-
ation and response is strengthened so that the response
will be more likely to occur the next time the animal is
in that situation. Through his animal research, Thorn-
dike had synthesized the teachings of James and Dewey;
i.e., animals formed or modified habits (James)
through trial-and-error interactions with their envi-
ronments (Dewey). Thorndike extended his research
to education, viewing instruction largely as a trial-and-
error process in which S-R bonds are strengthened by
such satisfiers as the words “Right” and “OK.”
Initiating the behavioral revolution in psychology,
John B. Watson (1878–1958) advocated banishing
from psychology terms such as “mind” and “con-
sciousness” that were prevalent in functionalism and
other schools of psychology. Watson, reinterpreting
psychology as the science of behavior, discussed habit
without reference to consciousness. After becoming
430 B Behavioral Approaches to Instruction
familiar with the work of Ivan P. Pavlov (1849–1936)
and Vladimir M. Bechterev (1857–1927), Watson made
the conditioned reflex the basic unit of habit. Watson
argued that behaviorism opens the door to many
empirical applications, including “experimental peda-
gogy.” However, he made no contributions to this field,
probably because the conditioned reflex is not a good
model for achieving most instructional objectives.
Thorndike’s model provided a better starting point.
Sidney L. Pressey (1888–1979) sought to automate
the application of Thorndike’s laws of learning through
a device called a teaching machine. The basic mecha-
nism consisted of a metal case containing four or five
buttons for responding to multiple-choice questions.
The student read an assignment and then took
a multiple-choice test using the machine. Whenever
the student made a correct response, the pressure on
the correct button mechanically caused the machine to
advance to the next question. If the student made an
incorrect response, the machine did not advance and
the student was required to try again until he or she had
made the correct response. This procedure provided
the “stamping in” of the correct stimulus-response
connection. Importantly, the machine provided
a written record of the student’s performance for
addressing any problems the student had with the
material, for improving the teaching program, and for
research purposes. Pressey also developed a paper ver-
sion of his teaching machine. The student made choices
on a chemically treated paper using a pen with a special
type of red ink that turned black when the student
made the correct choice but remained red when the
student made an incorrect choice.
B. F. Skinner (1904–1990) developed a different
design for teaching machines based on his research on
operant conditioning (a concept very similar to
Thorndike’s law of effect). Rather than relying on mate-
rial from an external source as Pressey’s teaching
machine did, Skinner’s teaching machine presented
small segments of material called frames in a sequence
designed to facilitate learning of the material (Skinner
1958). At the end of each frame was a question or
a statement in which one or more words were blanked
out. The student’s task was to write the answer to the
question or the missing word(s) from the information
in the frame. The student then pulled a lever causing
the correct response to be shown and the student’s
response to move under a transparent cover where it
could be viewed but not changed. The frames were
designed and sequenced to minimize errors because
Skinner believed that (1) errors tend to generate sub-
sequent errors, (2) the aversive by-products of errors
interfere with the learning process, and (3) the
matching of the student’s response with the correct
response provided the positive reinforcement needed
to maintain the student’s continued responding on the
machine. Skinner eschewed the use multiple-choice
questions for at least three reasons: (1) he believed
that being able to generate correct answers is more
useful than simply being able to recognize them,
(2) multiple-choice questions expose students to incor-
rect information that could interfere with subsequent
learning, and (3) it is possible for a student to make
a correct response to a multiple-choice question with-
out being under the control of the relevant educational
variables. Similar to Pressey’s teaching machines, Skin-
ner’s teaching machines provided written records of the
students’ performance for addressing any problems
students had with the material, for improving the
teaching program, and for research purposes. Skinner
developed a paper version in the form of a book called
a programmed textbook. Presentation of material in the
manner of a Skinnerian-type teaching machine is called
programmed instruction (PI). Computerized versions of
PI, unlike Skinner’s teaching machines, tend to use
multiple-choice questions.
Ogden R. Lindsley (1922–2004), who received his
doctorate under Skinner, developed a behavioral
method of instruction called precision teaching that
focuses on increasing the student’s response rate on
basic concepts and skills before going to more complex
concepts and skills (Lindsley 1991). Teachers are
encouraged to chart student performance on a special
logarithmic graphing paper, called a celeration chart, as
a means of evaluating the effectiveness of their teach-
ing. Precision teaching is often combined with
a method called direct instruction (developed by Sieg-
fried E. Engelmann) whereby teachers are given scripts
to follow in demonstrating concepts and skills to stu-
dents, prompting responses from students, and provid-
ing positive reinforcement for correct responses.
Fred S. Keller (1899–1996), a colleague and friend
of Skinner, developed a method called personalized
system of instruction (PSI) (Keller 1968). In this method
the course material (which may be in the form of
a textbook) is broken down into small units and students

are provided with study questions on each unit. Students
proceed through the units at their own pace, demonstrat-
ing mastery of the material in each unit by passing a test
on the unit before proceeding to the next unit. Because of
the frequency of testing required, students (called proc-
tors) from a more advanced course are used to grade the
unit tests and provide feedback to the students on those
tests. Computer-aided personalized system of instruction
(CAPSI), a computerized version of PSI, selects proc-
tors (called peer reviewers) from within the same course
(Pear et al. in press). Serving as a proctor or peer
reviewer enhances the proctor or peer reviewer’s learn-
ing as well as the learning of the students whose unit tests
they grade. Data are archived for addressing any prob-
lems the student has with the material, for improving
the teaching method, and for research purposes.
The final behavioral instructional method to be
considered here is called interteaching (IT) (Saville
et al. 2011). In this approach, students receive a prep
guide at the end of each class consisting of study
questions relating to the next assignment. At the next
class meeting each student teams up with a peer to
discuss their answers to the questions on the prep
guide and to prepare a report on any difficulties they
encountered with the material. The instructor moni-
tors the peer discussions and at the beginning of the
next class presents a lecture on material with which
students had difficulty. The method also includes
frequent tests based on questions in the prep guides.
Important Scientific Research and
Open Questions
Wherever behavioral instructional methods have been
tested they have proven to be more effective than tradi-
tional methods, which have typically been lectures. Why
then have they not been more widely adopted? Some
possible reasons to investigate are: (1) teachers receive or
perceive pressure from administrators, colleagues, and/
or students to use traditional methods; (2) teachers are
more comfortable teaching by methods by which they
themselves have been taught and with which they are
familiar; (3) much of the teaching done using behavioral
methods is less conspicuous than the work done by
teachers using traditional methods thereby resulting in
less reinforcement and/or punishment for the teacher;
(4) the audience effect is more reinforcing to the teacher
when traditional methods are used than are the changes
in the students’ behavior that occur when behavioral
Behavioral Approaches to Instruction B 431
methods are used; (5) behavioral methods require
more advance preparation on the part of the instructor;
and (6) traditional methods may be perceived as B
providing more opportunities than behavioral methods
for the instructor to further develop his or her own
knowledge about the subject matter being taught.
Typically the dependent variable that has been
examined most extensively in comparing behavioral
instructional methods with traditional methods is the
learning of basic facts and concepts. The effects of
behavioral instructional methods on the higher levels,
including critical thinking and creativity, need also to
be examined more extensively.
Comparisons should also be made of different
behavioral methods in comparable educational set-
tings; e.g., comparisons of PSI with IT, or of PI with
precision teaching. These comparisons will be most
useful if each method is tested in its maximally effective
form with respect to the dependent variables being
examined. This will require an analysis of the compo-
nents of the different methods.
Some examples of component effects that are
important to examine are: (1) the effects of multiple-
choice versus short-answer type questions; (2) the
effects of deadlines for completing course components;
(3) the effects of peer discussions; (4) the effects of
instructor, teaching assistant, and peer-reviewer grad-
ing versus computer grading; (5) the effects of methods
designed to improve peer reviewing and peer discus-
sions; and (6) the reinforcing effects of the numerical
points given toward the final grade for various behav-
iors outlined in the teaching methods.
Hybrid behavioral instructional methods should
also be developed and examined for at least three
reasons: (1) some students may do better with one
method while other students do better with other
methods, (2) students may do better under a variety
of methods than using one single method, and
(3) some methods may be better for developing some
types or levels of knowledge or skills while others are
better at teaching other types of levels or skills.
Cross-References
▶ Behavior Modification, Behavior Therapy, Applied
Behavior Analysis and Learning
▶ Behavioral Skills Training and Skill Learning
▶ Behaviorism and behaviorist learning theories
▶ Collaborative Learning Supported by Digital Media
432 B Behavioral Capacity Limits
▶ Critical Thinking and Learning
▶ Evidence-Based Learning
▶ Learner-Centered Teaching
▶ Learning by Doing
▶ Mastery Learning
▶ Online Collaborative Learning
References
Keller, F. S. (1968). “Good-bye teacher. . .”. Journal of Applied Behavior
Analysis, 1, 79–89.
Lindsley, O. R. (1991). Precision teaching’s unique legacy from
B F. Skinner. Journal of Behavioral Education, 1, 253–266.
Pear, J. J. (2007). A historical and contemporary look at psychological
systems. Mahwah: Lawrence Erlbaum.
Pear, J. J., Schnerch, G. J., Silva, K. M., Svenningsen, L., & Lambert, J.
(in press). Web-based computer-aided personalized system of
instruction. In W. Buskist, & J. E. Groccia (Eds.), Evidence-Based
Teaching. New Directions for Teaching and Learning, (Vol. 128).
San Francisco: Jossey-Bass.
Saville, B. K., Lambert, T., & Robertson, S. (2011). Interteaching:
Bringing behavioral education into the 21st century. The Psycho-
logical Record, 61, 153–166.
Skinner, B. F. (1958). Teaching machines. Science, 128, 969–977.
Behavioral Capacity Limits
SCOTT WEEMS1, JAMES REGGIA2
1
Center for Advanced Study of Language, University of
Maryland, College Park, MD, USA
2
Department of Computer Science, University of
Maryland, College Park, MD, USA
Synonyms
Scope of attention; Short-term memory capacity
Definition
Behavioral capacity limits refer to the resource-limited
nature of human information processing. While this
term can be used to refer to limitations in such infor-
mation-processing phenomena as sensory encoding,
short-term memory, and response selection, it is gen-
erally accepted to apply to any cognitive process involv-
ing a bottleneck of available resources. The most widely
referred to behavioral capacity limit involves short-
term memory capacity, which is believed to be
constrained to approximately four items, although
the nature of this limitation and its relationship with
attention is still an active source of debate.
Theoretical Background
The distinction between limited and unlimited capac-
ity components of memory has existed for at least
a century, ever since William James spoke of the differ-
ence between primary and secondary memory (1890).
Secondary memory, also known as long-term memory,
is assumed to be nearly infinite, or at least so large that
it is not likely to be exhausted during a human lifetime.
On the other hand, primary memory, otherwise known
as short-term memory, is much smaller. In fact, short-
term memory was first believed to be limited to seven
items (Miller 1956). Later, however, when factors such
as chunking and other strategic influences were con-
sidered, that number was found to be closer to four
items (Cowan 2000). This introduces an important
question: why do certain kinds of memory experience
such limitations but others do not?
Although processing limitations are still poorly
understood, one likely interpretation is what psychol-
ogists call a cognitive bottleneck. A bottleneck occurs
when some cognitive resource is needed for more than
one task, leading to a disruption of performance when
more than one cognitive process calls upon that resource
(Pashler 1993). Bottlenecks occur at all stages of cogni-
tive processing, not just memory. For example, bottle-
necks have been observed during detection of novel
visual stimuli; when people consciously identify
a visual stimulus, they may be unable to detect
a second stimulus for up to 500 ms after the first.
This effect is called attentional blink (AB), because
capacity to detect the second stimulus is briefly delayed
by the first. Another example of a resource bottleneck is
called the psychological refractory period (PRP). This is
observed when subjects perform two sensorimotor
tasks in rapid succession, with the second task often
demonstrating delayed responses due to lingering
resources still being controlled by the first.
Although the nature of the limited resource which
leads to the bottleneck in each of these situations is
poorly understood, with many simply calling it
a “resource” and others (Broadbent 1971) calling it
“an information processing channel with limited
capacity,” it is widely believed that attentional limita-
tions play a key role. For example, data support that the
PRP results from the postponement of attention-

demanding stages of the second task which are shared
and with the first. Attention also acts as an important
bottleneck for short-term memory; specifically, Cowan
has suggested that attention, by being limited in nature
and being a gateway to conscious awareness, deter-
mines the maximum number of items that can be
held in memory at any given time. Still, this leaves
many questions unanswered. For example, what stages
of visual processing are capacity unlimited (not
attentionally demanding) and what stages are capacity
limited (require attention)? Does attention limit short-
term memory prior to or during encoding? At what
stage of response selection does an attentional bottle-
neck occur?
Important Scientific Research and
Open Questions
Donald Broadbent was one of the first researchers to
use an information-processing approach to address
bottlenecks. His Filter Model proposed that informa-
tion from one’s environment is filtered early in the
perceptual process based on its physical characteristics,
with the filtering allowing the person to focus on spe-
cific aspects of the environment, albeit often at the
expense of others (and therefore introducing capacity
limitations). However, subsequent research demon-
strating that people still identify irrelevant and
unattended stimuli required the model be revised, lead-
ing Anne Treisman to propose her Attenuation Model
(1964). While this model also postulated an informa-
tion-processing bottleneck, it proposed one which
attenuates rather than eliminates those aspects of the
incoming signal which do not meet the filter criteria. In
this way, meaningful information can still reach atten-
tion provided it is salient enough to pass through the
filter. Lastly, much like Treisman’s Attenuation Model,
late selection models have also been developed to
explain the important role of arousal and stimulus
salience in attentional filtering. However, in contrast
to Treisman’s and Broadbent’s models, late selection
models assume that all stimuli are processed to the
level of their semantic meaning before relevance deter-
mines what reaches conscious awareness (Deutsch and
Deutsch 1963).
Today, many questions remain regarding not just
the timing of processing bottlenecks, but also their
physiological underpinnings and their unitary nature.
For example, some researchers have questioned
Behavioral Capacity Limits B 433
whether capacity limitations result from a single serial
bottleneck, a structure that queues items one at a time
for processing, or some abstract resource (e.g., B
processing device, storage units, communication
device, or energy) being distributed unequally between
tasks. Other researchers have argued that stimulus-
dependent influences on AB indicate that a single
resource cannot explain the effect, with some propos-
ing an alternate interpretation involving multiple chan-
nels of interference leading to processing capacity
limitations. Studies of the neurobiological nature of
the AB, PRP, and visual working memory (VWM)
have also shown that while AB and PRP share common
physiological sources in the lateral frontal/prefrontal
cortex, VWM appears more relatively localized to the
posterior parietal and occipital cortex. So it is possible
the bottlenecks associated with each may share not just
different neural loci, but also different cognitive infor-
mation-processing sources as well. Lastly, recent com-
putational work in the area of capacity limits has
suggested that such limits may not result from some
limited resource at all, at least with short-term mem-
ory, but simply be an emergent property resulting from
decay and interference in the human memory system
(Weems et al. 2009).
Cross-References
▶ Capacity Limitations of Memory and Learning
▶ Short-Term Memory and Learning
▶ Working Memory
References
Broadbent, D. (1971). Decision and stress. London: Academic.
Cowan, N. (2000). The magical number 4 in short-term memory:
A reconsideration of mental storage capacity. The Behavioral and
Brain Sciences, 24, 87–185.
Deutsch, J., & Deutsch, D. (1963). Attention: some theoretical con-
siderations. Psychological Review, 70, 80–90.
James, W. (1890). Principles of psychology. New York: Holt.
Miller, G. (1956). The magical number seven, plus or minus two:
some limits on our capacity for processing information. Psycho-
logical Review, 63, 81–97.
Pashler, H. (1993). Doing two things at the same time. American
Scientist, 81, 48–55.
Treisman, A. (1964). Verbal cues, language, and meaning in selective
attention. The American Journal of Psychology, 77, 206–219.
Weems, S., Winder, R., Bunting, M., & Reggia, J. (2009). Running
memory span: a comparison of behavioral capacity limits with
those of an attractor neural network. Cognitive Systems Research,
10, 161–171.
434 B Behavioral Cloning

an infant’s behavior to investigate infants’ understand-

Behavioral Cloning ing of the various components of language. Although
▶ Model-Based Imitation Learning the methodologies were initially developed to study
infant speech/phonemic perception, variations have
been used to study infants’ understanding of semantic
and grammatical components of language. The pre-
Behavioral Depression dominant methodologies include measuring infant
sucking rates and using infant orienting, gaze, and
▶ Learned Helplessness head-turn behaviors. Several versions of these method-
ologies have also used operant conditioning of specific
behavioral responses to demonstrate perceptual dis-
crimination in infants. The specific methodology
Behavioral Disorders
employed often depends on two constraints: (1) the
▶ Achievement Deficits of Students with Emotional age group under investigation (older infants possess
and Behavioral Disabilities many more measureable behavioral responses in
comparison to newborns) and (2) the specific nature
of the research question being examined (e.g., percep-
tual discrimination or perceptual preference).
Behavioral Engineering
▶ Behavior Modification, Behavior Therapy, Applied
Theoretical Background
The development of behavioral methodologies for the
Behavior Analysis and Learning
study of language development during infancy
stemmed from interest in psychoacoustic theories of
speech perception, which suggested that auditory per-
Behavioral Flexibility ception was a key factor in the development of speech
perception, and thus, language acquisition. Psycho-
▶ Impulsivity and Reversal Learning acoustic approaches involved examining the absolute
hearing sensitivities of listeners by manipulating stim-
uli and task factors. In the 1970s, the use of psycho-
acoustic theories and methodologies were used to
Behavioral Inhibition examine phenomena such as categorical perception of
▶ Neural Mechanisms of Extinction Learning and speech sounds and to test claims that specialized lan-
Retrieval guage mechanisms existed and were unique to the
human brain. In this way, behavioral methodologies
of infant language acquisition have evolved from basic
psychoacoustic paradigms.
Behavioral Methodologies in The high-amplitude sucking procedure has been
Infant Language Acquisition used to test perceptual discrimination abilities of new-
born infants in a variety of speech- and language-related
CHRISTINE D. TSANG studies. In this paradigm, a pacifier is connected to
Department of Psychology, Huron University College a pressure transducer, which in turn is connected to
at the University of Western Ontario, London, ON, a computer. The computer measures how often the
Canada infant produces a hard suck. In one version of
the sucking paradigm, infants are habituated to
a particular stimulus (e.g., the sound/ba/) such that
Definition each time the infant makes a hard suck, the stimulus is
Behavioral methodologies in infant language acquisi- presented. When the sucking rate decreases to some
tion generally involve measuring observable changes in habituation criterion (e.g., sucking rate decreases
 Behavioral Methodologies in Infant Language Acquisition
to 50% of initial rate), infants are presented with either
the same stimulus as heard during the habituation
phase, or the infant is presented with another novel
stimulus (e.g.,/da/). If the infant’s sucking rate increases
to the presentation of the novel stimulus but not to the
familiar stimulus, the infant detects the change in
the stimulus. Studies using the high-amplitude sucking
procedure have shown that young infants possess many
rudimentary speech perception abilities, such as
categorical perception of speech sounds in 2-month-
old infants.
Observation of infant orienting responses is also
a common methodology typically used with younger
infants (newborn to 4- or 5-month-olds). This para-
digm utilizes the orienting response normally displayed
by infants when attending to a new or novel event. The
specific response observed varies across infants, but
may include a momentary cessation of ongoing behav-
ior, orienting of eyes or head toward a new stimulus, or
changes in physiological measures, such as heart rate
deceleration. Infants’ gaze or looking behavior can also
be used as an assessment of language sensitivity. This
multimodal procedure measures the length of time
infants visually attend to a particular visual stimulus
while listening to another auditory stimulus. Variants
of this procedure include presentation of two different
visual images side by side while a single auditory stim-
ulus is played and measuring how long the infant
attends to one image or the other while the sound
plays, or presenting a single visual image for visual
fixation (e.g., checkerboard pattern) and presenting
two different auditory sound stimuli alternately and
measuring how long infants attend to the visual target
while one sound is presented in comparison to the
other sound. Generally, this paradigm has been utilized
to measure infants’ perceptual preference for
a particular stimulus; however, it can be modified
into a habituation/dishabituation procedure relatively
easily to measure discriminatory abilities. The head-
turn preference procedure is also a common method-
ology in studies of infant language discrimination, for
use with older infants capable of making a controlled,
voluntary head-turn (6 months and older). In this
paradigm, infants are seated (on a caregiver’s lap)
with an experimenter seated (or a video camera) facing
the infant. At the start of a trial, a visual target (e.g.,
a flashing light) located on one side of the infant causes
the infant to look to that side. When the infant looks to
B 435
that side, a sound stimulus is presented from a sound
speaker located behind the visual target. When the
infant looks away for more than 2 s, the sound presen- B
tation ends. The infant’s attention is once again
brought to the center, and the next trial begins with
the same visual target flashing on the other side of the
infant. When the infant looks to that side, a different
sound stimulus is presented and continues to play as
long as the infant attends to that side. A difference in
looking time to the two sides in response to one sound
stimulus in comparison to the other sound stimulus is
interpreted as an indication that the infant discrimi-
nates between the two sounds and that the infant pre-
fers one over the other.
Important Scientific Research and
Open Questions
The high-amplitude sucking procedure has shown that
newborn infants are capable of discriminating many
different speech sounds (e.g., Eimas et al. 1971). This
technique has also demonstrated that newborns have
learned some characteristics of maternal speech pat-
terns in utero. For example, newborn infants are not
only familiar with and are able to discriminate between
some characteristics of their mother’s speech patterns,
but generally prefer maternal voice and speech charac-
teristics that were learned while in utero (DeCasper and
Fifer 1980; Mehler et al. 1988). More recently, studies
using the sucking procedure have found that newborn
infants are highly sensitive to rhythmic and intona-
tional (prosodic) features of speech (Nazzi et al. 1998).
The use of the orienting response method has
been primarily used to obtain auditory threshold
measures across several different age groups of infants
(see Werner 1992), but has also been employed to
examine 2–4-month-old infants’ ability to localize
sounds (Morrongiello et al. 1990). Gaze- or looking-
based procedures have been used successfully to dem-
onstrate many aspects of early speech perception and
semantic development. For example, at 2 months of
age, infants can associate facial articulations with pro-
duced sounds (Kuhl and Meltzoff 1982), by 4 months,
infants are able to integrate visual and auditory infor-
mation in speech perception (Werker and Desjardins
2004) and by 6 months of age, infants have already
made an association between the sound patterns of
words and their meaning (Tincoff and Jusczyk 1999).
Head-turn methodologies have been used to
436 B Behavioral Methodologies in Infant Language Acquisition
demonstrate that 6–8-month-old infants are able to
segment word in fluent speech using only frequency-
based cues found in the statistical structure of speech
(e.g., Saffran et al. 1996), and by 14 months of age
infants are capable of learning the association between
new words and new objects (Werker et al. 1998).
The use of operant reinforcement procedures with
sucking and head-turn measures have yielded much
information about infants’ early language abilities.
Indeed, one of the variants of the high-amplitude suck-
ing procedure operantly conditions infants’ sucking
rates to show both perceptual discrimination and per-
ceptual preference for maternal features of speech. In
this version, infants are reinforced for either a high rate
of sucking (fast sucking) or a low rate of sucking (slow
sucking) by the presentation of either a reinforcing
stimulus (e.g., mother’s voice) or a non-reinforcing
stimulus (e.g., strange female voice).
For older infants (6 months and older), the operant
conditioned head-turn paradigm is often used to dem-
onstrate infants’ perceptual discrimination abilities. In
this paradigm, the infant is presented with a sound
stimulus (e.g.,/ba/) for which the orienting response
to the sound is rewarded usually by some visual rein-
forcement, such as a moving mechanical toy, or
a flashing light. Thus, the sound stimulus becomes
a discriminatory stimulus that predicts a reward, such
that if the infant has learned the association, the infant
should make a head-turn as soon as the sound stimulus
is heard. The difference between sound stimuli can be
made very small during the course of testing to exam-
ine infants’ thresholds for sound discrimination. This
method has been used to successfully examine infants’
abilities to discriminate between vowel categories as
well as infants’ discrimination of native language
speech contrasts compared to nonnative language
speech contrasts.
It is clear that the use of infant behavioral method-
ologies over the past 40 years has provided much infor-
mation about perceptual capacities that have effects on
both language and cognitive development during the
first 2 years of life. Modern psychologists are now
exploring the use of electrophysiological and neuroim-
aging techniques such as EEG, ERP, and fMRI, which are
increasingly common in studies examining infant lan-
guage and cognition. Indeed, many researchers are
attempting to combine electrophysiological and behav-
ioral methods to elucidate a more complete picture
of infant cognitive capacity. However, behavioral
methodologies such as the ones described here remain
the predominant research methodology for perceptual,
cognitive, and language research with infant
participants.
Cross-References
▶ Acoustic and Phonological Learning
▶ Audiovisual Learning
▶ Conditioning
▶ Developmental Cognitive Neuroscience and
Learning
▶ Habituation
▶ Habituation in Infant Cognition
▶ Language Acquisition and Development
▶ Phonological Representation
▶ Speech Perception and Learning
▶ Statistical Learning in Perception
References
DeCasper, A. J., & Fifer, W. P. (1980). Of human bonding: Newborns
prefer their mothers’ voices. Science, 208, 1174–1176.
Eimas, P. D., Siqueland, E. R., Jusczyk, P., & Vigorito, J. (1971). Speech
perception in infants. Science, 171(968), 303–306.
Kuhl, P. K. (1979). Speech perception in early infancy: Perceptual
constancy for spectrally dissimilar vowel categories. Journal of the
Acoustical Society of America, 66, 1668–1679.
Kuhl, P. K., & Meltzoff, A. N. (1982). The bimodal perception of
speech in infancy. Science, 218, 1138–1141.
Mehler, J., Jusczyk, P. W., Lambertz, G., Halsted, N., Bertoncini, J., &
Amiel-Tison, C. (1988). A precursor of language acquisition in
young infants. Cognition, 29, 144–178.
Morrongiello, B. A., Fenwick, K. D., & Chance, G. (1990). Sound
localization acuity in very young infants: An observer-based
testing procedure. Developmental Psychology, 26, 75–84.
Nazzi, T., Bertoncini, J., & Mehler, J. (1998). Language discrimination
by newborns: Towards an understanding of the role of rhythm.
Journal of Experimental Psychology: Human Perception and Per-
formance, 24, 756–766.
Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learning
by 8-month-old infants. Science, 274, 1926–1928.
Tincoff, R., & Jusczyk, P. W. (1999). Some beginnings of word com-
prehension in 6-month-olds. Psychological Science, 10, 172–175.
Werker, J. F., Cohen, L. B., Lloyd, V. L., Casasola, M., & Stager, C. L.
(1998). Acquisition of word-object associations by 14-month-
old infants. Developmental Psychology, 34, 1289–1309.
Werker, J. F., & Desjardins, R. N. (2004). Is the integration of heard
and seen speech mandatory for infants? Developmental Psycho-
biology, 45, 187–203.
Werner, L. A. (1992). Interpreting developmental psychoacoustics. In
L. A. Werner & E. W. Rubel (Eds.), Developmental psychoacous-
tics. Washington: American Psychological Association.

Behavioral Mimicry
▶ Mimicry in Social Interaction: Its Effect on Learning
Behavioral Pathology
▶ Psychopathology of Repeated (Animal) Aggression
Behavioral Skills Package
▶ Behavioral Skills Training and Skill Learning
Behavioral Skills Training and
Skill Learning
NANCY DIB, PETER STURMEY
Department of Psychology, Queens College and
The Graduate Center, Flushing, NY, USA
Synonyms
Behavioral Skills Package; Modeling, Rehearsal, and
Feedback
Definition
Behavioral Skills Training (BST) is a training package
that utilizes instructions, modeling, rehearsal, and
feedback in order to teach a new skill. Typically training
is implemented not for some fixed time, but rather to
some predetermined criterion. For example, a trainee
may be said to have acquired a skill when they have
emitted correct responses on 90% of three consecutive
training sessions. Although these four components
are common there are many procedural variations in
how researchers and practitioners apply them. For
example, modeling might be done live, in role play, or
through video-modeling. Feedback might be given
immediately or delayed, graphically or verbally, or in
combinations.
Behavioral Skills Training and Skill Learning B 437
Theoretical Background
BST comes from a behavioral background. Sometimes
research presents it in a relatively atheoretical manner as B
a strategy that appears to be robustly effective in skills
teaching. Applied behavior analysis approaches are likely
to be more explicit as to the possible functions of the
four components of BST. For example, instructions and
modeling may function as antecedent stimuli, rehearsal
functions as an opportunity to emit behavior, and feed-
back may function as consequences for trainee behavior.
That said evidence of the functions of each component is
minimal. For example, although feedback might at first
blush appear to function as a positive reinforcer for
correct responses, that is not the only possible function.
For example, it might function as a punisher for trainee
errors or as an antecedent stimulus for the next oppor-
tunity for the trainee to respond.
The issue of generalization and maintenance of
performance is important to behavioral skills training.
Training on one set of stimuli and responses should
produce desired behavioral effects with untrained stim-
uli and on untrained responses. For example, learning
to sight read music on a set of practice pieces should
produce accurate playing on new pieces of music with-
out practice (Dib and Sturmey 2011) or practicing
teaching one skill during role play should produce
accurate teaching on many teaching programs with
children with autism spectrum disorders (Ward-
Horner and Sturmey 2008). Change agents should
carefully program generalization through strategies
such as sampling multiple stimuli.
Important Scientific Research and
Open Questions
BST has been effective in teaching a wide variety of
skills. This has included teaching posture (Dib and
Sturmey 2007a) and sight-reading skills (Dib and
Sturmey 2011) to flute players as well as social skills
to people with and without various disabilities
(Frederiksen et al. 1976; Stewart et al. 2007.) It has
also been used to teach caregivers of children and adults
with disabilities to teach skills to the individuals they
work with. For example, Dib and Sturmey
(2007b) used BST to improve staff implementation of
discrete-trail teaching while working with children
with autism. Not only did the staff implementation
improve, but the stereotypical behavior of the
children they were working with decreased at the
438 B Behaviorism
same time. Although BST has been used for adults and
children with autism and other developmental disabil-
ities, the technique is just as useful in teaching new
skills to the general population.
There are several areas of active investigation. First,
it is unclear which of the four components of BST is
effective in producing behavioral change. Second,
although BST is relatively efficient compared to
instruction-based training, future research should also
develop more efficient training formats. These might
include group training, combinations of modeling and,
where necessary, feedback, and Web-based training.
Cross-References
▶ Behavior Modification as Learning
▶ Evidence-Based Learning
▶ Example-Based Learning
▶ Feedback and Learning
▶ Feedback in Instructional Contexts
▶ Learning by Doing
▶ Learning from Failure
▶ Learning in Autism
▶ Learning Skills
▶ Learning Tasks
▶ Model-Based Learning
▶ Model-Based Teaching
▶ Modeling and Simulation
▶ Models and Modeling in Science Learning
References
Dib, N. E., & Sturmey, P. (2007a). The effects of verbal instruction,
modeling, rehearsal, and feedback on correct posture during
flute playing. Behavior Modification, 31, 382–388.
Dib, N. E., & Sturmey, P. (2007b). Reducing student stereotypy by
improving teachers’ implementation of discrete-trial teaching.
Journal of Applied Behavior Analysis, 40, 339–343.
Dib, N. E., & Sturmey, P. (2011). The effect of general-case training,
instructions, feedback, and rehearsal on the reduction of sight-
reading errors of competent musicians. Journal of Applied Behav-
ior Analysis, 44, 599–604.
Frederiksen, L. W., Jenkins, J. O., Foy, D. W., & Eisler, R. M. (1976).
Social-skills training to modify abusive verbal outbursts in
adults. Journal of Applied Behavior Analysis, 9, 117–125.
Stewart, K. K., Carr, J. E., & LeBlanc, L. A. (2007). Evaluation of family-
implemented behavioral skills training for teaching social skills to
a child with asperger’s disorder. Clinical case Studies, 6, 252–262.
Ward-Horner, J. C., & Sturmey, P. (2008). The effects of general-case
training and behavioral skills training on the generalization of
parents’ use of discrete-trial teaching, child correct responses, and
child maladaptive behavior. Behavioral Interventions, 23, 271–284.
Behaviorism
A psychological movement which marked a move
away from descriptive studies toward a more scientific
analysis of observable and measurable behavior.
Emphasized habit formation and the use of reinforce-
ment to condition new behavior. Leading proponents
were Pavlov (classical conditioning) and Skinner
(Operant conditioning) in which behavior was modi-
fied by associations and habit formation.
Cross-References
▶ Behaviorism and Behaviorist Learning Theories
Behaviorism and Behaviorist
Learning Theories
DENIS C. PHILLIPS
School of Education and Department of Philosophy,
Stanford University, Stanford, CA, USA
Synonyms
Behaviorism; Science of behavior
Definition
Behaviorism is the name given to several approaches to
psychology, especially to the study of both animal and
human learning, which arose in – and flourished during –
the twentieth century. These approaches rejected the use
of introspective methods (wherein individuals reported
on their subjective experiences), and instead were based
upon the study of behavior, its modification, and its
observable antecedents and consequences – which were
taken to be the only scientifically objective, publicly
observable, sources of data. Consequently, behaviorists
rejected characterizations of psychology that were given
in terms of the study of mental events or of consciousness
or “mind,” and instead defined psychology as the
study of behavior and its modification. This focus, the
behaviorists felt, brought unity to the psychological study
of animals and humans.
Theoretical Background
The emergence of behaviorism as a broad movement
can be traced back to a variety of rich sources in the

eighteenth and nineteenth centuries. The British
empiricist philosopher John Locke (1632–1714) held
that “simple ideas” arose only from sense-experience,
but that the human mind had the capacity to combine
these in various ways to form “complex ideas”; this
philosophy eventually led to more detailed study of
the specific mechanisms of “association of ideas” and
thus to the birth of the psychological school of associ-
ationism. Some comparative psychologists working on
animal learning in the second half of the nineteenth
century seemed to conceive of themselves as studying
association – usually not of ideas per se, but of stimuli
and responses. (The ideas harbored by animals, of
course, cannot be directly accessed by researchers, but
Romanes, Lloyd Morgan, and others thought that they
could argue by analogy with human experience to
throw tentative light at least on the “inner” experience
being undergone by animals, especially those that were
close to humans on the phylogenetic scale.) The use of
associationist language spilled over to some of the early
work of behaviorists.
As suggested above, another important influence
was the work of the comparative psychologists (i.e.,
those psychologists who were interested in animal psy-
chology, and its relation – if any – to human psychol-
ogy). This domain of work became truly viable
following the publication of Darwin’s Origin of Species
in 1859, for Darwin was taken to have established the
“genetic continuity” of humans and animals – he
established mankind as part (no doubt in important
ways novel, but a part nonetheless) of the animal king-
dom. This in turn entailed that human capacities could
be found in more primitive forms in animals lower on
the evolutionary scale (for the theory of evolution held
that complex features did not spontaneously appear
but gradually evolved from simpler forms), and thus
the nature and functioning of these complex capacities
could be illuminated by close study of lower animals. In
short, the evolutionary argument led to the insight that
an understanding of human psychology could be
gleaned from the study of animals. However, there
was a related methodological principle that became
known as “Lloyd Morgan’s canon” (1894). It did not
take long for this to be called a “principle of parsimony,”
paralleling the one formulated by the medieval logician
and known to posterity as “Ockham’s razor.” (Recently
this identification of Morgan’s canon with Ockham’s
razor has generated a great deal of controversy.)
Behaviorism and Behaviorist Learning Theories B 439
Whatever its status in the history of Western thought,
the canon clearly had an impact on subsequent psycho-
logical work, including that of the behaviorists: “In no B
case may we interpret an action as the outcome of the
exercise of a higher psychical faculty, if it can be
interpreted as the outcome of the exercise of one which
stands lower in the psychological scale.”
The final influence worthy of mention here is the
long tradition of work (and speculation) concerning
reflexes and reflex action. For many researchers, from
the time at least of Descartes down to that of Pavlov,
reflexes were regarded as the basic psychological unit,
and animals often were conceived of as being “autom-
ata” that were bundles of reflexes. However, it was
Pavlov’s discovery that conditioning can take place
that was crucial in the history of behaviorism. As is
well-known, Pavlov discovered in the early years of the
twentieth century that a naturally occurring reflex – the
stimulus of the sight of food producing the response of
secretion of digestive fluid in a hungry animal – could
have another stimulus associated with it (e.g., the ring-
ing of a bell, if the bell was rung at the same time the
food was presented), so that eventually the so-called
conditioned stimulus (the sounding of the bell) could
itself, alone, produce the response of secretion. This
phenomenon of so-called stimulus substitution
became central in the work of one of the key founders
of behaviorism, John B. Watson.
From the sources discussed above, two different
forms of behaviorism emerged in the early decades of
the twentieth century. But it was a public address by
Watson, subsequently published in the Psychological
Review (1913), that formally gave birth – and name –
to the general movement. His opening lines were forth-
right in the extreme:
" Psychology as the behaviorist views it is a purely objec-
tive experimental branch of natural science. Its theo-
retical goal is the prediction and control of behavior.
Introspection forms no essential part of its methods,
nor is the scientific value of its data dependent upon
the readiness with which they lend themselves to inter-
pretation in terms of consciousness. The behaviorist, in
his efforts to get a unitary scheme of animal response,
recognizes no dividing line between man and brute.
(Watson 1913/1948, p. 457)
Watson treated human learning – acquisition of
behaviors – in terms of Pavlovian conditioning,
440 B Behaviorism and Behaviorist Learning Theories
whereby a new stimulus substitutes for a natural stim-
ulus in producing a response. In a notorious book
Psychological Care of Infant and Child (1928) he
asserted to parents that their modes of child-rearing
were entirely responsible for both the good and the bad
features their children developed, the key being the
regimens of conditioning that they had unwittingly
adopted. How did a child learn to cry or otherwise
misbehave at bedtime, for example? Probably because
of some mechanism such as the following: a loud noise
is the natural stimulus that produces a startled response
or fear; on some occasion after putting the infant into
bed, a parent may have slammed the bedroom door at
the same moment as turning out the light; the noise of
the slamming door is a natural stimulus that produces
fear (exhibited in the form of loud crying), but the
turning out of the bedroom light, which occurred at
the same time, has been conditioned and subsequently
substitutes for the natural stimulus – the result being
that whenever the bedroom light is doused, the infant
cries! Watson called such conditioned reflexes “home-
made fears.” It is apparent that Watson was
a determinist – by controlling the environmental con-
ditions under which a child developed, and thus by
manipulating the conditioning that occurred, the
nature of the child could be determined. In his book
Behaviorism (1924) he made a famous, extremely bold
determinist claim:
" Give me a dozen healthy infants, well-formed, and my
own specified world to bring them up in and I’ll guaran-
tee to take any one at random and train him to become
any type of specialist I might select (1924, p. 104).
Despite Watson’s optimism, his behaviorism was
limited in its potential by the fact that the mechanism
of stimulus-substitution conditioning was dependent
upon there being a relevant natural stimulus–response
connection upon which to build. The other form of
behaviorism that emerged in the early decades of the
twentieth century was much more flexible. The pioneer
here was E.L. Thorndike, whose work was in essence
built upon by B.F. Skinner among others; rather than
focusing on the stimulus that led to a responding
behavior (later called an operant by Skinner), Thorn-
dike investigated how the consequences of a behavioral
episode influenced the likelihood of it being repeated.
(If the behavior led to a reinforcing outcome, the prob-
ability of it being repeated – in appropriate
circumstances – was increased.) Thorndike’s important
book Animal Intelligence (1898) was based upon his
doctoral dissertation, and he adopted associationist
language throughout. His research made use of the
so-called puzzle box – essentially a cage with an escape
mechanism – in which an animal was enclosed, but
from which it could see some “reward” such as food.
On first enclosure in the box, the animal would flail
about and eventually hit the escape mechanism by
chance (and thereby gain access to the reward); Thorn-
dike recorded the times taken to escape on subsequent
trials, finding that the animal learned to do so more
quickly – a result that was recorded graphically as
a “learning curve.” On the basis of this work Thorndike
formulated the “law of effect” and the “law of exercise”
(see below).
Thorndike favored the use of cats; later B.F. Skinner
favored the use of rats and pigeons. Among other things
he investigated various “schedules of reinforcement,”
and found (for example), that the target behavior or
operant need not be reinforced every time – indeed,
intermittent reinforcement led to this behavior
persisting for long periods in the absence of the delivery
of the reward. Famously, Skinner (1961) was even able
to teach a pigeon to dance, by rewarding an accidental
behavior (a movement in a desired direction) so that
it became habitual, and then rewarding a further acci-
dental movement in the next desired direction, and so
on – the whole process of building up movements that
constituted a dance taking a remarkably short time.
Skinner wrote a novel, Walden Two (1948), showing
how a human utopia could be built upon behaviorist
principles, and he was a formidable controversialist in
support of the use of scientific methods in psycholog-
ical research – the position he adopted on the issues
here having been influenced by the logical positivist
philosophers.
Among other significant work, more than a decade
earlier E.C. Tolman (1932), working on “purposive
behavior” with rats learning to run through a maze to
get to a reward of food, produced evidence suggesting
that they developed a “cognitive map” of the maze
which allowed them to navigate it when several of its
passages had been obstructed.
Work in the various behaviorist traditions
flourished during the decades of the 1930s and the
1940s, but a crisis was precipitated by the 1957 publi-
cation of a book by Skinner in which he attempted to

account for human verbal behavior in terms of his
behaviorist mechanisms. In 1959 Chomsky published,
in the journal Language, “A Review of B.F. Skinner’s
Verbal Behavior,” which argued at some length and in
technical depth that the mechanism of reinforcement
was unable to account for the complexities of human
language use, and that the key concepts of “stimulus,”
“response,” and “reinforcement” were being expanded
in loose ways – a critique that is widely considered to
have been one of the key turning points in behavior-
ism’s fortunes:
" Careful study of this book . . . reveals however that
these astonishing claims are far from justified . . . the
insights that have been achieved in the laboratories of
the reinforcement theorist, though quite genuine, can
be applied to complex human behavior only in the
most gross and superficial way, and that speculative
attempts to discuss linguistic behavior in these terms
alone omit from consideration factors of fundamental
importance. (Chomsky 1959/1980, p. 48)
For example, one of the striking features of
the rapid learning of their native language by young
children that Skinner’s framework could not explain
was their ability to understand many sentences in their
native language that had a grammatical form which
they had never before encountered (which ruled out
the possibility that their learning resulted from rein-
forcement of prior correct responses). In the decades
following Chomsky’s critique, the growth of cognitive
science has made it increasingly harder to overlook
or discount the accumulating evidence for the existence
of information-processing mechanisms internal to
behaving organisms.
Important Scientific Research and
Open Questions
The research programs of the behaviorists led to many
contributions to the understanding of the complexities
associated with learning; and in addition, of course,
kept the issue of the scientific nature of psychology on
the intellectual agenda of the field – where it remains
a locus of contention today. Watson established that
many phenomena (such as young children’s likes and
dislikes, fears, and behavioral problems) could have
been acquired as the result of classical stimulus-
substitution conditioning (although this mechanism
was not as ubiquitous as Watson supposed); Skinner’s
Behaviorism and Behaviorist Learning Theories B 441
work on schedules of reinforcement also was impor-
tant, as was his development of the “teaching machine”
and “programmed” textbooks wherein a learner’s cor- B
rect responses to questions about the items of knowl-
edge that had been presented were reinforced virtually
instantly by the learner being given feedback that the
responses were, indeed, correct; and the work of
Tolman and others also was significant. But no doubt
the most famous contributions were Thorndike’s laws
of effect and exercise, which he put forward on the basis
of his work with cats in his “puzzle boxes”; the first of
these was as follows:
" Any act which in a given situation produces satisfaction
becomes associated with that situation, so that when
the situation recurs the act is more likely than before to
recur also. Conversely, any act which in a given situa-
tion produces discomfort becomes disassociated from
that situation, so that when the situation recurs the act
is less likely than before to recur. (1905)
The law of exercise stated that the more often
a response is made in a particular situation, the more
strongly it becomes associated with it, and conversely,
prolonged disuse tends to weaken the association.
Today, behaviorism lives on as a source for the
development of therapies, and of behavior manage-
ment techniques for autistic children, chronic schizo-
phrenics, and others – and the principles of
reinforcement of positive behavior and of successful
learning are part of the armamentarium of most suc-
cessful teachers.
Cross-References
▶ Animal Learning and Intelligence
▶ Approaches to Learning
▶ Association Learning
▶ Associationism
▶ Conditioning
▶ Habit Learning in Animals
▶ Human Learning
▶ Law of Effect
▶ Reinforcement Learning
▶ Verbal Behavior and Learning
References
Chomsky, N. (1959/1980). A review of B. F. Skinner’s verbal behavior.
In N. Block (Ed.), Readings in philosophy of psychology (Vol. 1,
pp. 48–63). Cambridge: Harvard University Press.
442 B Belief Formation
Skinner, B. F. (1961). Cumulative record (Enlargedth ed.). New York:
Appleton.
Thorndike, E. L. (1898/1948). Animal intelligence. In W. Dennis
(Ed.), Readings in the history of psychology (pp. 377–387).
New York: Appleton.
Watson, J. B. (1913/1948). Psychology as the behaviorist views it.
In W. Dennis (Ed.), Readings in the history of psychology
(pp. 457–471). New York: Appleton.
Watson, J. B. (1924/1966). Behaviorism. Chicago: University of
Chicago Press.
Tolman, E. C. (1932). Purposive behavior in animals and men.
New York: Century.
Belief Formation
J. MICHAEL SPECTOR
Learning and Performance Support Laboratory,
University of Georgia, Athens, GA, USA
Synonyms
Cognitive state; Epistemological state; Knowledge
claim; Mental model
Definition
In psychology, a belief is typically defined as an internal
mental state (e.g., a mental model) or a disposition that
is often associated with a particular response that could
be in the form of a statement or an action (see, e.g.,
Fishbein and Ajzen 1975). This is a somewhat vague
definition as it refers to states or dispositions and state-
ments or actions. Prior to the rise of cognitive psychol-
ogy Neisser (1967) in the 1900s, a belief was more
simply associated with a statement that made
a factual claim of some sort. Our understanding of
human reasoning is no longer so simple. Complicating
the definition of ‘belief ’ are several related consider-
ations. First, a belief is a particular human characteris-
tic that can rarely be considered in isolation. Other
beliefs are typically relevant to a particular belief
under scrutiny (see, e.g., Quine and Ullian 1978). For
example, when considering an individual’s stated pref-
erence for a political candidate (a representative belief),
it might be relevant to also consider that individual’s
dispositions with regard to race, gender, taxes, and
many other things. Moreover, it is not merely other
beliefs that might be relevant to understanding
a person’s particular belief about something – that
person’s attitudes and habits might also be relevant
(Fishbein and Ajzen 1975). Some beliefs might be
inferred based on the actions and decisions of an indi-
vidual, even when that individual would say he or she
was unaware of having such a belief. Finally, beliefs can
and do change over time and with experience. A belief
can no longer be simply defined as a discrete cognitive
state that can be represented with a simple declarative
proposition. Rather, cognitive psychologists and natu-
ralistic epistemologists suggest that an individual’s
beliefs comprise a complex and dynamic system that
consists of a set of subsystems that are loosely related
but that allow an individual to form new [tentative]
beliefs and take action in a wide variety of circumstances.
Theoretical Background
Ludwig Wittgenstein (1922) made the remarkable
observation in the Tractatus Logico-Philosophicus that
we picture facts to ourselves (remark #2.1). These
internal pictures might be called mental models or
beliefs depending on one’s purpose or academic train-
ing. Wittgenstein would have been more inclined to
regard these internal pictures as beliefs from the point
of view of epistemology. In the Tractatus, Wittgenstein
was mainly concerned with beliefs that had simple,
observable correlates; that is to say that he was
concerned with beliefs that could be shown to be true
or false with relative ease. However, Wittgenstein was
well aware of the interrelatedness of cognitive and
noncognitive factors when he wrote the Tractatus and
said (see remark #6.43) that the world waxes and wanes
as a whole and that the world of the happy person is not
the same as that of the unhappy person. Quine and
Ullian (1978) likewise argue that it is an entire set of
beliefs that we bring to experience and not a simple
discrete belief. When our belief system encounters an
anomaly, we must make alterations to the relevant
prior beliefs or drop an entire set of beliefs. Typically,
we are not so inclined to drop entire sets of beliefs. This
human characteristic might be called the conservation
of beliefs. Festinger (1957) makes it the foundation of
cognitive dissonance theory.
Cognitive psychologists and naturalistic epistemol-
ogists have long been trying to work out the nature of
the relationships between cognitive and noncognitive
factors on belief formation. Many of these relationships
have yet to be elaborated and explored. In any case, part

of the foundation for belief formation and mental
model development comes from naturalistic episte-
mology. A second important foundation for belief for-
mation is based on cognitive psychology and research
on mental models. Seel (2003) provides a general foun-
dational framework that links the philosophy of science
and theory construction with both cognitive psychol-
ogy (mental model development) and epistemology
(the logic of external representations). Seel does
include some noncognitive factors (e.g., the influence
of culture) and he explicitly addresses the implications
for learning and instruction.
While there seems to be a great deal of convergence
in the fields of cognitive psychology, epistemology and
the philosophy of science with regard to the develop-
ment of mental models and belief formation, there is
also some discord within these related research com-
munities. One primary point of difference is the extent
of rationality as a principle of belief formation. Quine
and Ullian (1978) argue that humans naturally want to
maintain a set of consistent beliefs. When confronted
with evidence that a particular set of beliefs might be
mistaken, the typical reaction is to make changes to
those beliefs in order to preserve consistency. In any
case, the implication is that humans cannot tolerate
inconsistency in their beliefs. Likewise, Festinger
(1957) argues that cognitive dissonance arises when
a person is confronted with conflicting beliefs or with
evidence that conflicts with an existing belief. The
person will then naturally resolve the conflict by chang-
ing a belief or possibly by misinterpreting or ignoring
the evidence that runs counter to established beliefs. In
any case, Festinger also believes that people naturally
abhor inconsistency in their beliefs. However, if one
expands the notion of a belief to include inferred beliefs
based on the actions and dispositions of a person, one
can find people saying one thing and doing something
that appears to conflict with a stated belief. While
humorists might say that this is the defining character-
istic of a politician, the ability to say one thing with
apparent sincerity while doing something that appar-
ently contradicts that belief is an all too familiar and all
too human characteristic (Spector 2000). In summary,
the principle of rationality is more of a goal than
a determining factor in belief formation. We would
like our beliefs to form consistent and coherent net-
works, but we may sometimes fall short of that goal for
any number of reasons.
Belief Formation B 443
Important Scientific Research and
Open Questions
When considering sets of beliefs, there is much that is B
not known. The interrelated effects of attitudes, beliefs,
habits, mental models, moods, motivation, predispo-
sitions, and volition on the formation of new beliefs
and their persistence have not been fully explored.
Theoretical and conceptual frameworks that accom-
modate all of these factors have yet to be constructed
and validated, although some models now integrate
a few of these factors (Keller’s ARCS model, for exam-
ple). Implementing such frameworks in learning and
instruction has yet to occur, although there is a push
toward more adaptive and personalized learning. Fac-
tors affecting the persistence of beliefs are especially in
need of further exploration. While it is now common to
find cognitive psychologists claiming that both cogni-
tive and noncognitive factors are relevant, there is still
not much research on the influence and role of
noncognitive factors on learning and instruction. It is
well-established that repetitive training with regard to
recurrent and rather straightforward procedures can facil-
itate the establishment of persistent performance and
associated beliefs. However, when the situation involves
complex problems and decisions, it is not always so clear
what beliefs are relevant, how those beliefs are formed,
and how they might be made less subject to fading and
forgetting once the situation has passed.
Cross-References
▶ Belief-Based Learning Models
▶ Dogmatism
▶ Dogmatism and Learning
▶ Model Facilitated Learning
▶ Naturalistic Epistemology
▶ Philosophy of Learning
▶ Situated Learning
References
Festinger, L. (1957). A theory of cognitive dissonance. Stanford:
Stanford University Press.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behav-
ior: An introduction to theory and research. Reading: Addison-
Wesley.
Neisser, U. (1967). Cognitive psychology. New York: Appleton.
Quine, W. V. O., & Ullian, J. S. (1978). The web of belief (2nd ed.).
New York: McGraw-Hill.
Seel, N. M. (2003). Model-centered learning and instruction. Tech-
nology, Instruction, Cognition and Learning, 1, 59–85.
444 B Belief Learning
Spector, J. M. (2000). Towards a philosophy of instruction. Educa-
tional Technology & Society, 3(3), 522–525.
Wittgenstein, L. (1922). Tractatus logico-philosophicus. London:
Routledge & Kegan Paul. Translated by C. K. Ogden.
Belief Learning
▶ Belief-Based Learning Models
Belief System
▶ Dogmatism
▶ Dogmatism and Learning
Belief-Based Learning Models
NICK FELTOVICH
University of Aberdeen Business School,
Aberdeen, UK
Synonyms
Belief learning
Definition
The belief-based learning models are a class of models
that intend to describe the way agents (such as
humans) learn over time in strategic situations
(games) that they play repeatedly. According to
a belief-based learning model, agents form beliefs
about the expected behavior of others, and choose
actions based on how they would perform against
opponents playing in this expected way. After playing
the game, they update their beliefs based on the feed-
back they receive.
An important aspect of belief-based learning is that
agents understand they are playing a strategic game,
allowing them to have a mental model about the way
opponents play. This is in contrast to reinforcement-
based learning models, where agents choose and update
based only on the payoffs received in the game, so that
it is not necessary to have a mental model of oppo-
nents’ play (in fact, under reinforcement learning,
agents need not realize that they are playing a game at
all). Additional assumptions underlying belief-based
learning models are that agents receive enough infor-
mation to be able to carry out their belief updating (in
particular, they can observe opponents’ choices), and
they have the cognitive ability to compute the perfor-
mance of their actions, given their beliefs.
While belief-based learning places some require-
ments on agents’ abilities, as mentioned above, it
assumes that agents do not have other abilities (or at
least, do not use them). Agents are assumed to be
myopic, in that they do not anticipate how their cur-
rent behavior affects payoffs in future plays of the game.
In particular, they do not engage in “strategic teach-
ing”: deliberately choosing an action that gives a lower
current payoff, in the anticipation that it beneficially
changes opponent behavior – through opponents’
learning – in future plays of the game, thus raising the
agent’s long-run payoff. Agents are also assumed to be
unsophisticated, in that they do not use information
about opponents’ payoffs or opponents’ information in
order to forecast which strategies the opponent is likely
to choose (or not to choose).
Theoretical Background
Early research into belief-based learning models did not
motivate them as models of the way agents learn, but
rather as a technique for finding solutions (primarily
Nash equilibria) of games. George Brown (1949) pro-
posed a belief-based learning model called fictitious
play, which for the case of two-player simultaneous-
move games, is characterized by the following properties:
1. Agents’ initial beliefs are characterized by
a nonnegative belief weight for each pure strategy
available to the opponent.
2. After a play of the game, beliefs are updated: the
belief weight corresponding to the pure strategy
chosen by the opponent is augmented by one.
3. Before a play of the game, the agent assesses the
probability of the opponent choosing a particular
pure strategy as the belief weight corresponding to
that strategy, divided by the sum of belief weights
over all pure strategies (the strength of beliefs). These
probabilities then form the agent’s belief about the
opponent’s mixed strategy.
 Belief-Based Learning Models B 445

4. For each play of the game, the agent chooses the strategies are identical to differences in the potential
pure strategy with the highest expected payoff, function from the same changes in strategies. Another
given his beliefs. If more than one pure strategy class is that of games with strategic complementarities, B
maximizes his expected payoff, he chooses ran- where the more likely an opponent is to choose
domly among the tied strategies. a particular strategy, the more attractive that strategy
becomes to an agent.
Note that if the opponent is playing a stationary
By the 1980s, researchers began to think of belief-
strategy (one that does not change over time), then the
based learning models as descriptions of the way agents
agent’s beliefs will converge to the opponent’s strategy.
actually learn. As a result, fictitious play came under
Fictitious play, like most belief-based learning
some criticism as being unrealistic, due to its assump-
models, typically conforms, at least weakly, to well-
tion that agents are capable of choosing strategies opti-
known principles of learning from psychology. Over
mally, given their beliefs. An alternative, called cautious
time, as beliefs about others’ behavior become more
fictitious play (also known as stochastic fictitious play),
accurate, agents’ decisions satisfy the Law of Effect (bet-
weakens this assumption by replacing Property #4
ter strategies become more likely). Also over time, the
above with 4a. For each play of the game, the agent
strength of beliefs increases, so that beliefs change
chooses the pure strategy with a probability that is
increasingly slowly; that is, learning is consistent with
positively related to its expected payoff, given his
the Power Law of Practice (learning slows as more
beliefs.
experience is gained).
Thus, instead of agents choosing optimally based
Early study of fictitious play revolved around the
on their beliefs, they are merely more likely to choose
conjecture that it would find a Nash equilibrium (a
better responses than worse ones (“better response”
collection of strategies – one for each agent – such
instead of best response), given their beliefs. Unlike
that each agent’s strategy maximizes his payoff against
fictitious play, cautious fictitious play typically cannot
the other strategies), in the sense that the historical
be used to find Nash equilibria, as its fixed points are
distributions of agents’ choices converge to a Nash
quantal response equilibria (a game-theoretic solution
equilibrium. In the case of pure-strategy Nash equilib-
concept that assumes agents make errors, and take into
rium, strong results have been obtained. For example,
account others’ errors when making their own choices)
pure-strategy Nash equilibria are absorbing states
instead of Nash equilibria. However, by incorporating
(once one is played, it will be played forever) for ficti-
bounded rationality, it promises a more realistic
tious play, and if fictitious play leads to a pure-strategy
description of how actual people learn.
profile being chosen all but a finite number of times, it
Another way of generalizing fictitious play involves
is a Nash equilibrium.
the manner in which beliefs are updated. Under ficti-
The case of mixed-strategy Nash equilibrium is
tious play, all past opponent choices are weighted
more complicated. Julia Robinson (1951) proved that
equally in forming beliefs, irrespective of when the
for the special case of two-player zero-sum games, the
choices were made: a choice in the most recent play of
historical distribution of agents’ choices under ficti-
the game counts just as much as a choice from the very
tious play converges to a Nash equilibrium of the
first play. Some variations of fictitious play (or cautious
game – even if the only equilibrium is one in mixed
fictitious play) allow for recent experience to count for
strategies – and a similar result for all two-player games
more than earlier experience, via a “discounting”
with two strategies for each player (2  2 games) soon
parameter f, with f 2 [0, 1]. In this case, Property
followed. However, Lloyd Shapley (1964) showed, via
#2 above is replaced by 2a. After a play of the game,
a counterexample using a 3  3 game, that no such
beliefs are updated: all belief weights are multiplied by
result held for general two-player games. Thereafter,
f, then the belief weight corresponding to the pure
researchers have sought to identify classes of games
strategy chosen by the opponent is augmented by one.
for which fictitious-play learning does converge to
If f = 1, all past opponent choices count equally, as
Nash equilibria. One such class is that of potential
in fictitious play. If f = 0, the agent “forgets” all
games, for which a “potential function” exists, such
opponent choices except for the most recent choice;
that differences in an agent’s payoff from changing
combined with choosing a best response to beliefs
446 B Belief-Based Learning Models
(Property #4), this yields the Cournot adjustment
model. Drew Fudenberg and David K. Levine (1998)
proposed a general belief-based model of learning that
included the above variations on fictitious play – better
response rather than best response, and discounting of
past experience – along with an “inertia” parameter,
where with some fixed probability, an agent simply
makes the same choice as in the previous play.
Important Scientific Research and
Open Questions
In the late 1980s and 1990s, learning models gained
popularity among experimental economists as a way of
describing behavior by experimental participants that
seemed inconsistent with standard game-theoretic pre-
dictions. Some researchers used well-known belief-
based models for this purpose, such as fictitious play
or the Cournot model, while others modified these
models to increase sophistication: for example, by
adding an assumption that when forming beliefs,
agents attribute zero probability to their opponents’
choosing a strictly dominated strategy (a strategy that
always performs worse than some other strategy).
By the late 1990s, researchers had begun using
experiments as tests between learning models. Initially,
these tests concentrated on different parameterizations
of the same general model (for example, fictitious play
versus Cournot adjustment), but quickly, experimental
economists started testing between types of learning
model: most notably, between belief-based and rein-
forcement-based learning models. In some cases, exper-
iments were designed specifically in order to test the
assumptions of learning models; for example, different
patterns of behavior have been found in an experimental
game depending on whether participants had sufficient
information to be able to execute belief-based learning
(that is, whether they were told only their own choices
and payoffs, or additionally told about opponents’ deci-
sions as well as the game’s payoff matrix). Such results
suggest that agents learn differently when this informa-
tion is available, though it need not be evidence in favor
of any of the common belief-based learning models.
Other researchers used data from one or more sets of
experiments in order to determine which learning model
best characterized the way experimental participants
learn. On the whole, this work failed to find overwhelm-
ing support for any one particular learning model;
rather, which model fits the data best has tended to
depend not only on the game being considered, but
also on what criteria of goodness of fit are used. Addi-
tionally, much of this work has subsequently been crit-
icized on the basis that when the experimental data
comprise multiple participants, and there is heterogene-
ity across these participants in their learning, then tests
of reinforcement-based versus belief-based models will
be biased in favor of the former, suggesting that belief-
based learning describes behavior better than most
studies have concluded.
More recent research has concentrated less on
differences between belief-based models and other
learning models than on their similarities. The most
influential such work is that of Colin Camerer and
Teck-Hua Ho (1999), who showed that versions of
belief-based and reinforcement-based learning could
be nested within a more general model which they
called experience weighted attraction (EWA). In their
model, an agent’s probability of choosing a particular
pure strategy is an increasing function of that strategy’s
“attraction.” In a simple reinforcement-based model,
attractions are nonnegative numbers which are
updated after each play of the game based on the payoff
that was earned in that play; the new attraction for the
strategy that was chosen is a weighted sum of the
previous attraction and the payoff, while the attractions
for the other strategies are unchanged. By contrast,
EWA allows for the possibility that unchosen strategies’
attractions are also updated, by the use of a “simulated
effect” parameter d, with d 2 [0,1]. The attraction for
the chosen strategy is updated in the same way as under
reinforcement, but now, the attractions for unchosen
strategies are updated by d times the payoff they would
have yielded if they had been chosen. If d = 0, EWA
reduces to a reinforcement-based learning model. Less
obviously, if d = 1, EWA reduces to a beliefs-based model
(including, for certain values of other parameters, ficti-
tious play or Cournot play). This was the key insight of
Camerer and Ho: having beliefs and choosing strategies
based on these beliefs is equivalent to a reinforcement-
based model in which all strategies’ attractions are aug-
mented by the payoff they would have received,
irrespective of which strategy was actually chosen.
Other similarities between belief-based and
reinforcement-based learning models involve their
dynamic properties. Ed Hopkins (2002) established
a connection between the dynamics of both cautious
fictitious play and reinforcement-based learning and the
 Beliefs About Language Learning B 447

replicator dynamics, a dynamic model used in evolution-

ary biology and evolutionary game theory. He also Beliefs
showed that under certain conditions, cautious fictitious ▶ Attitude Change Through Learning B
play and reinforcement-based learning have the ▶ Discourse and the Production of Knowledge
same asymptotic properties, though the former can
have an arbitrarily faster speed of convergence to the
asymptote.
One open area of research involving belief-based
learning models is simply the extension of work men- Beliefs About Knowledge
tioned above: when do belief-based models character-
ize learning well, and when do alternative models ▶ Epistemological Development and Learning
perform better? A second area, which also involves
generalizations of belief-based models such as EWA,
involves the reliability of these models across games.
If in one game, a particular model (including specific
Beliefs About Language
parameters) characterizes someone’s decisions best,
how well will this model predict the same person’s
Learning
decisions in a different game? More broadly, is there
EVA BERNAT
a connection between games and learning model
School of Education, Goodsell 109, The University of
parameters, such that given a new game, a researcher
New South Wales, Sydney, New South Wales, Australia
could predict which version of the learning model
would be expected to perform well? A third area con-
nects learning with the growing field of neuro-econom-
ics: does the brain operate in a different way when
Synonyms
Assumptions; Epistemologies; Metacognitive knowl-
agents are given sufficient feedback to execute belief-
edge; Perceptions; Suppositions
based learning than when they are not?

Cross-References
▶ Adaptive Game-Based Learning
▶ Belief Formation
▶ Learning and Evolutionary Game Theory
▶ Naturalistic Epistemology
▶ Reinforcement Learning
References
Brown, G.W. (1949). Some notes on computation of games solutions.
RAND Corporation research memorandum P-78.
Camerer, C., & Ho, T.-H. (1999). Experience-weighted attraction
learning in normal form games. Econometrica, 67(4), 827–874.
Fudenberg, D., & Levine, D. K. (1998). The theory of learning in
games (economics learning and social evolution). Cambridge:
MIT Press.
Hopkins, E. (2002). Two competing models of how people learn in
games. Econometrica, 70(6), 2141–2166.
Robinson, J. (1951). An iterative method of solving a game. The
Annals of Mathematics, 4(2), 296–301. Second Series.
Shapley, L. S. (1964). Some topics in two-person games. In M. Dresher,
L. S. Shapley, & A. W. Tucker (Eds.), Advances in game theory
(pp. 1–28). Princeton: Princeton University Press.
Definition
Defining a latent construct such as beliefs is always
a challenge. The plethora of different definitions avail-
able highlight its complex and multifaceted nature.
Scholars provide definitions that range from those
reflecting theoretical approaches in cognitive psychol-
ogy to those in line with socioconstructivist views. As
a result, learner beliefs have been defined, inter alia, as
cognitive constructs, implicit theories, personal episte-
mologies, folkloristic conceptions of learning, and
“general assumptions that students hold about them-
selves as learners, about factors influencing language
learning, and about the nature of language learning and
teaching” (Victori and Lockhart 1995, p. 224).
Theoretical Background
In recent decades, learner beliefs have been situated
predominantly within the study of epistemology.
Epistemology deals with affairs of the intellect, and
the deployment of intellect involves either mental acts
or states, or linguistic acts (public utterances),
448 B Beliefs About Language Learning
frequently both. Among mental states, beliefs are usu-
ally singled out by epistemologists, while among lin-
guistic acts, assertions are of central concern. Most
intellectual endeavors try to arrive at some belief on
a designated topic resulting in assertions, so epistemol-
ogy naturally focuses on either beliefs or assertions.
Beliefs such as those about the nature of knowledge,
beliefs about learning (including self-efficacy beliefs
regarding own abilities in learning), and beliefs about
language learning in particular, all consitute one’s belief
system, and have been found to form the basic blocks of
epistemology (Goldman 1986).
While traditionally beliefs have been linked with
information about an object, attitudes to the evalua-
tion of that object conceptualized beliefs as more than
just cognitions, capable of arousing affect around the
object of the belief and thus having a motivational
function. Clusters of beliefs form attitudes that are
functionally and cognitively connected to the value
system. Value, then, refers to a single proscriptive or
prescriptive belief that transcends specific objects or
situations, while attitude refers to an organization of
several beliefs focused on a specific object or situation –
for example, a foreign language. According to this view,
values are more central concepts than attitudes, are
determinants of attitudes, and are more resistant to
change. Together they constitute the value-attitude sys-
tem, embedded in the wider belief system (Fishbein
and Ajzen 1975).
The recognition of the role of learners’ epistemo-
logical beliefs across various disciplines, including lan-
guage learning, contributed to a growing body of
evidence suggesting that they play a central role in
learning experience and achievements, have
a profound influence on learning behavior as well as
learning outcomes, and act as very strong filters of
reality. Interdisciplinary research shows how one’s
belief systems, social cognitions, and metacognitions
are a driving force in intellectual performance, and that
learners may be directly influenced by their perception
of success in learning and levels of expectancy, with
realistically high expectations helping to build confi-
dence, and low (or unrealistically high) expectations
leading to de-motivation and disappointment.
In the language classroom context, the epistemo-
logical beliefs that students bring with them to the
learning situation have been recognized as
a significant contributory factor in the learning process
and ultimate success. For example, second or foreign
language students may hold strong beliefs about the
nature of the language under study, its difficulty, the
process of its acquisition, the success of certain learning
strategies and teaching methods, the existence of lan-
guage aptitude, and their own expectations about
achievement. Preconceived beliefs may directly influ-
ence or even determine a learner’s attitude or motiva-
tion, and precondition the learner’s success or lack of
success. Supportive and positive beliefs help to over-
come problems and thus sustain motivation, while
negative or unrealistic beliefs can lead to decreased
motivation, frustration, and anxiety. While many suc-
cessful learners develop insightful beliefs about lan-
guage-learning processes that have a facilitative effect
on learning, others can have “mistaken,” uninformed,
or negative beliefs that may lead to a reliance on less
effective strategies, resulting in a negative attitude
toward learning and autonomy, even classroom anxi-
ety. Students who believe, for example, that learning
a language primarily involves learning new vocabulary
will spend most of their energy on vocabulary acquisi-
tion, while older learners who believe in the superiority
of younger learners probably begin language learning
with fairly negative expectations of their own ultimate
success. In addition, an unsuccessful learning experi-
ence may likely lead students to the conclusion that
special abilities are required to learn a foreign language
and that they do not possess these necessary abilities.
Such beliefs can also inhibit learners’ perceptiveness to
the ideas and activities presented in the language class-
room, particularly when the approach is not consonant
with the learners’ experience.
Identification of these beliefs and reflection on their
potential impact on language learning and teaching in
general, as well as in more specific areas such as
learners’ expectations and strategies used, can inform
syllabus design and teacher practice. Indeed, pedagogy
has the capacity to provide the opportunities and
conditions within which learner contributions are
found to have a positive effect upon learning and may
be more fully engaged. Discovering students’ attitudes
and beliefs is possible, as it is generally accepted that
language learners are capable of bringing this knowl-
edge to consciousness and articulating it. So far,
research studies have employed predominantly quanti-
tative research methods to investigate learner beliefs via
self-report surveys, although qualitative methods

involving various forms of narrative enquiry (e.g., see
discourse analysis) have become more prevelant in
recent years (Kalaja and Barcelos 2003), and have
shed new light on the nature, depth, and strength of
learner beliefs – aspects of a latent construct which can
be difficult to capture via surveys alone.
Important Scientific Research and
Open Questions
The challenge for the field of research on language
learner beliefs is threefold. The first is to continue to
embrace various multidisciplinary approaches to
empirical studies in the field, taking advantage of devel-
opments in cognitive and social research fields, among
others. Such integration could provide a more holistic
framework, while enabling the triangulation of data
obtained via different methodological means in empir-
ical studies. While the current diversity of theoretical
perspectives has steadily begun to create a rich tapestry
of complimenting studies, there is a need for more
plurality in this area given the scope and depth of
one’s beliefs (Bernat 2008). Combinations of
approaches and methods utilizing both qualitative
with quantitative data could provide an even greater
insight into a multitude of potentially interacting socio-
cultural/contextual, affective/psychological, cognitive/
neurobiological, and ideological factors that have been
found to determine – to a variable extent – learner beliefs
about language learning.
The second challenge is to investigate the impact of
learner beliefs on their language-learning processes as
well as outcomes. While much has been reported on the
nature and strength of learner beliefs, less has been said
about their actual impact in the classroom and beyond.
A question of great interest for future research will be to
determine at which point or in which contexts beliefs
start to be used as resources for self-regulation in lan-
guage learning. This could perhaps be done via elicita-
tion techniques such as stimulated recall, particularly
when cognitive dissonance (dissonance between beliefs
and actions) might have occurred.
The mismatch between teacher and learner beliefs
presents a third challenge for the field. Future research
needs to investigate ways of dealing with this mismatch
in the classroom, as the majority of literature reports
significant belief differences between teachers and their
learners. For example, students tend to hold a higher
preference for learning grammar, translation,
Beliefs About Language Learning B 449
vocabulary and pronunciation exercises, and often
hold unrealistic expectations about the length of time
it takes to learn a foreign language. Significant differ- B
ences between student and teacher beliefs might create
tension in the classroom, thus emphasizing the need for
studies to explore the most productive ways of mini-
mizing this gap. There is a need to investigate possible
intervention measures and their effectiveness in chang-
ing or restructuring learner beliefs in the classroom
context, should they be found to be unproductive,
unrealistic, or hampering the language acquisition pro-
cess. Beliefs have been found to be strongly held and
resistant to change, and there is evidence of
a progressive construction and crystallization of beliefs
about learning, and such beliefs are argued to influence
increasingly the more situationally specific learning
behavior (Cantwell 1998), where this situation speci-
ficity could easily reflect a language-learning context.
Whether learner beliefs can (and even ought to) be
changed in the classroom context is a currently
much under-researched area, reflected by the paucity
of literature on the issue. Most often studies do not
examine long-term changes in beliefs by readministering
dependent measures over time; yet, a number of theories
exist which could provide a theoretical framework for
future studies exploring the possibilities of belief change
and change continuity in the language-learning context.
However, some scholars caution that a better under-
standing and a stronger foundation of learner beliefs is
needed, before any intervention measures are applied.
Whether the field has already gained enough knowledge
in this area in order to successfully intervene with theo-
retically sound intervention methods is still, in the view
of some, a matter for debate and further research.
Cross-References
▶ Belief Formation
▶ Beliefs About Learning
▶ Metacognition and Learning
References
Bernat, E. (2008). Beyond beliefs: Psycho-cognitive, sociocultural and
emergent ecological approaches to learner perceptions in foreign
language acquisition. Asian EFL Journal, 10(3), Article 1. Available
from http://www.asian-efl-journal.com/September_08_eb.php
Cantwell, R. H. (1998). The development of beliefs about learning
from mid-to-late adolescence. Educational Psychology, 18(1),
471–495.
450 B Beliefs About Learning
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behaviour:
An introduction to theory and research. Reading: Addison-Wesley.
Goldman, A. I. (1986). Epistemology and cognition. Cambridge: Har-
vard University Press.
Kalaja, P., & Barcelos, A. M. F. (Eds.). (2003). Beliefs about SLA:
New research approaches. Dordrecht: Kluwer.
Victori, M., & Lockhart, W. (1995). Enhancing metacognition in
self-directed language learning. System, 23(2), 223–234.
Beliefs About Learning
CHANMIN KIM
The Department of Educational Psychology &
Instructional Technology, The University of Georgia,
Athens, GA, USA
Synonyms
Epistemic beliefs; Epistemological beliefs
Definition
Beliefs about learning refer to a person’s subjective
judgments about a relation between learning and his
or her values or attributes (Fishbein and Ajzen 1975).
For example, a person’s beliefs about learning calculus
involve subjective judgments about the nature and
importance of calculus, that person’s interest in calcu-
lus, and that person’s beliefs about his or her ability to
learn calculus.
Theoretical Background
Students’ beliefs about learning and knowledge have
been researched to explain students’ responses to learn-
ing contexts (Hofer and Pintrich 1997). Some research
has focused on how beliefs develop over time while
other research has highlighted how beliefs influence
the cognitive processes involved in learning. Research
has also investigated how beliefs mediate the factors of
attitudes toward learning (i.e., interest, willingness,
etc.) and study habits (as exemplified by effort, persis-
tence, etc.), which are considered to indirectly impact
learning and achievement (e.g., Buehl et al. 2002; Hofer
2006; Pintrich and Schunk 2002). The research with
regard to the relation of beliefs to attitudes and study
habits is relatively newer and there are fewer studies than
one finds in research regarding the development of
beliefs and the relation of cognitive processes to beliefs.
In addition, research has been conducted on how to
change students’ beliefs about knowledge and learning
to more availing beliefs that presumably facilitate
learning (e.g., Kim and Keller 2010). “Availing” is the
term first used in Muis (2004) to characterize some
beliefs as implying more possibilities to enhance stu-
dents’ learning. For example, the belief that ability to
learn is not innate or fixed but it is developed over time
is an availing belief, which would influence a student’s
decision making regarding effort in a challenging task.
This term has the same meaning as “sophisticated” and
“mature” beliefs in Schommer (1994) and “appropri-
ate” beliefs in Schoenfeld (1988), as opposed to “naı̈ve
and inappropriate” beliefs.
In short, researchers who consider students’ beliefs
about knowledge and learning directly or indirectly
related to learning outcomes through influencing cog-
nitive processes, motivation, attitudes, behavior, and
efforts have studied the development of the beliefs and
possible ways to improve the beliefs in a variety of
learning contexts. Meanwhile, most research on stu-
dents’ beliefs about knowledge and learning has been in
light of epistemological beliefs. One of the most influ-
ential scholars in the field of epistemological belief
research, Schommer, has conceptualized and refined
a framework of epistemological belief research (e.g.,
see Schommer 1998). Schommer (1998) has argued
that tacit beliefs about knowledge and learning play
a critical role in guiding how learners think and solve
problems. She proposed that a person’s epistemological
beliefs consist of multidimensional beliefs, which are to
some degree independent of each other, and she vali-
dated this notion with students in various learning
contexts and at different grade levels. Her five dimen-
sions of students’ beliefs about knowledge and learning
consist of: (1) the structure of knowledge, ranging from
isolated pieces to integrated concepts; (2) the source of
knowledge, ranging from authority to reasoning;
(3) the stability of knowledge, ranging from certain
knowledge to changing knowledge; (4) the speed of
learning, ranging from quick learning to gradual learn-
ing; and (5) the ability to learn, ranging from fixed at
birth to improvable (e.g., see Schommer 1998).
Her five dimensions of epistemological beliefs along
with her measurement tool, the Epistemological Belief
Questionnaire (EBQ), provide researchers with
a foundation to study students’ beliefs about knowl-
edge and learning. Nonetheless, the last two beliefs (i.e.,

beliefs about speed of learning and ability to learn)
seem to be directly related to learning rather than to
knowledge, compared to the other three representing
beliefs about knowledge itself. Considering that episte-
mological beliefs refer to beliefs about knowledge, the
inclusion of beliefs about ability to learn and speed of
learning in the category of epistemological beliefs has
been debated as these two beliefs are focused more
specifically on learning rather than knowledge (Hofer
and Pintrich 1997).
In addition to research on the multidimensionality
of epistemological beliefs, there has been research on
the domain generality specificity of beliefs (see Buehl
et al. 2002; Hofer 2006; Muis 2004). Most of the early
research on students’ epistemological beliefs were based
on the assumption that the same beliefs would be
observed across disparate domains. However,
researchers have begun to question the domain general-
ity of epistemological beliefs, proposing that students’
epistemological beliefs need to be separately investigated
within specific subject domains. In fact, several
researchers have found that a person’s epistemological
beliefs are different across domains. For example, Buehl
et al. (2002) found that there were prominent differ-
ences in students’ epistemological beliefs about knowl-
edge and learning between history and mathematics. In
addition, Hofer (2006) argued that epistemologies
were not the same across domains owing to the unique
characteristics of academic disciplines, although she
criticized several studies supporting domain specificity
on methodological grounds.
Important Scientific Research and
Open Questions
Muis (2004) indicated that beliefs about learning
develop and change over time in light of Schommer’s
(1994) argument of recursive development of episte-
mological beliefs rather than sequential development.
Support for this argument can be found in the research
literature. For example, Kloosterman and Cougan
(1994) interviewed K-6 elementary students about
their beliefs about mathematics, including a question
about whether they believed that everybody could learn
mathematics. The authors found that most students
showed availing beliefs about their ability to study
mathematics, which was inferred from the success of
a 2-year project to improve teaching mathematics.
Franke and Carey (1997) also found that 1st graders
Beliefs About Learning B 451
had availing beliefs about mathematics knowledge
acquisition in the classrooms where problem solving
was highlighted. As they regarded this finding as B
“reform-minded” (p. 8), they indicated that these
beliefs were developed as a result of the uniqueness of
the classroom environment.
Beliefs about learning have been recognized as an
important factor for students’ learning and achievement
by influencing both their attitudes toward learning,
motivation, and persistence. A further inquiry would
be then on what kinds of efforts should be made to
improve learning environments to positively influence
the development of students’ beliefs about learning.
Cross-References
▶ Belief Formation
▶ Epistemic Beliefs About Learning
▶ Epistemology
▶ Epistemological Beliefs and Learning
▶ Motivation and Learning: Modern Theories
References
Buehl, M. M., Alexander, P. A., & Murphy, P. K. (2002). Beliefs
about schooled knowledge: Domain general or domain specific?
Contemporary Educational Psychology, 27(3), 415–449.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, and behavior: An
introduction to theory and research. Reading, MA: Addison-Wesley.
Franke, M. L., & Carey, D. A. (1997). Young children’s perceptions of
mathematics in problem-solving environments. Journal for
Research in Mathematics Education, 28, 8–25.
Hofer, B. K. (2006). Beliefs about knowledge and knowing: Integrat-
ing domain specificity and domain generality: A response to
Muis, Bendixen, and Haerle 2006. Educational Psychology Review,
18(1), 67–76.
Hofer, B., & Pintrich, P. (1997). The development of epistemological
theories: Beliefs about knowledge and knowing and their relation
to learning. Review of Educational Research, 67, 88–140.
Kim, C., & Keller, J. M. (2010). Motivation, volition, and belief
change strategies to improve mathematics learning. Journal of
Computer Assisted Learning, 26(5), 407–420.
Kloosterman, P., & Cougan, M. C. (1994). Students’ beliefs about
learning school mathematics. Elementary School Journal, 94,
375–388.
Muis, K. R. (2004). Personal epistemology and mathematics:
A critical review and synthesis of research. Review of Educational
Research, 74(3), 317–377.
Pintrich, P. R., & Schunk, D. H. (2002). Motivation in education:
Theory, research, and applications (2nd ed.). Upper Saddle
River, NJ: Merrill Prentice Hall.
Schoenfeld, A. H. (1988). When good teaching leads to bad results:
The disasters of, “well-taught” mathematics classes. Educational
Psychologist, 23(2), 145–166.
452 B Beliefs and Values
Schommer, M. (1994). Synthesizing epistemological belief research:
Tentative understandings and provocative confusions. Educa-
tional Psychology Review, 6(4), 293–319.
Schommer, M. (1998). The role of adults’ beliefs about knowledge in
school, work, and everyday life. In M. C. Smith & T. Pourchot
(Eds.), Adult learning and development: Perspectives from educa-
tional psychology (pp. 127–143). Mahwah, NJ: Erlbaum.
Beliefs and Values
▶ Culture in Second Language Learning
Belonging
▶ Identity and Learning
Benefit Finding
▶ Posttraumatic Growth
Best Practices
In education: Strategies, activities, and/or approaches
that have been shown through research and evaluation
to be effective in teaching and learning.
Best-Fit Techniques
In the context of Myers-Briggs Type Indicator (MBTI)
“best-fit” techniques refer to the matching of person-
ality type data produced by the 93 point questionnaire
with the respondents’ own knowledge and awareness of
their type. Data for the latter is collected by face to face
discussion with qualified MBTI personnel.
Beta Power
▶ Learning-Related Changes of b-Activity in Motor Areas
Big Five Factor Model of
Personality
The five-factor model of personality is a currently pop-
ular model of personality. It includes the factors extra-
version, agreeableness, conscientiousness, neuroticism,
and openness to experience.
Big Five Personality and
Prejudice
BO EKEHAMMAR1, NAZAR AKRAMI2
1
Department of Psychology, Stockholm University,
Stockholm, Sweden
2
Department of Psychology, Uppsala University,
Uppsala, Sweden
Synonyms
Big five personality: Five-factor personality theory;
Discrimination; Personality trait theory; Prejudice:
Intolerance
Definitions
The Big Five (five-factor) personality model (McCrae
and Costa 2008) is probably the current most widely
accepted model of personality structure, encompassing
neuroticism (emotional instability), extraversion, open-
ness to experience, conscientiousness, and agreeableness.
These factors reflect core personality because of their
substantial heritability coefficients and their early
expression in temperament in human infants and in
other animal species.
The term prejudice refers to prejudgment and
originates from the Latin praejudicium. An early defi-
nition was proposed by Allport who suggested that
prejudice is “an antipathy based upon a faulty and
inflexible generalization” (1954, p. 9). In line with
Allport, more recent definitions suggest that prejudice
is composed of negative judgments, beliefs, and feelings
about people because of their social group member-
ship. Distinguished from prejudice, stereotypes refer to
the cognitive component of prejudice and comprise
beliefs about the characteristics, attributes, and behav-
iors of members of a particular social group.

Discrimination, however, means that people act or
behave in accord with their prejudiced beliefs and ste-
reotypes toward out-group members.
Theoretical Background
Why are some people more prejudiced than others?
The answer to this question has been a point of dis-
agreement in psychological research. One camp
advocates the view that prejudice is best explained by
personality psychology whereas the other maintains
that the answer can be found in social psychology.
The personality explanation holds factors within the
individual as the major causes of prejudice (e.g.,
Adorno et al. 1950; Allport 1954; Ekehammar and
Akrami 2003). The social psychology approach, on
the other hand, links prejudice to factors in the outside
world. Thus, people’s group membership, self-catego-
rization, social identity, social position, or contextual
factors like social norms or social threat, are seen as the
major determinants of expressing prejudice (e.g.,
Guimond et al. 2003). As the title of this text
indicates, the present chapter puts forth evidence that
personality really matters for explaining prejudice.
Theoretically, there are two major supporting lines
of argument for the personality–prejudice relationship.
First, the personality approach to the study of prejudice
was developed around the notion that prejudice can
be generalized from one target to another. Thus, if
a person is anti-homosexuals, he or she is likely to be
anti-immigrants, antiblacks, and so forth (e.g., Allport
1954). Interestingly, such generalized prejudice appears
even in the absence of real out-groups. Thus, classical
works examined attitudes toward various groups and
found high correlations between fictitious groups
(e.g., Pireneans) and real social groups (e.g., Jews,
Communists). This would suggest that there is a
within-individual component of prejudice that is inde-
pendent of target and social context.
Second, the Big Five personality factors are indica-
tors of dispositions that provide the individual with
domain-specific tendencies to behave in certain ways in
the social world (McCrae and Sutin 2009). Although all
five factors might in some way be related to interpersonal
relationships, the Agreeableness and Openness to
Experience (henceforth Openness) factors can be consid-
ered strong candidates for guiding our behavior toward
people around us. Agreeableness comprises components
like tender-mindedness and nonhostility as well as
Big Five Personality and Prejudice B 453
sympathy and concern for others. Correspondingly,
Openness comprises components that have to do with
nonconformity and unconventionality and is positively B
related to liberal sociopolitical values. The characteristics
of these factors would imply a negative relation with
prejudice. According to the model of McCrae and Costa
(2008), these factors reflect “basic tendencies,” posi-
tioned first in a causal chain. Also, McCrae and Costa
denote observable behavior, such as prejudice
displayed in discrimination, as “objective biography,”
positioned at the end of this chain. Whereas Allport
(1954, p. 73) argued that prejudice is a personality
factor by itself, our suggestion (e.g., Ekehammar and
Akrami 2003) is that prejudice is an outcome of core
personality like the Big Five factors and subfactors.
Important Scientific Research and
Open Questions
Indeed, there is empirical support for a concept of
generalized prejudice as reflected in relatively strong
correlations between prejudices toward various targets.
Thus, in line with the classical works recent research has
disclosed that attitudes toward homosexuals, Blacks,
women, and elderly people are correlated and that racism,
sexism, and antigay attitudes are highly correlated as well.
Recently, Zick et al. (2008) reported high correlations
among nine prejudice targets which they refer to as
a syndrome of group-focused enmity. Taken together,
the high correlations obtained between different prej-
udice targets in past and more recent research lend
support to the idea of prejudice as a unit of personality.
Recent studies employing factor analyses on four
potential prejudice targets (immigrants, women, gays
and lesbians, and disabled people) arrived at a single
generalized prejudice factor explaining 50–60% of the
variance (e.g., Ekehammar and Akrami 2003).
Although some researchers suggest that choosing
other targets can give multifactor solutions (Duckitt
and Sibley 2007), the replicated one-factor solution
seems to hold for traditional prejudice targets.
Ekehammar and Akrami (2003) were probably the
first to examine the personality–prejudice issue by ana-
lyzing the relation of the Big Five personality factors
with generalized prejudice using a variable-centered as
well as a person-centered approach. They showed that
two of the Big Five factors (Openness and Agreeable-
ness) displayed rather strong (r  0.45) negative
zero-order correlations with generalized prejudice.
454 B Big Five Personality: Five-Factor Personality Theory
When using all Big Five factors for predicting general-
ized prejudice in a linear multiple regression analysis,
Ekehammar and Akrami arrived at a fairly high pre-
dictive power (R = 0.56) with Agreeableness and Open-
ness showing the largest regression weights. Quite a few
studies have since then disclosed similar results (see
meta-analysis of Sibley and Duckitt 2008).
Because most trait models of personality are
hierarchical, there is always a question which level to
choose for the study of personality and its relations to
other variables. Some researchers have made a strong
case for suggesting the trait rather than the factor level
for predicting, and understanding, various external
behaviors and constructs. Using this idea, Ekehammar
and Akrami (2007) were probably the first to use Big
Five facets (subfactors) in personality–prejudice
research. Indeed, they found the Big Five facet scores
to have significantly higher power when predicting
prejudice than the Big Five factor scores. Also, the
scores on the facets Tender-Mindedness (underlying
the Agreeableness factor) and Values (underlying the
Openness factor) were the most powerful single pre-
dictors of prejudice, outperforming corresponding fac-
tor scores as single predictors.
Finally, it is important to note that the theoretical
background and research findings presented here do not
imply that prejudice can be explained by personality
alone. Prejudice is a multifaceted phenomenon that
needs to be explained by multiple approaches. What
seems to be clear in more recent research, however, is
that personality and social psychological variables explain
distinct aspects of prejudice (Akrami et al. 2011).
Cross-References
▶ Learner Characteristics
▶ Personality and Learning
▶ Personality Effects on Learning
References
Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N.
(1950). The authoritarian personality. New York: Harper & Row.
Akrami, N., Ekehammar, B., & Bergh, R. (2011). Generalized preju-
dice: Common and specific components. Psychological Science,
22, 57–59.
Allport, G. W. (1954). The nature of prejudice. Reading: Addison-
Wesley.
Duckitt, J., & Sibley, C. G. (2007). Right-wing authoritarianism,
social dominance orientation and the dimensions of generalized
prejudice. European Journal of Personality, 21, 113–130.
Ekehammar, B., & Akrami, N. (2003). The relation between person-
ality and prejudice: A variable- and a person-centred approach.
European Journal of Personality, 17, 449–464.
Ekehammar, B., & Akrami, N. (2007). Personality and prejudice:
From Big Five personality factors to facets. Journal of Personality,
75, 899–926.
Guimond, S., Dambrun, M., Michinov, N., & Duarte, S. (2003). Does
social dominance generate prejudice? Integrating individual and
contextual determinants of intergroup cognitions. Journal of
Personality and Social Psychology, 84, 697–721.
McCrae, R. R., & Costa, P. T., Jr. (2008). The five-factor theory of
personality. In O. P. John, R. W. Robins, & L. A. Pervin (Eds.),
Handbook of personality: Theory and research (3rd ed.,
pp. 159–180). New York: Guilford.
McCrae, R. R., & Sutin, A. R. (2009). Openness to experience.
In M. R. Leary & R. H. Hoyle (Eds.), Handbook of individual
differences in social behavior (pp. 257–273). New York: Guilford.
Sibley, C. G., & Duckitt, J. (2008). Personality and prejudice: A meta-
analysis and theoretical review. Personality and Social Psychology
Review, 12, 248–279.
Zick, A., Wolf, C., Küpper, B., Davidov, E., Schmidt, P., & Heitmeyer,
W. (2008). The syndrome of group-focused enmity: The inter-
relation of prejudice tested with multiple cross-sectional and
panel data. Journal of Social Issues, 64, 363–383.
Big Five Personality: Five-
Factor Personality Theory
▶ Big Five Personality and Prejudice
Bilingualism and Learning
HENRIK SAALBACH1, SEBASTIAN KEMPERT2
1
Institute for Behavioral Sciences, Research on
Learning and Instruction, ETH Zurich, Zurich,
Switzerland
2
Department of Educational Sciences, Goethe-
University Frankfurt, Frankfurt am Main, Germany
Definition
Bilinguals are individuals or groups of individuals who
have gained communicative skills, with various degrees
of proficiency, in oral and/or written forms, in order to
interact with speakers of one or more languages.

Theoretical Background
Bilingualism and learning offer a vast range of topics.
In the present article, we focus on the cognitive side of
the issue. The relation between bilingualism and learn-
ing may also be mediated by sociocultural factors or
variables related to motivation, attitude, and identity.
Research on learning is typically based on the sim-
plifying assumptions that learners are taught in one
language in which they are proficient and have one
conceptual system to store knowledge. Recent develop-
mental and cognitive research has shown that there is
an interaction between language and thought in most
cognitive domains, particularly in learning and devel-
opment (for an overview, see Gentner and Goldin-
Meadow 2003; see below). Taking into account this
finding for examining the relation between bilingual-
ism and learning, we need to ask whether and to what
extent learning processes are affected when students
speak more than one language, when learners have to
frequently switch between different languages, when
their proficiency in the language of instruction is low,
and when the language of instruction is different from
the language of retrieval. Do these scenarios support or
impede the learning process? Potential answers to these
questions, as provided by research, emphasize the
importance of language for examining the basic pro-
cesses that underlie school-related learning.
The relation between bilingualism and learning can
be viewed in a number of quite different perspectives
focusing on different aspects, e.g., emphasizing either
cognitive costs or cognitive benefits. On the one hand,
there are reports highlighting the serious school prob-
lems encountered by students from immigrant fami-
lies. On the other hand, bilingual teaching programs
become more and more popular, presumably because
these programs have generally been considered to be
a very efficient way of teaching since students improve
their foreign language skills as a side effect, without any
additional instructional effort. Furthermore, profi-
ciency in two languages has been associated with ben-
efits in terms of thinking about language as well as in
terms of basic cognitive functions (e.g., Bialystok et al.
2009). An approach that accounts for different perspec-
tives on the outcomes of bilingual learning has been
suggested by Cummins (2004). In his Threshold
Hypothesis, he claims that the quality of the cognitive
consequences of bilingualism depends on the degree of
proficiency in the two languages. Individuals who are
Bilingualism and Learning B 455
highly proficient in L1 and L2 (so-called balanced bilin-
guals) are likely to experience positive cognitive effects,
as compared to monolinguals, whereas individuals B
with low proficiency in both languages (so-called
semilinguals or weak bilinguals) are likely to face nega-
tive consequences, as compared to monolinguals. Thus,
in order to avoid negative cognitive consequences or to
profit by positive cognitive consequences, learners’
bilingual proficiency needs to exceed a minimum
threshold and to approach a higher threshold,
respectively.
Important Scientific Research and
Open Questions
The learning benefits that bilingual children may expe-
rience have been well identified. It has been shown that
children who are proficient in more than one language
(balanced bilinguals) and bilingual children who are
more proficient in one language than in the other (so-
called dominant bilinguals) outperform their monolin-
gual peers with respect to their ability to think and
reflect about language – competencies that are gener-
ally referred to as metalinguistic awareness. Interest-
ingly, advantages in terms of executive control
processes of bilinguals, as compared to monolinguals,
have not only been reported in linguistic tasks but also
in nonlinguistic domains. Bialystok and colleagues
(2009) argue that due to the intensive practice of inhi-
bition processes that go along with switching from one
language to another, the advantage of bilingual chil-
dren can be traced back to their improved ability to
selectively attend to specific aspects of a situation, par-
ticularly in ambiguous situations. This cognitive bene-
fit has shown to be relatively domain-independent,
thus giving learners an advantage in various fields
such as literacy, reading, and third language acquisition
as well as mathematics.
Findings concerning the positive effects of bilin-
gualism as well as the enthusiastic promotion of bilin-
gual education programs contrast, however, with
reports that highlight the serious school problems
encountered by students from immigrant families in
many developed countries. In fact, international com-
parative studies on school achievement such as PISA
revealed that bilingual students from immigration fam-
ilies performed significantly worse than their native
peers in all three domains of assessment: reading,
mathematics, and science (e.g., OECD 2006).
456 B Bilingualism and Learning
Throughout their school career, the risk for these stu-
dents to be retained is two to three times as high as that
of their “native” classmates. Can the problems faced by
students from immigrant families be linked to what
Cummins called Semilingualism or Weak Bilingualism?
The potential mechanisms underlying the cognitive
disadvantages, as proposed in Cummins’ model, have
not been as clearly defined as the advantages associated
with balanced and dominant bilingualism. Doubts
have been expressed whether the disadvantage is as
general as the advantage. Analyses of the PISA studies
have shown that the striking differences between immi-
grant children and their “native” classmates cannot be
attributed to general cognitive deficits, nor can they be
entirely traced back to their socioeconomic or cultural
backgrounds. Rather, mastering the instruction lan-
guage is the greatest barrier for students from immi-
grant families (OECD 2006). Deficits in mastering the
language of instruction have a cumulative effect in
most subjects, thus resulting in impaired competencies
in all academic fields for students with poor language
skills. This suggests that language proficiency is an
important determinant for academic success and that
teaching core subjects in a language in which students
are not proficient may be bound up with certain costs –
which may be a problem not only for students from
immigration families but also for students in immer-
sion programs. However, language proficiency may not
be the only potential source of costs associated with
bilingual learning environments. The promotion of bi-
or multilingual teaching implicitly assumes that the
knowledge we gain is represented in a system that is
independent of the specific language of instruction
and, thus, can be easily accessed no matter which lan-
guage the learner uses. This assumption, however, is far
from being empirically confirmed. It may well be that
languages not only serve as a medium of input for
learning, but also as a medium of representation for
the knowledge thus acquired. As a consequence, costs
may occur when knowledge needs to be transferred
from one representational system to another, that is,
when the language of learning is different from the
language that is used to apply what has been learned.
Evidence for this proposal comes from studies on the
relation between language and thought, which show
that the representation of information people take up
during their lifetime is affected by features of their
native language (Gentner and Goldin-Meadow 2003).
Furthermore, it has been found that bilingual learning
environments in various domains, including mathe-
matics, chemistry, and history, are bound up with
significant costs when the information taught has to
be transferred from one language to the other. In par-
ticular, when new facts are learned in one language,
these facts are more efficiently retrieved in the language
of instruction than in another language, no matter
whether it is L1 or L2 (e.g., Marian and Fausey 2006).
Given these findings, it is important for bilingual
educational programs to take into account the possible
costs and benefits of bilingual learning. There are
roughly three common forms of bilingual education
(cf. Baker 2006): Submersion programs, transitional
programs, and immersion programs. Submersion pro-
grams try to compensate the possible costs of language-
based problems as they are primarily designed to
improve immigrant children’s proficiency in the
instructional language. The idea is to have minority
language students learn the majority language by sim-
ple exposure to this language in the classroom. Typi-
cally, no special support is provided, and minority
language children and majority language children are
taught together. However, there are also programs that
provide native language support and give special sup-
port to minority language students by means of “pull-
out” classes.
Transitional programs also aim to promote profi-
ciency in the majority language. Unlike Submersion,
however, students in transitional programs are tempo-
rarily allowed to use their native language. Here, the
rationale is to gradually decrease native language use
during classes until students are able to cope with the
educational content in their second language. In other
words, children are initially taught in their native lan-
guage (e.g., the language in which they are proficient)
to prevent them from underperforming with respect to
their majority language peers. Thus, programs are
based on the assumption that concepts acquired in
one language are easily accessible in the other language.
However, the necessity for teachers to speak more than
one language limits the variety of students’ native lan-
guages. Hence, these programs are only suitable for
relatively homogenous language groups.
Finally, Immersion is a competence- and benefit-
orientated form of bilingual education. The explicit
goal of immersion programs is to reach high profi-
ciency in two languages (balanced bilingualism),

including bi-literacy. A key difference from Submer-
sion and transitional programs is that here, the decision
to raise children bilingually tends to be a voluntary one
and, as such, is opposed to the case of immigrant
children who have to learn the majority language in
order to take part, and function, in the society. One-
way Immersion is characterized by students from one
language group being taught in two languages. Chil-
dren are typically monolingual or start out from an
equal level of second language experience. Early Total
Immersion starts in Kindergarten, and the ratio of
using the two languages is balanced in most cases.
Middle and late Immersion start with one language
only and switch to a bilingual mode in grade two or
four, often via a step-by-step decrease of the initial
language. In Two-way Immersion programs, two lan-
guage groups, typically an equal number of majority
language and minority language students, make up one
classroom. Language use is balanced, but during each
period of instruction, only one language is used.
How well these programs are suited to compensate
for the special needs of disadvantaged bilingual stu-
dents and to promote the cognitive benefits associated
with balanced and dominant bilingualism is a question
for future research. In any case, the relation between
program characteristics and learning success may not
be onefold. Besides the design of the learning environ-
ment, the social, economic, political, and cultural con-
texts play an important role. Furthermore, to evaluate
the outcomes of bilingual educational programs, it is
important to not only look at cognitive and basic
educational measures such as oracy, literacy, and
numeracy but also at variables such as self-concept,
drop-out rates, tolerance, and integration.
Cross-References
▶ Language Acquisition and Development
▶ Literacy and Learning
▶ Psycholinguistics and Learning
References
Baker, C. (2006). Foundations of bilingual education and bilingualism
(4th ed.). Clevedon: Multilingual Matters.
Bialystok, E., Craik, F. I. M., Green, D. W., & Gollan, T. H. (2009).
Billingual minds. Psychological Science in the Public Interest, 10
(3), 89–129.
Cummins, J. (2004). Language, power and pedagogy: Bilingual children
in the crossfire. Clevedon/Buffalo/Toronto/Sydney: Multilingual
Matters.
Biographical Learning B 457
Gentner, D., & Goldin-Meadow, S. (2003). Language in mind:
Advances in the study of language and thought. Cambridge, MA:
MIT Press.
Marian, V., & Fausey, C. M. (2006). Language-dependent memory in B
bilingual learning. Applied Cognitive Psychology, 20, 1025–1047.
OECD. (2006). Where immigrant students succeed. A comparative
review of performance and engagement in PISA 2003. Paris: OECD.
Bimodal Learning
▶ Multimodal Learning
Biographical Learning
METTE KROGH CHRISTENSEN
Institute of Sports Science and Clinical Biomechanics,
University of Southern Denmark, Odense M, Denmark
Synonyms
Agency; Life course; Lifelong learning
Definition
The term “biographical learning” is used to describe
the study of the relationships that exists between learn-
ing and biography, the influence of biography on learn-
ing processes and practices, and biography as a mode of
learning (Tedder and Biesta 2007, p. 3). The word bio
means life, and it comes from the Greek word bios. The
word graphy means written or told, and it comes from
the Greek word grafia and grafein, which means
a record, a note, or a memorandum. A biography is
most often referred to as the story of a person’s life
written or told by somebody else, for example, “The
Education of John Dewey: A Biography” by Jay Martin.
In this way of understanding the word, a biography is
a literary genre and a style of writing like the novel and
the poem. Yet, the biographer (the subject) who is
writing the biography (the object) is likely to prefer
facts about a person’s life history instead of fiction even
if biographies may entail fictional elements. Biogra-
phers generally rely on a wide variety of documents
and viewpoints in the writing of a person’s biography.
By comparison, an autobiography (the word auto
458 B Biographical Learning
means self or same, and it comes from the Greek word
autos) may be based entirely on the writer’s life and his
or her own memory of that life. Unlike a biography, an
autobiography (and a memoir) is a life history of
a person written or told by that person himself or
herself, and thus intertwining the subject (the biogra-
pher) and the object (the biography). Some sociologists
and psychologists have noted that autobiography
offers the author the ability to revisit history,
because the autobiography is a way of telling, retelling,
and reinterpreting one’s own life history and
thereby re-creating a new biographical history.
According to some historians, the ability to both de-
and reconstruct a life also applies to the biography
(Bale et al. 2004, p. 22).
Biographical learning connotes the abovementioned
senses of the words biography and autobiography. There
does not yet seem to be an agreed-upon definition of
biographical learning, yet the two influential authors in
the field of biographical learning, the German sociolo-
gists and educationalists Peter Alheit and Bettina
Dausien, suggest the following definition:
" If we conceive of biographical learning as a self-willed,
“autopoietic” accomplishment on the part of active sub-
jects, in which they reflexively “organise” their experi-
ence in such a way that they also generate personal
coherence, identity, a meaning to their life history and
a communicable, socially viable lifeworld perspective for
guiding their actions, it becomes possible to compre-
hend education and learning both as individual identity
work and as the “formation” of collective processes and
social relations (Alheit and Dausien 2002, p. 17).
More simply expressed, biographical learning refers to
the ways in which a person’s biography both structures
and is structured by that person’s learning processes.
These learning processes are often characterized by
communicative situations, in which learners and teachers
(or other persons) meet in biographically oriented activ-
ities, dialogues, or learning environments. The starting
point of these processes are personal narratives (i.e., the
contextualized and intensified story of a given experi-
ence), life stories, and the transitional potential of re-
interpretation of these narratives and stories. In this
way, biographical learning is about understanding
changes in personal and social identity, as well as bodily
identity, as a potential for growth and ownership of
one’s own life story and “biographicity” (Alheit et al.
1995, p. 61; Alheit and Dausien 2002, p. 3), that is to say
the “hidden capacity” to lead one’s own life. According to
Alheit, the term “biographicity” means a person’s innate
capacity to redesign again and again, from scratch, the
contours of his/her own life within the specific contexts in
which he/she (has to) spend it, and in which he/she
experiences these contexts as “shapeable” and designable
(Alheit et al. 1995, p. 65). Therefore, biographical learn-
ing differs from the broader concept of lifelong learning,
which may be defined as a concept spanning an entire
lifetime in a process of transforming experience into
knowledge, skills, attitudes, values, emotions, beliefs,
and the senses (Smilde 2008).
Theoretical Background
The philosophies and assumptions underpinning bio-
graphical learning derive from and influence sociology,
pedagogy, and anthropology. Since John Dewey
launched the concept of educative experience in his
book “Education and Experience” many interpreta-
tions of experience have entered the discourse of learn-
ing, including biographical learning. At the outset,
Dewey emphasized the continuity of experience, mean-
ing that every experience both takes up something from
those which have gone before and modifies in some
way the quality of those which come after. He also
pointed out that only when the experience is the object
of reflection is it considered to be an educative experi-
ence. The concept of biographical learning has central
points of resemblance to Dewey’s thoughts on experi-
ence and progressive education, especially with regards
to the intertwining of past and future experiences in
a person’s biographical learning processes, and the
point of view that the organization of biographically
oriented learning (processes) must be communitarian,
lifeworld-oriented, project-related, and interest-related
(Alheit 1994, p. 293).
In the early 1970s, the life history approach in
education stressed the importance of the emancipatory
aspect of including the (often contradictory) experi-
ences and life histories of the learner in educational
connexions (Elias and Merriam 1980). The aim of this
experience-oriented and often critical pedagogy is to
describe and argue in support of the individual life
history as a potential for the awakening and emanci-
pating qualities of the learning process. Oskar Negt and
Paulo Freire, who were very influential and critical
educationists in the 1970s, have inspired the viewpoint

that experience is an active, continuous, critical, and
creative process and that experience as such is collec-
tive, which means it is tied to social structures.
The point is that by taking a biographical approach
in adult education, it is possible to reveal distinctive
processes that on the one hand emerge from
certain specific conditions and therefore are similar
(e.g., discrimination as a cultural, social, or political
phenomenon), but on the other hand differ from
other processes (e.g., discrimination as a personal
experience). This is a central argument for using
a biographical approach in pedagogical matters: the
intimate and already existing interrelation between
individuality and society, for which the French sociol-
ogist Pierre Bourdieu in particular stands. According to
Bourdieu, the interrelation between individuality and
society exists in the “habitus,” that is, the individual’s
embodied social history and perceptual dispositions
developed through engagement in social life and espe-
cially educational contexts (Bourdieu 1998, p. 25).
Following Bourdieu’s argument, it can be said that
learning and teaching is always a matter of how
biographicity and society meet in educational connex-
ions or in learning processes (Christensen 2005).
Contemporary theories of subjectivity and individ-
ualization argue that modern society is giving a new
importance to the biographical perspective of the agent
and to biographicity as the underlying structure
of meaning construction and learning. Subjectivity
and the ongoing process of making identities are seen
as dominant features of today’s society which actualizes
a biographical approach to adult education and the
concept of biographical learning in understanding
learning processes in modern society. Concurrently,
the turn to biographical research methods in social
sciences marks a trend toward the agent perspective in
order to reveal the smaller history as a representation
of the bigger history, because the biographical
approach lays stress on valuing knowledge of
a personal history in arriving at an understanding of
how people learn, what they learn, and the choices
which people make.
Biographical Learning as
a Pedagogical Method
In the biographical approach to education, the life
history of the single learner plays a central role, because
continuous holistic narratives told in everyday
Biographical Learning B 459
language is assumed to reach deeper layers in the
social reality ways of functioning. When a person tells
about his or her life, it is done with the help of explan- B
atory models that are legitimate in his or her
culture (Christensen 2007). In this way, insights in
written or told life histories allow for an understanding
of a person’s way of structuring his or her life. They also
provide a connected picture of the different practical-
ities and limitations which structure a person’s
life. Biographical learning as a pedagogical method,
by which a teacher encourages learners to develop
a personal, sensuous language about their experiences,
may support learners’ involvement in dialogues
and narrative activities, which form the basis for
personal narratives surrounding concrete meaningful
experiences from everyday life (Christensen 2007).
The pivotal concern in the biographical approach is
subjectivity, represented in individual ways of doing
and ways of thinking. According to Bourdieu (1998),
subjectivity is embedded in the habitus: in the individ-
ual’s embodied life history and perceptual dispositions
developed through engagement in social life. As such
subjectivity is a person’s capability and intention to
understand and to experience and evaluate oneself in
a life course perspective. Thus, the biography of
a learner can be seen as a faculty of experience, through
which he or she learns to deal with subject matters in
relation to education. Following this point of view,
assimilation of educational aims has to be seen as
integrated with the self-identity, life history, and bio-
graphical reflections of a learner.
Important Scientific Research and
Open Questions
A biographical approach in studying lifelong learning
and learning careers is widespread in the literature on
teacher education, teacher development, and profes-
sional identity (e.g., Goodson 2003), mostly by the
use of biographical research methods such as in-depth
life-history interviews. However, research into bio-
graphical learning as such is rather infrequent. How-
ever, there are a number of important studies. The
work of Alheit et al. (1995) and Alheit and Dausien
(2002) showed that in a risk society the biographical
approach in adult education is all-important, because
biography itself has become a “laboratory” in which
new constellations of ideas, knowledge, and self-
identity have to be anticipated and even experienced.
460 B BIOL - Bonding and Identification-Based Observational Learning
Also, findings from The Learning Lives (see www.
learninglives.org), a 3 year research project focusing
on the interrelationships between learning, identity,
and agency in people’s lives suggest among other things
that learning about one’s agent-related orientations
and learning how to reframe a particular agent-related
“constellation” can be important in shaping our
responsiveness and hence in achieving agency. The
authors argued that for such learning to occur, it is
important that actors can distance themselves from
their actions in order to explore and evaluate them
(Biesta and Tedder 2007; Tedder and Biesta 2007).
Furthermore, the recent research into specific subject
fields such as physical education (Christensen 2007),
music (Smilde 2008), and university sociology
teaching (McLean and Abbas 2009) suggest that
biographical learning viewed as a pedagogical method
by which the learners’ life stories and narratives forms an
entrance to teaching and learning specific subject mat-
ters in educational connexions. Several questions remain
open: Are narratives a necessary condition for biograph-
ical learning? What is the learning potential of different
kinds of narratives? How and with which outcomes
are biographically oriented teaching methods employed
in formal educational settings? Is biographical learning
a European/Western phenomenon exclusively for
adult learners or does the concept have broader scope?
Cross-References
▶ Adult Learning Styles
▶ Adult Learning Theory
▶ Adult Learning/Andragogy
▶ Adult Teaching and Learning
▶ Autobiographical Memory
▶ Education of Teacher Educators
▶ Lifelong and Worklife Learning
▶ Life-Long Learning
References
Alheit, P. (1994). The “biographical question” as a challenge to
adult education. International Review of Education, 40(3–5),
283–298.
Alheit, P., & Dausien, B. (2002). The double face of lifelong learning:
Two analytical perspectives on a ‘silent revolution’. Studies in the
Education of Adults, 34(1), 3–22.
Alheit, P., Bron-Wojciechowska, A., Brugger, E., & Dominicé, P.
(Eds.). (1995). The biographical approach in European adult edu-
cation. Wien: Verband Wiener Volksbildung.
Bale, J., Christensen, M. K., & Pfister, G. (2004). Writing lives in sport.
Biographies, life-histories and methods. Aarhus: Aarhus University
Press.
Biesta, G., & Tedder, M. (2007). Agency and learning in the lifecourse:
Towards an ecological perspective. Studies in the Education of
Adults – Special Issue, 39(2), 132–149.
Bourdieu, P. (1998). Practical reason. On the theory of action. Cam-
bridge: Polity.
Christensen, M. K. (2005). Experience as biographical learning:
Inspirations for physical education. Moving Bodies, 1(3), 21–42.
Christensen, M. K. (2007). Biographical learning as health promotion
in physical educations? A Danish case study. European Physical
Education Review, 13, 5–24.
Elias, J. L., & Merriam, S. B. (1980). Philosophical foundations of adult
education. Huntington: Robert E. Krieger Publishing Company.
Goodson, I. F. (2003). Professional knowledge, professional lives:
Studies in education and change. Buckingham: Open University
Press.
McLean, M., & Abbas, A. (2009). The ‘biographical turn’ in university
sociology teaching: A Bernsteinian analysis. Teaching in Higher
Education, 14(5), 529–539.
Smilde, R. (2008). Lifelong learners in music; research into musi-
cians’ biographical learning. International Journal of Community
Music, 1(2), 243–252.
Tedder, M., & Biesta, G. (2007). Learning from life and learning for
life: Exploring the opportunities for biographical learning in the
lives of adults. Working paper 7, Learning Lives Website, www.
learninglives.org. Learning Lives Website.
BIOL - Bonding and
Identification-Based
Observational Learning
▶ Selective Attention in Social Learning of Vervet
Monkeys
Biological Function of Learning
▶ Adaptation and Learning
Biological Groundings of
Reinforcement Learning
▶ Basal Ganglia Learning

Biological or Evolutionary
Constraints on Learning
MICHAEL DOMJAN
University of Texas, Austin, TX, USA
Synonyms
Adaptive specializations in learning; Constraints on
learning
Definition
A biological or evolutionary constraint on learning is
a limitation on classical or instrumental conditioning
that is observed despite the use of procedures that
would be expected to produce successful learning.
Such constraints on learning reflect the evolutionary
history of the organism and are usually contrasted with
examples of unusually rapid learning with condition-
ing procedures that employ a different conditioned
stimulus or reinforcer.
Theoretical Background
Traditionally, investigators of basic learning phenomena
such as classical and instrumental conditioning sought to
identify universal rules of learning that apply irrespective
of the specific stimuli and responses the subjects are
learning about. The assumption that learning processes
are universal and independent of the specific stimuli and
responses involved is called the equipotentiality principle
and is the basis for general process learning theory. The
equipotentiality principle encouraged investigators to
focus on a few well-developed laboratory paradigms
such as rats lever-pressing or running down a straight
alley for food or pigeons pecking a response key for
food. The assumption was that the results obtained
using standard laboratory procedures and species
would be applicable to other species and other stimuli,
responses, and reinforcers.
Problems with the principle of equipotentiality
became evident early in the study of instrumental con-
ditioning when E. L. Thorndike discovered that unlike
pulling a string or releasing a latch, grooming responses
such as scratching and yawning did not increase in
frequency with food reinforcement. These observations
by Thorndike were put aside for more than 50 years, as
Biological or Evolutionary Constraints on Learning B 461
most of the major learning theorists of the twentieth
century espoused some version of equipotentiality in
their advocacy of a general process approach to learn- B
ing. However, ignoring biological constraints on learn-
ing became increasingly difficult with the discovery of
numerous additional examples in the 1960s (Domjan
1983; Hinde and Stevenson-Hinde 1973; Shettleworth
1972). As one might expect, these discoveries occurred
when investigators examined learning in unusual spe-
cies or situations. Keller and Marian Breland, for exam-
ple, attempted to condition a raccoon to put a coin into
a slot for food reinforcement, as a part of an entertain-
ment display. As training progressed, the raccoon
became increasingly reluctant to let go of the coin and
let it drop into the slot. When required to put two coins
into the slot, the raccoon rubbed the coins together
instead of releasing them into the slot. Attempts to
train a pig to put a coin into a piggy bank met with
similar failure. The pig would root the coin on the
ground rather than put it into the slot.
Another major constraint on learning was discov-
ered by John Garcia in the course of his work on the
biological effects of x-irradiation. Garcia and his col-
leagues found that foot-shock will readily condition an
aversion to an audiovisual cue but not a taste cue in
rats. In contrast, illness induced by irradiation or
a drug injection will condition an aversion to a taste
but not to an audiovisual stimulus. Constraints on
learning were also found in avoidance conditioning.
Bolles found that rats are more likely to learn to freeze,
jump out of a box, or run in a wheel to avoid shock than
they are to rear or press a lever. In pigeons, key pecking
is much more difficult to condition as an instrumental
shock-avoidance response than is pressing a treadle.
The constraints on learning discovered in the 1960s
were often contrasted with especially rapid learning
using similar procedures. For example, although rats
have difficulty learning to press a response lever to
avoid shock, they readily freeze if freezing is designated
as the avoidance response. Although pigeons have dif-
ficulty learning to peck to avoid shock, pecking is
quickly learned to obtain food. The failure of rats to
learn an aversion to an audiovisual cue paired with
illness contrasts with very rapid learning of an aversion
to a taste that is paired with illness. In fact, taste aver-
sions can be learned in a single trial even if the illness
follows the taste exposure by several hours. Because
462 B Biological or Evolutionary Constraints on Learning
biological constraints on learning are often discussed in
contrast to examples of highly rapid learning with
similar training parameters, they are assumed to reflect
an adaptive specialization that facilitates some
instances of learning and disrupts others.
Thorndike proposed the concept of belongingness to
explain the failure of grooming responses to be
reinforced with food. He suggested that an instrumen-
tal response has to “belong” with the reinforcer for
instrumental learning to occur. The Brelands intro-
duced the concept of instinctive drift to explain the
failures of conditioning they observed, because the
difficulties that raccoons and pigs had in releasing
coins into a slot to obtain food seemed to be due to
the intrusion of incompatible instinctive behaviors
related to feeding in these species.
Belongingness and instinctive drift were offered to
explain specific instances of constraints on learning.
A more comprehensive approach was provided by
Seligman, who introduced the concept of preparedness
(Seligman and Hager 1972). According to this concept,
learning phenomena can be ordered along
a continuum of preparedness, with instances of learn-
ing that occur with relatively little training or input at
one end and instances of learning that occur very
slowly and only with a great deal of training at the
other end. Highly prepared forms of learning (e.g.,
taste–illness associations) presumably reflect evolu-
tionary preparation for this type of learning. In con-
trast, contraprepared forms of learning at the other end
of the continuum presumably reflect tasks that are
contrary to the evolutionary legacy of the organism.
The concept of preparedness provided a convenient
way to describe both constraints on learning and
contrasting instances of highly rapid learning. How-
ever, the concept of preparedness turned out to be
merely descriptive rather than explanatory. The degree
of preparedness could only be identified by how rapidly
or slowly something was learned. This made prepared-
ness a circular explanation for constraints on learning,
because a constraint could only be identified by a slow
learning rate. The inherent circularity of the concept of
preparedness made it difficult to predict new instances
of constraints or adaptive specializations of learning in
the absence of measurements of learning rate.
The next major development in efforts to under-
stand biological constraints on learning was behavior
systems theory developed by Timberlake (2001).
Behavior systems theory assumes that evolution has
shaped behavior into functionally organized units
that enable organisms to successfully accomplish bio-
logically important tasks such as feeding, defense, and
reproduction. Each behavior system consists of
a hierarchy of response modes and modules, each of
which determines the responses the organism is likely
to perform and stimuli it is likely to pay attention to.
Learning procedures are superimposed on the behavior
system that is activated at the time of training. For
example, food deprivation and food reinforcement
activate the feeding system. During food deprivation,
responses and stimulus relations that are compatible
with the feeding system will be easily learned, whereas
those that are incompatible with the feeding system will
be especially difficult to learn.
Behavior systems theory avoids the circularity of
the concept of peparedness because the response
modes and modules of a behavior system can be iden-
tified and characterized independently of a particular
classical or instrumental conditioning procedure. Fea-
tures of the feeding system, for example, can be char-
acterized by observing an animal’s behavior while it is
food deprived and periodically receiving food. The
response modes and modules identified during such
investigations of the feeding system can then be used to
predict how the animal will respond to a classical or
instrumental conditioning procedure that involves
food reinforcement.
Behavior systems theory acknowledges that
response modes and modules have evolved because of
their adaptive functions. However, it focuses on the
outcomes of that evolutionary process rather than the
ecological context in which evolution occurs. Another
approach, proposed by Domjan (2008), emphasizes
that evolutionary selection cannot be divorced from
the ecological context in which the selection takes
place. Behavioral processes that are adaptive and con-
tribute to reproductive success in one ecological niche
may not be as successful in another, and may even be
deleterious. These considerations encourage looking
for biological influences on learning by examining
how learning takes place in an organism’s natural
habitat.
An organism’s natural habitat will limit the types
of learning than are likely to take place. Consider, for
example, classical conditioning. We know that organ-
isms will associate a conditioned stimulus (CS) with
 Blended Learning B 463

food only if that CS is reliably encountered just before
food. For this to occur in nature, the CS has to be
a natural precursor of food or a component of the
causal chain of events that invariably leads to food.
Evolutionary selection that improves an organism’s
ability to predict and ingest food is likely to favor the
conditioning of stimuli that are natural precursors of
food rather than the conditioning of arbitrary
unrelated events. These considerations suggest that
conditioning of natural precursors of food should
occur more quickly and be less sensitive to disruptions
than the conditioning of arbitrary cues. The most
extensive tests of these ideas have been carried out in
studies of sexual conditioning. Experiments have
shown that in comparison to arbitrary cues, naturalis-
tic conditioned stimuli become conditioned more rap-
idly, come to elicit a broader range of conditioned
responses, and learning about naturalistic cues is resis-
tant to increases in the CS–US interval, the blocking
effect, and extinction.
(Ed.), Learning theory and behavior. (Vol 1, J. Byrne (Ed.) Learn-
ing and memory: A comprehensive reference (pp. 327–340)).
Oxford: Elsevier.
Hinde, R. A., & Stevenson-Hinde, J. (1973). Constraints on learning. B
New York: Academic Press.
Seligman, M. E. P., & Hager, J. L. (1972). Biological boundaries of
learning. New York: Appleton.
Shettleworth, S. J. (1972). Constraints on learning. In D. S. Lehrman,
R. A. Hinde, & E. Shaw (Eds.), Advances in the study of behavior
(Vol. 4, pp. 1–68). New York: Academic.
Timberlake, W. (2001). Motivational modes in behavior systems. In
R. R. Mowrer & S. B. Klein (Eds.), Handbook of contemporary
learning theories (pp. 155–209). Mahwah: Erlbaum.
Biologically Inspired
Computing
▶ Learning to Sing Like a Bird: Computational Devel-
opmental Mimicry

Important Scientific Research and

Open Questions
Biological constraints on learning have encouraged Bi-sensory Learning
investigators to accept that evolution has significantly
shaped and constrained what animals learn about and ▶ Audiovisual Learning
how learning occurs. This in turn has led to abandon-
ment of the principle of equipotentiality. However,
much remains to be discovered about the specific
ways in which evolutionary adaptation shapes learning
mechanisms. In particular, we need to examine in
Blended Instruction
greater detail how the particulars of an organism’s ▶ Integrated, Multidisciplinary, and Technology-
ecological niche determine the course of evolution in Enhanced Science Education
producing specialized forms of learning.

Cross-References
▶ Classical Conditioning
▶ Instrumental Conditioning Blended Learning
▶ Learning and Instinct
▶ Natural Learning DIRK IFENTHALER
Institut für Erziehungswissenschaft, Albert-Ludwigs-
References University Freiburg, Freiburg, BW, Germany
Domjan, M. (1983). Biological constraints on instrumental and clas-
sical conditioning: Implications for general process theory. In
G. H. Bower (Ed.), The psychology of learning and motivation
(Vol. 17, pp. 215–277). NewYork: Academic.
Synonyms
Domjan, M. (2008). Adaptive specializations and generality of the Blended online learning; Flexible learning; Hybrid
laws of classical and instrumental conditioning. In R. Menzel learning; Mixed eLearning
464 B Blended Learning
Definition
Blended learning combines face-to-face with technol-
ogy-based learning and instruction. The terms
“blended learning” and “hybrid learning” are used
interchangeably in instructional science. However,
a blended learning scenario is also regarded as
a combination of different computer-based technolo-
gies, a combination of different pedagogical approaches
and theories, or a combination of instructional technol-
ogy with specific tasks in order to facilitate desired skills
and competencies.
Theoretical Background
The digital age and its rapid introduction of informa-
tion and communication technologies (ICT) opened
up new possibilities for learning and instruction. Par-
allel to the introduction of technology-based learning
environments, the term “blended learning” was intro-
duced. However, a precise meaning of blended learning
is still discussed (Graham and Dziuban 2008). Accord-
ingly, various definitions of blended learning can be
found in educational science (Seel and Ifenthaler
2009).
The most accepted definition of blended learning is
a combination of traditional face-to-face with com-
puter-based learning and instruction (Kerres and de
Witt 2003). Traditional face-to-face learning and
instruction requires learners and teachers being at the
same place at the same time in order to communicate
and exchange information. Using computer-based
learning and instruction, learners and teacher do not
have to meet at the same place and the same time to
communicate and exchange information. Hence, infor-
mation and communication technologies are used to
facilitate learning and instruction. Meanwhile, not only
desktop computers are used for blended learning sce-
narios. Newer technological developments allow the
implementation of blended learning environments on
cellular phones, television sets, or games consoles.
Another definition of blended learning is the com-
bination of tools, applications, and media in
a computer-based or web-based learning environment.
Here, the traditional face-to-face learning and instruc-
tion is disregarded and a mixture of various media
application is used to facilitate learning and instruction
in an optimal combination. Media applications include
interactive texts, forums, chats, wikis, portfolios,
podcasts, simulations, games, quizzes, assessment
tools, authoring tools, as well as video- and audio-
streaming. Yet, this definition is criticized because any
learning situation includes a variety of media what
makes this explanation of blended learning redundant
(Oliver and Trigwell 2005).
One more definition of blended learning is the
combination of pedagogical approaches or learning
theories. However, a mix of pedagogic approaches
prompt new questions for educational researchers:
What are the benefits of combining different pedagog-
ical approaches? Is such a combination more effective
than using only one alternative approach? Does such
a combination affect the learning objectives negatively?
Last, Kerres and de Witt (2003) introduce plausible
reasons for blended learning environments. Blended
learning environments provide pedagogical richness,
give flexible access to information, induce social inter-
action and collaboration, enable a fast and economic
revision of learning content, provide alternative assess-
ment possibilities, and are to a great extent cost-effec-
tive. However, learners and instructors need to bring
along specific skills and competencies to being able to
successfully utilize blended learning environments.
Still, the theoretical foundation of blended learning
needs to be developed in future. Especially, cognitive,
affective, and social aspects of learning need to be
integrated in a well-defined theoretical basis of blended
learning.
Important Scientific Research and
Open Questions
Blended learning is lively discussed in educational and
instructional science. Most importantly, a blended learn-
ing environment is always dependent on its combined
components, e.g., face-to-face learning and instruction
and technology-based learning and instruction. Accord-
ingly, all components need to be optimally calibrated in
order to facilitate learning and instruction at its best. Key
research on blended learning is conducted including the
following perspectives: Focus on students, focus on
instructors, as well as focus on organizations and insti-
tutions (Graham and Dziuban 2008).
Researchers investigate the relation of motivation
and satisfaction of students induced by blended learn-
ing environments. Studies report high motivation and
satisfaction of students in blended learning environ-
ments. Other researchers have studied collaboration
and the sense of community within blended learning

environments. Rovai and Jordan (2004) found a higher
sense of community existed within blended learning
environments compared to face-to-face and full online
learning environments. Concerning instructors,
research studies investigate models for supporting
instructors to implement blended learning environ-
ments and how satisfied instructors are. A problem
found is the double role of being a face-to-face teacher
and an online facilitator. This requires new skills and
competencies from instructors using blended learning
environments. Another important research question
investigated includes the relation of workload of
instructors and blended learning environments. Using
technology-based learning and instruction requires
more time and effort (Kaleta et al. 2007). As
a consequence, specialized trainings for instructors
using blended learning environments are inevitable to
optimize the implementation of technology-based ele-
ments into courses. Furthermore, a successful imple-
mentation of blended learning environments is highly
dependent on the organizational conditions. Little
research is available on the conditions needed for suc-
cessfully introducing blended learning into institu-
tions. Here, new theories of practice are required to
give those responsible the necessary background for the
implementation of blended learning environments.
Open questions arise from the development of
future technologies. Will blended learning environ-
ments change with regard to their combination of
face-to-face and technology-based elements? Will dif-
ferent organizations, e.g., school, universities, compa-
nies, etc., use blended learning environments in
different ways? What skills and competencies will stu-
dents and instructors need to successfully learn and
teach in blended learning environments? What impact
will future blended learning environments have on
traditional academic institutions?
Cross-References
▶ Collaborative Learning
▶ Computer-based Learning
▶ Distance Learning
▶ Distributed Technologies
▶ e-Learning
▶ Learning Management Systems
▶ Learning with and from Blogs
▶ Online Learning
▶ Seamless Learning
Blocking B 465
References
Graham, C. R., & Dziuban, C. (2008). Blended learning environments.
In J. M. Spector, M. D. Merrill, J. van Merrienboer, & M. P. Driscoll B
(Eds.), Handbook of research on educational communications and
technology (pp. 269–276). New York: Taylor & Francis.
Kaleta, R., Skibba, K., & Joosten, T. (2007). Discovering, designing,
and delivering hybrid courses. In A. G. Picciano & C. Dziuban
(Eds.), Blended learning: Research perspectives (pp. 111–144).
Needham: Sloan Consortium.
Kerres, M., & de Witt, C. (2003). A didactical framework for the
design of blended learning arrangements. Learning, Media and
Technology, 28(2 & 3), 101–113.
Oliver, M., & Trigwell, K. (2005). Can “blended learning” be
redeemed? E-Learning, 2(1), 17–26.
Rovai, A. P., & Jordan, H. M. (2004). Blended learning and sense of
community: A comparative analysis with traditional and fully
online graduate courses. International Review of Research in Open
and Distance Learning, 5(2), 1–3.
Seel, N. M., & Ifenthaler, D. (2009). Online lernen und lehren.
München: Ernst Reinhardt Verlag.
Blended Online Learning
▶ Blended Learning
Blended Research on Learning
▶ Mixed Methods Research on Learning
Blocking
The decrease in conditioned responding that results
from reinforcing a novel stimulus in the presence of
a trained stimulus instead of in the presence of another
novel stimulus. First reported by Leo Kamin in 1969,
blocking challenged the view that contiguous stimulus
pairings are sufficient for Pavlovian learning. It is one
of a class of cue-competition phenomena that led to the
view that conditioned responding results when
a stimulus predicts an increase in reinforcer likelihood.
References
Kamin, L. J. (1969). Predictability, surprise, attention, and conditioning.
In B. A. Campbell & R. M. Church (Eds.), Punishment and aversive
behavior (pp. 279–296). New York: Aappleton-Century-Crofts.
466 B Bloom’s Domain
Bloom’s Domain
▶ Bloom’s Taxonomy of Learning Objectives
Bloom’s Model of School
Learning
NORBERT M. SEEL
Department of Education, University of Freiburg,
Freiburg, Germany
Synonyms
Conceptual model of school learning
Definition
Bloom’s model of school learning emphasizes the cycli-
cal character of instruction, the outcomes of one school
learning task becoming the input to another. In
Bloom’s model all learning outcomes are determined
by four dimensions: Students’ cognitive entry behav-
iors, affective entry characteristics, the learning task(s),
and the quality of instruction. The learning outcomes
or results are specified as level and type of achievement,
or as rate of learning, or as affective outcomes. The
model relates to series of learning tasks.
Theoretical Background
In the 1970s, Bloom (1976) introduced a model of
school learning that transcends his former model of
learning for mastery (Bloom 1971) and focuses on
characteristic group situations which are representative
for school learning. Bloom’s model aims at changes of
the educational system in order to achieve a substantial
improvement of the student’s outcomes and
performances.
In Bloom’s model, all learning outcomes are
determined by four dimensions: Cognitive predisposi-
tions (cognitive entry characteristics), affective entry
characteristics, the learning task(s), and the quality of
instruction. More detailed, the model can be illustrated
as in Fig. 1.
With regard to the characteristics of students,
Bloom adopts the traditional distinction between cog-
nitive and affective components. Cognitive entry
characteristics are defined as knowledge, abilities and
competences necessary for completing particular tasks.
They constitute specific preconditions and constraints
for given learning tasks. As in Carroll’s (1963) model of
school learning, generalized cognitive capabilities, such
as general intelligence and reading abilities, are consid-
ered as influential on a variety of learning tasks. Beyond
these general cognitive capabilities, Bloom refers to
Glaser (1970) and discusses accordingly five categories
of cognitive entry characteristics: (1) task learning
already acquired, (2) prerequisite learning or prior
knowledge, (3) cognitive styles, (4) task-specific abili-
ties, and (5) general mediating abilities.
Affective entry characteristics include motivation,
enthusiasm for given learning tasks, as well as a positive
self-concept with regard to achievement capabilities and
positive attitude toward schooling and learning.
Bloom argues that cognitive entry characteristics
have the critical influence on accomplishment of single
tasks, and therefore, on the learning outcomes in
general. More specifically, even the best instructional
quality is ineffective in the case of lacking the relevant
task-specific entry characteristics. The quality of
instruction contains four constituents: Cues, reinforce-
ment, participation, and feedback/correctives.
Cues refer to all information concerning the pre-
sentation and explanation of the learning task. Most of
them are verbal but occasionally they are also pictorial.
Bloom points out that this aspect of instructional qual-
ity corresponds with Carroll’s concept of quality of
instruction and includes presentation, explanation,
and sequencing of tasks in correspondence with the
characteristics of individual learners.
Participation corresponds with the efforts in
accomplishing a task or with the engagement in
a task. Regularly, participation is expressed in measures
of active learning time (similar to Carroll’s
concept of actual time needed for learning). Bloom
argues that participation or active learning can occur
covertly (referring to internal thought processes) or
overtly (i.e., as observable interactions with learning
material).
Reinforcement refers to affective responses to stu-
dents’ behavior and contains praise, blame, supportive
or discouraging statements of the teacher. The effective
application of reinforcement depends on the individual
student, his/her advancements and the difficulty of the
task to be mastered.
 Bloom’s Model of School Learning B 467

Students’ Instruction Outcomes

characteristics of learning

Cognitive entry
Level and type
of achievement
B
characteristics

Learning
task(s) Rate of learning

Affective entry Affective

characteristics outcomes
Quality of
Instruction
Cues

Reinforcement

Participation

Feedback/
correctives

Bloom’s Model of School Learning. Fig. 1 Bloom’s model of school learning

Feedback/correctives are included in the course of
instruction as a kind of particular cues and reinforce-
ments in order to provide the learners with additive
information for successful accomplishment of the
learning tasks. In regular classroom instruction, the
feedback often consists in formative tests; corrective
procedures contain specific encouragements or chal-
lenges for students to improve learning. Feedback
refers to the identification and notifying of that what
the students has learned, whereas correctives are hints
and activities aiming at the improvement of a student.
Reinforcement and feedback are closely related to each
other in instructional practice.
With regard to the outcomes of learning Bloom
distinguishes between (a) level and type of achieve-
ment, (b) rate of learning, and (c) affective outcomes.
The character of achievement is determined through
objectives and contents of the tasks. Interestingly,
Bloom argues in favor of a criterion-referenced testing
of learning outcomes. Rate of learning is defined in
terms of activity quantity within a unit of a time
period. Unfortunately, Bloom did not specify the con-
cept rate of learning in more detail and it remains
unclear whether it can be improved by instruction.
Concerning the affective learning outcomes it can be
assumed that the effects of instruction have presumably
the strongest effects on particular attitudes.
Learning tasks are at the center of Bloom’s model of
school learning. Alternatively, learning tasks are also
named learning units. A learning task or unit can
contain one lesson up to ten lessons or also a book
chapter. Learning tasks or units can be independent of
each other or interrelated with each other, or they can
be organized hierarchically. It is in dependence on the
degree of their structuredness can learning tasks be
related to prior steps in the learning process.
A particular strength of Bloom’s model is the
uncovering of the cumulative interdependence of
learning tasks in schooling. Each task precedes another
one so that mastering a task becomes a precondition for
mastering the following tasks. This feature of the model
facilitates the explanation of cumulative effects of
instruction on learning and constitutes a fundamental
basis for the analysis of long-term effects of instruc-
tional strategies.
Important Scientific Research and
Open Questions
Bloom’s model of school learning explains cumulative
learning in the classroom in dependence on learner
468 B Bloom’s Model of School Learning
characteristics and instructional quality, discussed
in terms of learning tasks or units. However, due
to some lack of clarification of interdependences
between the various components of the model it can
be criticized for shortcomings (Harnischfeger and
Wiley 1978).
Bloom argues, for instance, that accomplishment of
tasks presupposes specific cognitive entry characteris-
tics considered as dichotomous (i.e., noncontinuous)
variables. He also maintains that all learners who pos-
sess these specific entry conditions can master each
learning task if they are sufficiently motivated and if
the quality of instruction corresponds with their needs
for learning. Moreover, Bloom asserts that there is only
a small variation of learning tempo as well as in the
level of achievement if the learners possess the appro-
priate entry conditions. This corresponds with Bloom’s
idea of learning for mastery. Cognitive entry character-
istics can change for a particular task due to changes of
the instructional method. As long as content and objec-
tives do not change, a substantial change of the learning
task is not necessary. This results in an adaptation of
instruction to the needs and characteristics of learners.
This includes implicitly a dichotomy of the entry
conditions in accordance with task-specific conditions
and conditions of particular instruction.
In addition to cognitive preconditions, affective
entry characteristics play an important role in Bloom’s
model inasmuch as they affect achievement and rate of
learning. The structure of affective entry characteristics
corresponds with the level of specificity of its relation-
ship with the learning task. The lowest level of speci-
ficity is associated with affects related to subject matter
of instruction; the next levels correspond with school-
related affects and, finally, with achievement-related
self-concept. Bloom maintains that these levels can be
distinguished during the first years of schooling but
merge to a comprehensive characteristic over time.
Bloom argues that students differ significantly with
regard to the time they need for accomplishing learning
tasks. Harnischfeger and Wiley (1978) believe in two
reasons for these differences of the needed learning time.
On the one side, students vary in the contingent of
time allowed for task learning, and therefore, in the
overall active time of learning within a particular
time interval. On the other side, there is a different
efficiency of using time allowed for task learning. It
remains unclear whether Bloom has distinguished
between these two concepts or whether the active
time of learning contains the efficiency of learning.
In view of Bloom’s discussion of the problem results
an immediate interpretation: When cognitive
entry characteristics are completely given then the learn-
ing efficiency or rate of learning does not vary. There-
fore, it is possible that learning efficiency is determined
only by the cognitive entry characteristics and the qual-
ity of instruction.
Despite these critical comments it is noteworthy
that the contributions of Carroll and Bloom to the
theoretical underpinning of research on time and
learning cannot be overemphasized as Goodman
(1990) says. Actually, the time concept, first introduced
by Carroll, has been translated into a working model
for classroom instruction by Bloom who explains how
student characteristics (cognitive and affective) and
instructional variables interact with time and partici-
pation to produce effective learning for mastery. Time
is both an intervening and outcome variable inasmuch
as it contributes to the quality of instruction as well as
to instructional effectiveness (Bloom 1974). Compara-
ble with Carroll’s approach, the working model of
Bloom integrates a complex set of student, instruc-
tional, and situational variables that interact in order
to facilitate cognitive and affective learning outcomes.
Within Bloom’s model it is the degree of student par-
ticipation in learning which intervenes between
instruction and learning outcomes and the rate of
learning is viewed as an indicator of learning efficiency.
Optimal learning conditions, including sufficient time
for accomplishing the learning task(s) as well as an
adaptation to cognitive and affective characteristics,
increase student achievement and reduce the amount
of time needed to achieve mastery. In Goodman’s
(1990) view, Bloom’s model can be considered a good
example for the methodology of process–product
research which provides the empirical framework for
the study of cause and effect relationships within the
instructional setting. The process–product paradigm
focuses on instructional dynamics at the school, class-
room, and student levels in order to achieve effective
learning in the classroom.
Cross-References
▶ Academic Learning Time
▶ Carroll’s Model of School Learning
▶ Learning Tasks

References
Bloom, B. S. (1971). Learning for mastery. In B. S. Bloom, J. T.
Hastings, & G. F. Madaus (Eds.), Handbook on formative and
summative evaluation of student learning (chap. 3). New York:
McGraw-Hill.
Bloom, B. S. (1974). Time and learning. The American Psychologist,
29, 682–688.
Bloom, B. S. (1976). Human characteristics and school learning.
New York: McGraw-Hill.
Carroll, J. B. (1963). A model for school learning. Teachers College
Record, 64, 723–733.
Glaser, R. (1970). Evaluation of instruction and changing educational
models. In Proceedings of the Symposium on Problems in the
Evaluation of Instruction, University of California, Los Angeles,
December, 1967. Los Angeles: Center for the Study of Evaluation
of Instructional Programs, University of California.
Goodman, L. (1990). Time and learning in the special education
classroom. Albany: State University of New York Press.
Harnischfeger, A., & Wiley, D. E. (1978). Conceptual issues in models
of school learning. Journal of Curriculum Studies, 10(3), 215–231.
Bloom’s Taxonomy of Learning
Domains
▶ Bloom’s Taxonomy of Learning Objectives
Bloom’s Taxonomy of Learning
Objectives
AYTAC GOGUS
Center for Individual and Academic Development,
Sabanci University, Istanbul, Turkey
Synonyms
Bloom’s domain; Bloom’s taxonomy of learning
domains; Classification of levels of intellectual behavior
in learning; The classification of educational objectives;
The taxonomy of educational objectives
Definition
Taxonomy means a scientific process of classifying
things and arranging them into groups. Learning objec-
tives are statements of what a learner is expected to
know, understand, and/or be able to demonstrate
after completion of a process of learning. In 1956,
Bloom’s Taxonomy of Learning Objectives B 469
Benjamin S. Bloom (1913–1999) and a group of edu-
cational psychologists developed a hierarchy of educa-
tional objectives, which is generally referred to as B
Bloom’s Taxonomy, and which attempts to identify six
levels within the cognitive domain, from the simplest to
the most complex behavior, which includes knowledge,
comprehension, application, analysis, synthesis, and
evaluation. Bloom’s Taxonomy refers to a classification
of the different learning objectives that educators set
for learners.
Theoretical Background
Following the 1948 Convention of the American Psy-
chological Association, Benjamin S. Bloom took a lead
in formulating a classification of educational activities.
The three domains of educational activities were iden-
tified as the cognitive domain, affective domain, and
psychomotor domain. The cognitive domain represents
the intellectual skills and knowledge processing. The
affective domain represents objectives that are
concerned with attitudes and feelings. The objectives
in the psychomotor domain concern what students
might do physically (Anderson and Krathwohl 2001;
Bloom 1956; Krathwohl et al. 1964). The three domains
can be represented as below:
● Cognitive Domain: Acquisition of knowledge and
intellectual skills (Knowledge)
● Affective Domain: Integration of beliefs and ideas
(Attitude)
● Psychomotor Domain: Acquisition of manual and
physical skills (Skills)
Bloom’s Taxonomy refers to six levels, sub-domains
within the cognitive domain, which are knowledge, com-
prehension, application, analysis, synthesis, and evalua-
tion. The six levels are classified hierarchically from the
simplest action to the high-order thinking actions (see
Table 1). The two subdivisions of cognitive domain are
lower cognitive (knowledge and comprehension) and
higher cognitive domains (application, analysis, syn-
thesis, and evaluation). Also, the last three levels of
higher cognitive domain (analysis, synthesis, and eval-
uation) concentrate on critical-thinking skills. Table 1
outlines these levels and the associated sample verbs
used to write learning objectives and learning outcomes
(Bloom 1956; Clark 2009).
Bloom’s Taxonomy uses observed student behavior
to infer the level of cognitive achievement. Bloom’s
470 B Bloom’s Taxonomy of Learning Objectives

Bloom’s Taxonomy of Learning Objectives. Table 1 Bloom’s taxonomy of learning objectives for the cognitive domain
Levels Sample verbs
Lower 1. Knowledge: Deals primarily with the ability Define, list, identify, show, describe, label, outline,
cognitive to memorize and recall specific facts. recall, state, match, outline, select, recognize, and
reproduce.
2. Comprehension: Involves the ability to Comprehend, convert, contrast, distinguish,
interpret, and demonstrate students’ basic defend, infer, explain, extend, generalize,
understanding of ideas. interpret, predict, translate, summarize.
Higher 3. Application: Involves the ability to apply Apply, solve, change, compute, construct,
cognitive concepts and principles to novel practical demonstrate, discover, manipulate, modify,
situations. operate, predict, prepare, produce, use.
Critical 4. Analysis: Involves the ability to analyze Analyze, break down, compare, contrast, diagram,
thinking concepts and separate concepts or principles deconstruct, differentiate, infer, separate, select,
into components. arrange, discriminate.
5. Synthesis: Involves the ability to blend Integrate, categorize, combine, compile, design,
elements and parts to form a whole. modify, device, compose, rearrange, organize,
generate, create, adopt, revise.
6. Evaluation: Involves the ability to make Decide, defend, appraise, interpret, justify,
judgments of the value of a work. summarize, judge, convince, rank, evaluate,
critique, conclude, criticize, explain.

general approach is called mastery learning that offers
the basis for the competency-based education model
(Bloom 1956). Mastery learning proposes that all stu-
dents can learn when appropriate learning conditions
such as enough time and appropriate instruction are
supplied in the classroom (Bloom 1956). According to
Bloom (1956), enough time and appropriate instruc-
tion are determined by the student’s cognitive entry
behaviors. In addition, the availability of prior knowl-
edge is a crucial factor that influences the learning
process. Bloom’s Taxonomy as a hierarchical frame-
work is used by educators to discuss comparisons in
student achievement, improve standardized testing,
and assess the level of learning (Athanassiou et al.
2003). According to Krathwohl (2002), Bloom saw the
Bloom’s Taxonomy as more than a measurement
tool and believed that the Bloom’s Taxonomy could
serve as a
● Common language about learning goals to facilitate
communication across persons, subject matter, and
grade levels
● Basis for determining for a particular course or
curriculum the specific meaning of broad educa-
tional goals, such as those found in the currently
prevalent national, state, and local standards
● Means for determining the congruence of educa-
tional objectives, activities, and assessments in
a unit, course, or curriculum
● Panorama of the range of educational possibilities
against which the limited breadth and depth of any
particular educational course or curriculum could
be contrasted (p. 212)
Important Scientific Research and
Open Questions
Benjamin S. Bloom, as the editor, and four others, Max
D. Engelhart, Edward J. Furst, Walker H. Hill, and David
R. Krathwohl, comprised a Committee of College and
University Examiners, who undertook the analysis of the
cognitive domain. Taxonomy of Educational Objectives:
The Classification of Educational Objectives Handbook I:
Cognitive Domain, was published in 1956. David R.
Krathwohl, Benjamin S. Bloom, and Bertram B. Masia
were coauthors of Taxonomy of Educational Objectives:
The Classification of Educational Objectives Handbook
II: Affective Domain, published in 1964. The committee
members did not write a third handbook on the
psychomotor domain. Although no taxonomy of
the psychomotor domain was compiled by Bloom
and his coworkers, several competing taxonomies for
 Bloom’s Taxonomy of Learning Objectives B 471

the psychomotor domain have been created over the Anderson and Krathwohl (2001) defined the new
years (Clark 2009). terms and subcategories of the cognitive dimension in
Bloom’s Taxonomy is a scheme for classifying the revised taxonomy as: B
educational goals, objectives, and standards. According
to Krathwohl (2002), Bloom’s Taxonomy provides an 1. Remembering – Retrieving relevant knowledge from
organizational structure and a common meaning to long-term memory
learning objectives classified in one of its categories, (a) Recognizing
thereby enhancing communication. The six levels of (b) Recalling
Bloom’s Taxonomy were arranged in a cumulative hier- 2. Understanding – Determining the meaning of
archical framework, that is, achievement of the follow- instructional messages, including oral, written,
ing, more complex skill or ability required achievement and graphic communication
of the prior one (Krathwohl 2002). Lorin W. Anderson (a) Interpreting
and David R. Krathwohl revisited the cognitive domain (b) Exemplifying
in the learning taxonomy in order to reflect a more (c) Classifying
active form of thinking and made some changes such as (d) Summarizing
changing the names in the six categories from noun to (e) Inferring
verb forms, and slightly rearranging them (Anderson (f) Comparing
and Krathwohl 2001). In contrast to the single dimen- (g) Explaining
sion of the original taxonomy, the revised framework is 3. Applying – Carrying out or using a procedure in
two-dimensional. The two dimensions are cognitive a given situation
process and knowledge (Anderson and Krathwohl (a) Executing
2001). The cognitive process dimension contains six (b) Implementing
categories from cognitively simple to cognitively com- 4. Analyzing – Breaking material into its constituent
plex: remember, understand, apply, analyze, evaluate, parts and detecting how the parts relate to one
and create. The knowledge dimension contains four another and to an overall structure or purpose
categories from concrete to abstract: factual, concep- (a) Differentiating
tual, procedural, and metacognitive (Anderson and (b) Organizing
Krathwohl 2001). In the revised taxonomy, the cogni- (c) Attributing
tive process dimension has six levels that are arranged 5. Evaluating – Making judgments based on criteria
in a hierarchical structure, but not as rigidly as in the and standards
original taxonomy (Krathwohl 2002). In combination, (a) Checking
the knowledge and cognitive process dimensions form (b) Critiquing
a very useful table, the taxonomy table, as depicted in 6. Creating – Putting elements together to form
Table 2. Table 2 also shows the comparison of the original a novel, coherent whole or make an original
taxonomy with the revised taxonomy for cognitive product
domain (Anderson and Krathwohl 2001; Clark 2009). (a) Generating
Krathwohl (2002) emphasizes the benefits of the (b) Planning
taxonomy table as below: (c) Producing (Krathwohl 2002, p. 215)
" Using the table to classify objectives, activities, and Anderson and Krathwohl (2001) defined the new
assessments provides a clear, concise, visual represen- terms and subcategories of the knowledge dimension in
tation of a particular course or unit. Once completed, the revised taxonomy as:
the entries in the Taxonomy Table can be used to
examine relative emphasis, curriculum alignment, and 1. Factual Knowledge – The basic elements that stu-
missed educational opportunities. Based on this exam- dents must know to be acquainted with a discipline
ination, teachers can decide where and how to or solve problems in it
improve the planning of curriculum and the delivery (a) Knowledge of terminology
of instruction (Krathwohl 2002, p. 218). (b) Knowledge of specific details and elements
472 B Bloom’s Taxonomy of Learning Objectives

Bloom’s Taxonomy of Learning Objectives. Table 2 The comparison of the original taxonomy by the revised taxonomy
for cognitive domain and the taxonomy table (Anderson and Krathwohl 2001; Clark 2009)

Original Revised The Knowledge Dimension

Taxonomy Taxonomy
Factual Conceptual Procedural Meta-cognitive

Knowledge Remembering

Comprehension Understanding

Application Applying

Analysis Analyzing

Synthesis Evaluating

Knowledge Remembering

2. Conceptual Knowledge – The interrelationships
among the basic elements within a larger structure
that enable them to function together
(a) Knowledge of classifications and categories
(b) Knowledge of principles and generalizations
(c) Knowledge of theories, models, and structures
3. Procedural Knowledge – How to do something;
methods of inquiry, and criteria for using skills,
algorithms, techniques, and methods
(a) Knowledge of subject-specific skills and
algorithms
(b) Knowledge of subject-specific techniques and
methods
(c) Knowledge of criteria for determining when to
use appropriate procedures
4. Metacognitive Knowledge – Knowledge of cognition
in general as well as awareness and knowledge of
one’s own cognition
(a) Strategic knowledge
(b) Knowledge about cognitive tasks, including ap-
propriate contextual and conditional knowledge
(c) Self-knowledge (Krathwohl 2002, p. 214)
Bloom’s Taxonomy has been used internationally
within the evaluation community and used widely for
policy purposes, sparingly in schools of education,
largely in support of curriculum development and
teaching strategies, and not at all by practicing teachers
in K-12 learning environment (Athanassiou et al.
2003). In college teaching, Bloom’s Taxonomy
has been used as a framework for discussing the dimen-
sions of effective teaching and effective design, imple-
mentation of international experience-based learning
and assessment of higher order cognitive skills.
Athanassiou et al. (2003) investigated the use of Bloom’s
ideas to support student-centered classroom in college
level and conducted a research study to measure the
effect of using Bloom’s Taxonomy as a feedback mecha-
nism in an effort to build students’ critical-thinking
skills. Athanassiou et al. (2003) summarized the use of
Bloom’s Taxonomy in management education:
" A review of the management-education literature
indicates a growing awareness of the taxonomy’s
potential usefulness and richness among college and

university-level educators in curriculum design,
student assessment, and instruction evaluation. In
curriculum discussions, the taxonomy serves well as
a common language to describe increasing levels of
cognitive sophistication within the competency-based
curriculum. These discussions include both the curric-
ulum itself and student mastery of it.. . . Student mas-
tery in these learning settings can involve the more
traditional, faculty-centered assessment of student-
demonstrated learning; student self-assessment . . . or
possibly both. (p. 537)
In addition, Athanassiou et al. (2003) state that the
revised taxonomy focuses more closely on “planning
curriculum, instruction, assessment, and the alignment
of these three” (Anderson and Krathwohl 2001, p. 263)
since the revised taxonomy “eliminates synthesis as
a dimension and adds the dimension of creation beyond
that of evaluation” (p. 537). Athanassiou et al. (2003)
emphasize that the use of the Bloom’s Taxonomy
empower students as self-responsible learners in the
classroom.
Cross-References
▶ Learning Objectives
▶ Mastery Learning
▶ Outcomes of Learning
References
Anderson, L. W., & Krathwohl, D. R. (2001). A taxonomy for learn-
ing, teaching, and assessing: a revision of Bloom’s taxonomy of
educational objectives. In P. W. Airasian, K. A. Cruikshank, R. E.
Mayer, P. R. Pintrich, J. Raths, & M. C. Wittrock (Eds.), A
taxonomy for learning, teaching, and assessing: a revision of
Bloom’s taxonomy of educational objectives. New York: Longman.
Athanassiou, N., McNett, J. M., & Harvey, C. (2003). Critical thinking
in the management classroom: Bloom’s taxonomy as a learning
tool. Journal of Management Education, 27(5), 533–555.
Bloom, B. S. (1956). Taxonomy of educational objectives: the
classification of educational goals; Handbook I: Cognitive
domain. In M. D. Engelhart, E. J. Furst, W. H. Hill, & D. R.
Krathwohl (Eds.), Taxonomy of educational objectives: the classi-
fication of educational goals; Handbook I: Cognitive domain. New
York: David McKay.
Clark, D. (2009). Bloom’s taxonomy of learning domains. Retrieved
from the Web on Dec 1, 2009: http://www.nwlink.com/
~Donclark/hrd/bloom.html.
Krathwohl, D. R. (2002). A revision of bloom’s taxonomy: an over-
view. Theory into Practice, 41(4), 212–218.
Krathwohl, D. R., Bloom, B. S., & Masia, B. B. (1964). Taxonomy of
educational objectives: the classification of educational goals
(Affective domain, Vol. Handbook II). New York: David McKay.
Bootstrapping: How Not to Learn B 473
Bootstrapping: How Not to
Learn B
LANCE J. RIPS
Psychology Department, Northwestern University,
Evanston, IL, USA
Definition
(Quinean) Bootstrapping is a proposed method for
learning a new conceptual system from an old and
incommensurable one. The bootstrapping process
includes the following characteristics (Carey 2009,
p. 418):
1. Relations among symbols [in the new system] are
learned directly, in terms of each other.
2. Symbols are initially at most only partly interpreted
in terms of antecedently available concepts.
3. Symbols serve as placeholders.
4. Modeling processes – analogy, inductive inference,
thought experiments, limiting case analysis,
abduction – are used to provide conceptual
underpinnings for the placeholders.
5. These modeling processes combine and integrate
separate representations from distinct domain-
specific conceptual systems.
6. These processes create explicit representations of
knowledge previously embodied in constraints on
the computations defined over symbols in one or
more of the systems being integrated.
Theoretical Background
Learning requires more than adding new information
to memory. It often requires adjusting old information
in light of the new. New facts may conflict, directly or
indirectly, with old ones. People may be forced to
modify old facts, delete them entirely, modify the new
facts, or refuse to believe them. They may find it
unclear which of these strategies or combinations of
them is best. Learning is belief change, and belief
change is difficult, for both computational and theo-
retical reasons (see, e.g., Rott 2001). Similarly, learning
often entails extrapolating – applying information
beyond the range of observed cases. But people can
extrapolate the same information in many incompati-
ble ways, and by hypothesis, nothing in the data tells
474 B Bootstrapping: How Not to Learn
them which is correct. By contrast, not learning
is a simple matter. In my own case, I have effortlessly
not learned many individual facts, such as whether
there are infinitely many twin prime numbers
or whether life exists in southern Indiana, to cite
just two.
The difficulties just mentioned apply to learning
single facts, but most education research is concerned
with how students learn entire systems or domains of
knowledge, such as the rational numbers, Newtonian
mechanics, computer programming, French history, or
macroeconomics. Learning these domains poses
extraordinary problems, not only because they consist
of many individual facts, but also because of their
conceptual distance from students’ prior knowledge.
Students who have mastered the natural numbers
(0, 1, 2, . . .) understand that for each such number,
exactly one number immediately follows. But this prin-
ciple, which helps define the natural numbers, does not
apply to the rationals, leaving a conceptual gap between
these number systems. Moreover, because some prin-
ciples from the old system carry over to the new,
whereas others do not, interference between the sys-
tems is likely (Hartnett and Gelman 1998). Develop-
mental psychologists have shown that children learn
even entry-level domains, such as the natural numbers,
against a background of assumptions that are often
incorrect of the new information. This makes learning
new domains seem like a hopeless endeavor.
If a student already possesses the elementary con-
cepts of a domain, then learning the domain may call
for rearranging these elements, promoting some ele-
ments and demoting others. Depending on the
entrenchment and incoherence of the original assem-
bly, learning may be simple or vastly difficult. Learning
Newtonian physics, for example, may require students
to reorganize a relatively unstructured set of concepts,
so that some concepts (e.g., the geometry of forces)
come to have a dominant position, whereas others
(e.g., actions of an agent) become less relevant. The
difficulty that students have in understanding mechan-
ics testifies to the difficulties in achieving this reorga-
nization, despite the fact that learning may occur by
ordinary hypothesis testing or strengthening (diSessa
1993). For example, many college students apparently
believe that if a ball rolls off a cliff, it will continue to
have horizontal velocity for a short time but then begin
to fall straight down.
Suppose, though, that prior to learning a new
domain, a student’s knowledge is a richer, more tightly
organized structure. In principle, learning might be no
more difficult than in the case of loosely structured
elements. As long as students can disassemble and
reassemble the elements in the right way, they may be
able to express the to-be-learned facts in their old
vocabulary. By analogy, if a person knows how to
express the fact that the dancer dazzled the financier
and has to learn the new fact that the financier dazzled
the dancer, then he or she already has the concepts
needed to express the new information. What happens,
however, if students cannot formulate facts about the
new domain in the vocabulary of the old? This predic-
ament would seem to make learning, not just hard, but
impossible (as Fodor 1975 has argued). If students
cannot formulate a hypothesis about what might be
true in the new domain, they cannot marshal evidence
to confirm or disconfirm it. If they cannot formulate
a hypothesis – either because their vocabulary is too
impoverished or because their representational power
is too weak – where does the new information come
from? The appropriate analogy might be trying to learn
a sentence in a language one does not know – say,
Tagalog – by looking up the words in a dictionary
written entirely in Tagalog.
Important Scientific Research and
Open Questions
Carey (2009) has advanced the thesis that children’s
learning sometimes takes place by moving from one
state of knowledge to an incommensurable state, and
she has proposed a form of learning, Quinean
bootstrapping, that effects this change. This thesis can
be illustrated with Carey’s proposal about how 2- to
4-year-old children learn the meanings of the terms for
the first few natural numbers: from “one” to the end of
their current list of numerals (e.g., “ten”). Figure 1
diagrams the steps of this bootstrapping process, the
most fully worked out of Carey’s examples. At the
beginning (Step 1), children have two relevant kinds
of mental representations from their innate knowledge.
The first are representations of individual objects, and
the second, representations of sets of objects. The rep-
resentation of individuals is limited to only three
objects; the representation of sets can handle more
than three items within a set, provided the items
move as a group. Step 2 occurs during initial language
 Step 1 (pre-linguistic representation from innate sources): Step 5 (three-knower stage):
Representation g:
Representation a:
object1, object2, object3 in Working Memory (“Parallel Individuation”) “one” “two” “three” “four” “ten”

Representation b: “a” dual marker trial marker

{object1, object2,..., objectk} in Working Memory (“Set based Quantification”)

{object} {object1, object2} {object1, object2, object3} {object1, object2,....}

Step 2 (initial language learning): Step 6 (four-knower stage):

Representation c: (similar to the representation in Step 5)

“One” “two” “three” “ten” in Long-term Memory (“Count List”)

Representation d:
“a” “some”

{object} {object1, object2,....} in Long-term Memory (“Enriched Parallel

Individuation”)

Step 3 (one-knower stage): Step 7 (pre-bootstrap stage):

Representation e: Representation i:
“one” “two” “three” “ten” “one” “two” “three” “four” “ten”

“a” “some” “a” dual marker trial marker

{object} {object1, object2,....} in Long-term Memory {object} {object1, object2} {object1, object2, object3} {object1, object2,....}

Step 4 (two-knower stage): Step 8 (post-bootstrap, Cardinal-Principle-knower stage):

Representation f:
“one” “two” “three” “ten” Representation j:
“one” “two” “three” “four” “ten”

“a” dual marker

{object} {object1, object2} {object1, object2,....} in Long-term Memory {object} Cardinality(n + 1) = Cardinality(n) + 1

Bootstrapping: How Not to Learn. Fig. 1 Steps in children’s learning of the meanings of “one” through “ten,” according to the bootstrapping proposal of Carey (2009)
Bootstrapping: How Not to Learn
B
475

B
476 B Bootstrapping: How Not to Learn
learning at around age two, and it puts two
more representations into play. One is the memorized
list of number words, in order from “one” to, for
example, “ten.” The other is a mapping between the
word “a” (as in “a book”) and a symbol for an individ-
ual. Similarly, there is a mapping between the word
“some” and the symbol for a set. This is children’s
representation of the singular/plural distinction, and
it exists in long-term memory as the meaning of “a”
and “some.”
In the next few steps (Steps 3–6 in Fig. 1), children
successively learn the meanings of “one” through
“three” or “four” by connecting these words with men-
tal representations of sets containing the corresponding
number of objects. First, the children think that “one”
means whatever “a” means and that the rest of the
number words mean whatever “some” means. So
“one” gets connected to {object}, and all the rest of
the number words get connected to the symbol for an
arbitrary-sized set. Over a period of about a year, they
learn to differentiate this set. They connect “two” to
{object1, object2}. But “three,” “four,” . . ., “ten” are still
associated with an arbitrary set of more than two ele-
ments. Then they learn “three” and later “four” in the
same incremental way. Carey believes this process
accompanies learning the quantifier system in lan-
guage; so in languages that have a dual marker denoting
pairs of objects, children learn the dual marker and
then learn that “two” means the same thing.
Finally, in Steps 7 and 8, children notice the relation
between the sequence of numerals in the count list
(Representation c in Fig. 1) and the sequence of sets
of objects (i.e., {object1}, {object1, object2}, {object1,
object2, object3}, . . .). They learn this relation in enu-
merating objects – for example, counting to “three”
while pointing sequentially to three horses in a picture
book. At long last, then, the children can infer that
advancing by one in the count list is coordinated with
adding one object to a set. They now have a rule that
gives them the meaning of any of the numerals on their
count list: the meaning of the next term on the list is one
plus the set size associated with the previous term.
The learning episode in Fig. 1 extends children’s
ability to enumerate objects in response to verbal
requests. But post-bootstrap children still do not
know many important facts about the positive integers.
For example, they have not learned that the integers go
on infinitely instead of stopping at a largest item,
like the letters in the alphabet, or looping around, like
the days of the week (Rips et al. 2008). All they can do is
produce a set of ten things by counting them out.
Moreover, a paradox exists about claims for radical
conceptual change, including bootstrapping. To the
extent that it is possible to understand the change on
a step-by-step basis, the less the new representations
seem genuinely incommensurable with the old ones
and the less evidence there is for conceptual change.
Imagine pre-bootstrap children who are at Step 7 of
Fig. 1. While enumerating objects, they might notice
the correspondence between saying a new numeral and
adding one element to a set. This is the key learning
process the children have to go through – the bootstrap
itself – which advances them to Step 8. But are the
children not learning the new rule (next numeral in
the count list corresponds to adding one to the set)
through their pre-bootstrap concepts? “Next numeral
in the count list” is already there (back in Step 2). The
notion of “one more element” is not so clearly old, but
Carey explains it by saying that the children can compare
the sets in long-term memory to determine that one set
contains one more than the last. That is, the children can
compare {object1} to {object1, object2} and can com-
pare {object1, object2} to {object1, object2, object3} to
get the add-one-more ordering. Thus, all the elements in
the children’s rule can apparently be pronounced in their
old conceptual vocabulary. That is just what Fodor
(1975) would predict and what Carey (2009) denies:
Children must formulate the hypothesis in their old terms
before they can learn it. In evaluating bootstrapping, one
must distinguish the claim that (a) children cannot learn
the new representation in terms of their old representa-
tions, from (b) they cannot learn the new representation
in any one of the old representations. The latter claim
might be true, while the former is false.
Cross-References
▶ Belief Formation
▶ Cognitive Models of Learning
▶ Conceptual Change
▶ Learning Numerical Symbols
References
Carey, S. (2009). The origin of concepts. Oxford: Oxford University Press.
diSessa, A. A. (1993). Toward an epistemology of physics. Cognition
and Instruction, 10, 105–225.
Fodor, J. A. (1975). The language of thought. New York: Crowell.

Hartnett, P. M., & Gelman, R. (1998). Early understandings of num-
bers: Paths or barriers to the construction of new understand-
ings? Learning and Instruction, 8, 341–374.
Rips, L. J., Bloomfield, A., & Asmuth, J. (2008). From numerical
concepts to concepts of number. Behavioral and Brain Sciences,
31, 623–642.
Rott, H. (2001). Change, choice, and inference: A study of belief revision
and nonmonotonic reasoning. Oxford: Oxford University Press.
Boredom in Learning
JENNIFER J. VOGEL-WALCUTT, JULIAN ABICH, SAE SCHATZ
Institute for Simulation and Training-3100
Technology, University of Central Florida, Orlando,
FL, USA
Synonyms
Apathy; Disinterest; Disengagement; Dissatisfaction;
Ennui; Meaninglessness; Monotony; Purposelessness;
Relentlessness; Under arousal; Weariness
Definition
The construct of boredom can be divided into two
main subsets, trait boredom and state boredom. Trait
boredom, also known as boredom proneness, refers to
an individual’s inclination to experience feelings of bore-
dom regardless of setting or condition. Individuals who
are more boredom prone seem to have lower thresholds
for situations that may induce boredom. Substantial
research has been conducted in this area, finding extro-
version and sensation seeking correlate with this person-
ality trait. Trait boredom is a significant concern because
it can negatively affect individuals in a number of ways
including increased work absenteeism and decreased
attention and motivation to learn.
The definition of state boredom, on the other hand,
has been highly debated with no clear agreement in the
literature to date. Across empirical research, state bore-
dom tends to be operationally defined as the subjective
perception of negative affect, the objective assessment of
low arousal, or a combination of the two. Descriptions
outside these are few but include changes in speech
patterns, association with certain types of body move-
ments, disproportion between challenge and competence,
and feelings of alienation and low assertiveness. State
Boredom in Learning B 477
boredom has most commonly been a focus of research
in the work productivity and instructional design
domains, with these investigations seeking to establish B
recommended guidelines that will support motivation
to work or learn, and decrease feelings of boredom.
Theoretical Background
Over time, the topic of boredom has attracted the atten-
tion of researchers across multiple disciplines, each with
unique goals. Due to these diverse contributions, bore-
dom has been conceptualized amorphously, its descrip-
tion influenced by the primary background of the
researcher, the goal of the study, or the decade in
which the research was conducted. Initial definitions of
boredom were typically constructed in response to study
goals and current environmental needs. For example, in
the 1920s, researchers concentrated on improving effi-
ciency for factory output, thus investigating boredom to
better understand the impact of monotonous tasks on
work productivity. This led researchers to define bore-
dom as an experience of monotony. The concept became
more inclusive throughout the 1950s, 1960s, and early
1970s and began to incorporate affective responses and
arousal changes. Investigators generally adopted the sug-
gestion that arousal may play a significant role in affect-
ing feelings of boredom, but whether high or low arousal
correlated with feelings of boredom remained unclear.
Unfortunately, empirical work during these decades was
also influenced by unusual experimental methodologies
and was limited in sample size and sample characteris-
tics. Therefore, the validity and reliability of these stud-
ies, and consequently developed measures, remain in
question. Toward the end of the 1970s, researchers
began to incorporate the study of environmentally
influenced feelings of boredom (later described as state
boredom) and began focusing on factors rooted in an
individual’s personality (later defined as trait boredom).
Considering boredom as a personality trait, or the idea
that feelings of boredom could be experienced indepen-
dent of the current environment, became the focus of
research during this time.
Trait Boredom
Studying different personality traits and their associa-
tions with boredom was a growing area of research
toward the end of the 1970s. For example, some inves-
tigators found differences between extroverts and
introverts. This idea gained significant support from
478 B Boredom in Learning
the literature (Vodanovich 2003) and, consequently,
many assessment tools were developed (e.g., Boredom
Proneness Scale, Boredom Susceptibility Scale, Job
Boredom Scale, Leisure Boredom Scale) to identify
those individuals.
Personality characteristics, or the inherent traits of
the individual, influence how they interpret the world.
Specifically, those with high levels of boredom prone-
ness focus on or view life as largely boring. They easily
experience boredom, even in generally stimulating sit-
uations. Jurich (2004) suggests that boredom prone-
ness is an individual trait that influences a person’s
perception of a given situation. Boredom proneness
varies on a continuum; individuals who score higher
will more likely find situations boring, compared to
those who rate lower on the continuum.
According to Hunter and Csikszentmihalyi (2003),
adolescents generally fall into two groups: those with
chronic interest (i.e., low boredom proneness) and
those who experience widespread boredom (i.e., high
boredom proneness). The first group experiences
a high degree of durable positive self-concept, while
those with high levels of trait boredom demonstrate
more negativity and an unstable self-assessment. Fur-
ther, boredom proneness has been correlated with pro-
crastination. Similarly, positive relationships between
shyness and boredom have also been found suggesting
that those who are less likely to engage in a social
situation or participate in class are also plagued by
increased feelings of boredom.
State Boredom
Research on the construct of boredom, toward the
1980s, was expanded to include definitions of negative
affect, low arousal, or a combination of the two. Affect
is an automatic emotional response to stimuli that may
be influenced by cognition or prior experience. Arousal
is an individual’s physiological response to stimuli. As
investigators refined the scientific definition of state
boredom, the apparatus used to measure it also
improved. Researchers began to rely less upon use of
subjective self-assessments and instead began incorpo-
rating objective neurophysiological measures of bore-
dom to better assess the construct.
Such neurophysiological studies support the view-
point that boredom, like other emotions, exists on
a two-dimensional affective scale (Russell 1980; Chanel
et al. 2008). For instance, Russell (1980), along with
other contemporary researchers, posits that boredom
involves both negative emotions as well as negative
arousal. Based on his empirical research, Russell devel-
oped the Circumplex Model, within which each emo-
tion is placed in one of four quadrants on a Cartesian
plane. Boredom is positioned in the same quadrant as
other theoretically under-arousing, unpleasant emo-
tions such as depression, sadness, and gloominess.
Much contemporary research (D’Mello, Taylor and
Graesser 2007; Posner et al. 2009) is based upon this
theoretical definition. Thus, many researchers now
suggest that (low) arousal and subjective (negative)
emotion combined define state boredom. However,
other researchers still maintain that boredom is better
defined by only affect or arousal, and consequently,
a fully unified theory of state boredom has not yet
been adopted by the research community.
Learning and Boredom
Learning environments are a common place in which
feelings of boredom arise. Researchers have found that
boredom increases when the perceived skill of the indi-
vidual exceeds the level of task challenge, when little
choice is provided to the learner during the learning
process, and when the learning style does not match the
learners’ method for information processing. As
a result, boredom can reduce learning by up to 25%
(Craig et al. 2004), decrease students’ achievement
scores, and reduce their grade point averages.
Boredom can be substantially detrimental to the
learning process, with as much as two-thirds of high-
school students experiencing it, and 56% of students
spending time off the tasks while in the classroom
(Craig et al. 2004). In short, boredom disengages
many learners from educational activities, and it seri-
ously decreases learners’ abilities to acquire knowledge.
Students who have feelings of boredom often lose
interest in the material being covered, disrupt the
pace of the class, and interrupt other learners. Ongoing
research may be able to provide insight on this matter,
such as identifying the conditions under which bore-
dom is most likely to occur, but today’s investigations
have yet to provide well-defined, measurable instruc-
tional assessments and matching strategies to improve
state boredom. In time, more valid and reliable assess-
ment tools will likely lead to more effective boredom
mitigation strategies. Having the ability to correctly
diagnose those affected by boredom will allow the

learning environment and manner in which informa-
tion is presented to be catered to those experiencing
feelings of either state or trait boredom.
Important Scientific Research and
Open Questions
The overarching problem in the literature today is the
lack of an agreed-upon definition of the construct of
state boredom. As a result, suitable assessment tools
and interventions are also lacking. A review of research
across disciplines and domains may aid researchers in
developing a more solidified and empirically supported
definition. Once a research-supported definition is
established, improved diagnosis and recommendations
for interventions should follow.
A second, but no less important, area of upcoming
research involves the use of neurophysiological sensors to
more objectively and efficiently detect feelings of bore-
dom. Major advances in technology over the past few
decades have given investigators the ability to measure
physiological changes in real-time and correlate those
changes with observed or reported emotions. As a result,
research involving individuals’ physiological responses to
stimuli can be combined with subjective psychological
interpretations of their emotions to better measure feel-
ings of boredom. Empirical studies investigating the
neuro-physiological basis of boredom have generally
supported the two-dimensional viewpoint; however,
continued research to confirm this finding is warranted.
Cross-References
▶ ARCS-Model of Motivation
▶ Assessment of Academic Motivation
▶ Interests and Learning
▶ Learner Characteristics
▶ Motivation to Learn
▶ Personality and Learning
References
Chanel, G., Rebetez, C., Betrancourt, M., & Pun, T. (2008). Boredom,
engagement and anxiety as indicators for adaptation to difficulty
in games. Proceedings of the Mindtrek Conference, October 7–9,
2008, Tampere, 13–17.
Craig, S. D., Graesser, A. C., Sullins, J., & Gholson, B. (2004). Affect and
learning: an exploratory look into the role of affect in learning with
AutoTutor. Journal of Educational Media, 29(3), 241–250.
Hunter, J. P., & Csikszentmihalyi, M. (2003). The positive psychology of
interested adolescents. Journal of Youth and Adolescence, 32(1),
27–35.
Bottom-Up Learning and Top-Down Learning B 479
Jurich, D. (2004). Attribution of boredom: attentional factors and
boredom proneness. (Doctoral Dissertation, New School
University, 2004).
Russell, J. A. (1980). A circumplex model of affect. Journal of Person- B
ality and Social Psychology, 39, 1161–1178.
Vodanovich, S. J. (2003). Psychometric measures of boredom:
a review of the literature. The Journal of Psychology, 137(6),
569–595.
Bottom-Up Learning and
Top-Down Learning
RON SUN
Cognitive Science Department, Rensselaer Polytechnic
Institute, Troy, NY, USA
Synonyms
Bottom-up learning is also known as implicit-to-
explicit explicitation; Top-down learning is also
known as explicit-to-implicit implicitation (or
assimilation)
Definition
Bottom-up learning refers to learning implicit knowl-
edge first and then learning explicit knowledge on that
basis (i.e., through “extracting” implicit knowledge).
Top-down learning refers to learning explicit
knowledge first and then learning implicit knowledge
on that basis (i.e., assimilating explicit knowledge into
an implicit form).
Theoretical Background
The idea of two systems in the human mind (implicit
and explicit) that are rather separate for representing or
learning different types of knowledge or skills can be
traced back to early work in psychology, for example,
on classical and instrumental conditioning (without
subjective conscious awareness) and so on.
In particular, Arthur Reber demonstrated very early
on (in the 1970s and 1980s) that subjects could mem-
orize letter strings that followed certain patterns and
after that discriminated valid from invalid novel strings
without conscious awareness of the basis for their
judgments. Similar work has been carried out by
Donald Broadbent and others in other experimental
settings. Some general claims made about such implicit
480 B Bottom-Up Learning and Top-Down Learning
learning, which have been controversial, were that
implicit learning can occur without conscious aware-
ness, that such learning can occur automatically (i.e.,
without involving limited cognitive resources), and
that such learning involves abstracting the underlying
structure of the stimuli. Empirical (psychological)
results show that while the notion of implicit learning
is solid these specific claims often need qualifications to
take into account the complex interactions between
implicit and explicit processes (Sun et al. 2005).
In the late 1990s, on the basis of the large body of
work on implicit learning, Ron Sun and others (see,
e.g., Sun et al. 2001, 2005) focused on the very process
of the interaction between implicit and explicit learn-
ing, and proposed the notion that skill learning can go
from implicit learning and implicit knowledge to
explicit learning and explicit knowledge, which was
termed bottom-up learning or implicit-to-explicit
explicitation (see also, e.g., Stanley et al. 1989). This
style of learning is distinct from the more commonly
recognized way of skill learning going from explicit
knowledge to implicit knowledge, that is, top-down
learning or, as it has been termed before, assimilation
or implicitation (or even “proceduralization”, which,
unfortunately, confounds the issue of implicit versus
explicit learning with the issue of procedure versus
declarative learning).
To account for the detailed (“computational”) process
of bottom-up learning, a novel comprehensive model of
the human mind (known as a “cognitive architecture” in
the cognitive science parlance) has been developed since
then (named CLARION; see Sun 2002), which provides
detailed mechanistic explanations of the processes of
bottom-up learning and top-down learning (based,
in part, on machine learning algorithms for reinforce-
ment learning and rule learning that have been devel-
oped rather recently; see Sun 2002 for details).
This model has since been used to account for a large
variety of empirical data related to implicit learning,
bottom-up learning, and top-down learning.
Important Scientific Research and
Open Questions
A major claim from the research on bottom-up and
top-down learning is that learning can occur in either
way (or both): (1) learning can occur through trial-
and-error implicitly, without explicit knowledge to
begin with; implicit skills may be acquired before
explicit knowledge emerges; explicit knowledge may
in fact be learned through the mediation of already
acquired implicit knowledge (i.e., through extracting
implicit knowledge in a sense); and (2) learning can
also occur through acquiring explicit knowledge first
and then, with practice, assimilating explicit knowledge
into implicit forms.
The significance of this distinction above lies in the
fact that bottom-up learning has been very much
a neglected topic (even its very existence was ignored
in cognitive psychology for a long time until recently;
Sun et al. 2001), while top-down learning has been
over-emphasized. Given the culturally created systems
of schooling, apprenticeship, and other forms of
guided (or instructional) learning, top-down learning
is quite prevalent in society. However, bottom-up
learning is more fundamental. It is more fundamental
in two senses: the ontological sense and the ontogenetic
sense.
Ontologically, explicit conceptual knowledge needs
to be obtained in the first place before it can be
imparted to people (e.g., to enable top-down learning).
Therefore, bottom-up learning, which creates new
explicit knowledge, is more fundamental. Only after
bottom-up learning (and/or other types of learning)
created explicit conceptual knowledge, can top-down
learning be possible. Ontogenetically, there seem to
be some empirical indications that children learn
sensory-motor skills (as well as knowledge concerning
concepts) implicitly first, and then acquire explicit
knowledge on that basis (see Sun et al. 2001 for
a review of the relevant psychological literature).
Therefore, bottom-up learning is also more important
ontogenetically (i.e., developmentally).
Of course, instead of bottom-up learning, it is pos-
sible that one can learn explicit knowledge directly. One
reason why bottom-up learning has been emphasized
here is because it has not been sufficiently emphasized
in the past. Furthermore, a cognitive advantage that
comes with bottom-up learning, as opposed to directly
learning explicit knowledge, is the reduction of “com-
putational” cost of learning. For one thing, employing
this two-stage approach may be a more efficient way of
learning explicit knowledge (in a “computational”
sense), because, guided by implicit knowledge, the
search space for explicit knowledge is narrowed down
and an online, incremental search can then be more
easily performed (as has been demonstrated through

modeling and simulation using CLARION; Sun 2002).
This fact may, in part, explain why evolution has cho-
sen this approach (Sun 2002). There have been human
data reported in the literature that indicate that it is
likely that humans do engage in bottom-up learning
(see, e.g., Stanley et al. 1989; Sun et al. 2001, 2005; Sallas
et al. 2007). Therefore, bottom-up learning is believed
to be cognitively realistic.
This research also has significant implications for
educational practice (Sun et al. 2007). Most educa-
tional settings focus on teaching conceptual, explicit
knowledge rather than providing opportunities for
gaining substantial experiential, that is, mostly
implicit, knowledge first and then bottom-up learning
on that basis that goes from implicit knowledge to
explicit knowledge. While this may be advantageous
for some subject areas, others may often involve com-
plex skills and knowledge (e.g., features of a convoluted
system or some ill-structured categories) that are better
learned (at least initially) through extensive experience.
In general, repeated practice, memorization of examples,
laboratory exploration, and so on may help to promote
implicit (and/or procedural) learning and consequently
bottom-up learning on that basis, while classroom lec-
tures and textbooks often promote learning of explicit,
conceptual knowledge first and top-down learning on
that basis. While we acknowledge the importance of
explicit knowledge, the significance of implicit learn-
ing/knowledge and bottom-up learning processes in
education should not be downplayed either.
Current research in this area investigates the inte-
gration of implicit and explicit knowledge in complex
skill learning through bottom-up and top-down learn-
ing. Psychological experiments with human partici-
pants are being conducted that explore different
ways for, and different effects of, the integration of
implicit and explicit knowledge through bottom-up
and top-down learning (e.g., Sallas et al. 2007).
They also explore different methods of facilitating bot-
tom-up and top-down learning in order to enhance
skill acquisition. In relation to such research, “cognitive
architectures” (e.g., CLARION, as mentioned before;
Sun 2002) are being further developed for capturing the
fine process details of bottom-up and top-down learning.
Through exploring a variety of empirical data, a unified
and comprehensive cognitive architecture may shed bet-
ter light on plausible mechanistic (“computational”) pro-
cesses of bottom-up and top-down learning, as well as the
Bounded Rationality B 481
interaction and integration of implicit and explicit pro-
cesses in general. The contribution of the research lies in
some useful theories that may explain a wide range of B
human data in terms of bottom-up and top-down learn-
ing and in terms of the interaction and integration of the
two types of cognitive processes. The practical implica-
tions of bottom-up versus top-down learning in enhanc-
ing education, learning, and training may also be better
understood as a result.
Cross-References
▶ Human Cognitive Architecture
▶ Implicit Learning
▶ Skill Learning
References
Sallas, B., Mathews, R., Lane, S., & Sun, R. (2007). Developing rich
and quickly accessed knowledge of an artificial grammar. Mem-
ory and Cognition, 35(8), 2118–2133.
Stanley, W., Mathews, R., Buss, R., & Kotler-Cope, S. (1989). Insight
without awareness: On the interaction of verbalization, instruc-
tion and practice in a simulated process control task. Quarterly
Journal of Experimental Psychology, 1215(41A), 553–577.
Sun, R. (2002). Duality of the mind. Mahwah: Lawrence Erlbaum.
Sun, R., Merrill, E., & Peterson, T. (2001). From implicit skills to
explicit knowledge: A bottom-up model of skill learning. Cogni-
tive Science, 25(2), 203–244.
Sun, R., Slusarz, P., & Terry, C. (2005). The interaction of the explicit
and the implicit in skill learning: A dual-process approach.
Psychological Review, 112(1), 159–192.
Sun, R., Mathews, R., & Lane, S. (2007). Implicit and explicit processes in
the development of cognitive skills: A theoretical interpretation with
some practical implications for science education. In E. Vargios
(Ed.), Educational psychology research focus (pp. 1–26).
Hauppauge: Nova.
Bottom-Up Learning Is Also
Known as Implicit-To-Explicit
Explicitation
▶ Bottom-Up Learning and Top-Down Learning
Bounded Rationality
The notion that, due to cognitive limitations in knowl-
edge and computational capacity, decision makers
482 B Bounded Rationality and Learning
cannot aim for optimal choices, but have to be content
with good-enough solutions – that is, they have to
“satisfice.”
Bounded Rationality and
Learning
FERNAND GOBET1, PETER C. R. LANE2
1
School of Social Sciences, Centre for the Study of
Expertise, Brunel University, Uxbridge, Middlesex, UK
2
School of Computer Science, University of
Hertfordshire, Hatfield, Hertfordshire, UK
Synonyms
Procedural rationality
Definition
▶ Bounded rationality is a term proposed by Nobel
Prize winner Herbert A. Simon (1916 – 2001) to
emphasize that a ▶ decision maker’s rational choice is
affected by cognitive limitations, in particular limita-
tions in knowledge and limitations in computational
capacity. Limitations in knowledge relate to what
a decision maker knows about their domain; how
long have they been learning, and from which sources?
Limitations in computational capacity refer to the abil-
ity to identify all relevant factors in a problem, work
out their consequences, and come to a conclusion in
a reasonable period of time. Psychological explanations
for the latter include small span of attention, limited
capacity of short-term memory, slow learning rates,
and slow rate of searching problem spaces. All these
limitations are particularly problematic given the lim-
ited time available for most decisions. Learning is an
important way in which the impact of these limitations
can be alleviated, at least to some extent.
Theoretical Background
The seeds of Simon’s conception can be found in
Administrative Behavior (Simon 1947), a book derived
from his PhD thesis, in which he showed that the
type of unbounded or ▶ global rationality assumed
by classic economics and statistical decision theory
could not account for behavior in organizations.
Bounded rationality, and its impact on human deci-
sion-making, guided Simon’s research in many
domains, including cognitive psychology, artificial
intelligence, and the use of information technology in
organizations.
Simon introduced the concept of bounded ratio-
nality in his analysis of organizational behavior (Simon
1947), although he was clear that his conclusions
applied to the social sciences in general, including in
particular economics and psychology. His interest cen-
tered on the following question: Can the kind of ratio-
nality assumed in classic economics – unlimited access
to the relevant information and unlimited resources to
carry out the computations necessary for maximizing
utility – reflect the way executives make decisions in
business, government, and other organizations? Based
on a critical analysis of the extant literature, Simon
concluded that this was not the case, for two main
reasons. First, access to information is strictly limited,
and indeed involves a cost. Thus, lacking full informa-
tion, executives and other decision makers cannot list
all possible alternatives, compute their utility, and
choose the optimal solution, as required by unbounded
rationality. Second, the human cognitive system is
characterized by a number of rather strong limitations.
There is a limit of attention: the amount of information
which can be perceived, read, or simply absorbed at any
one time is restricted by constraints, such as the size of
our visual field, the rate at which we can read, and the
amount of information which can be communicated
verbally. Further limits include the capacity of short-
term memory, which limits the number of ideas we can
hold in mind at any one time, and also the rate with
which information can be stored in long-term memory.
Given these limits, attention can normally focus on
only one thing at a time, and the search for solutions
to problems is carried out essentially serially. Thus, even
assuming unlimited access to the information required
to make a decision, these cognitive limitations make it
impossible for humans to carry out the computations
necessary for making optimal decisions, except in very
simple cases. Rather, executives and humans in general
display bounded rationality, that is, they make decisions
that are rational given the limited information and com-
putational resources available. They do not try to find an
elusive optimal solution, but they satisfice and stop their
search once a good-enough decision has been met. In
this context, ▶ satisficing means that a solution to

a problem is accepted once all criteria for comparing
alternatives are above a given threshold. These criteria
are set a priori but can then be revised as search pro-
gresses and aspiration levels are modified, either down-
ward or upward.
In later writings (e.g., Simon 1997), Simon also used
the terms substantial rationality and ▶ procedural ratio-
nality, which closely correspond to global rationality
and bounded rationality, respectively. Substantial
rationality refers to the type of rationality used in
economics, where the emphasis is not only on maxi-
mizing utility but also on analyzing the situation rather
than the decision maker. With procedural rationality,
the emphasis is reversed: the interest is about the pro-
cesses leading to a decision, and the focus is on the
decision maker rather than on the situation. Thus,
procedural rationality matches the type of rationality
studied in psychology.
Given this rather grim description of human cogni-
tive abilities, where the emphasis seems to lie primarily
on shortcomings, one is legitimately entitled to wonder
how our species was able to produce such developments
as the invention of calculus and the discovery of DNA.
Simon’s answer is that humans (partly) sidestep the
limits imposed by bounded rationality by storing previ-
ous solutions in long-term memory (through learning),
by using powerful heuristics (rules of thumb), and by
combining their forces in efficient and typically hierar-
chical organizations, where decision-making control can
be centralized or decentralized as a function of the
demands of the internal and external environments.
They can also use the environment as an external mem-
ory or even processor, for example, by order of sophis-
tication: a cue, such as a road sign, to remind one to
carry out an action, paper on which to write notes, and
computer memory.
Two major types of decisions can be identified:
programmed decisions and non-programmed decisions.
Programmed decisions are well-specified sequences of
activities, those repetitive decisions that are routine and
well-learned and for which clear-cut procedures have
been developed. Thus, it is not necessary to devise new
solutions where such decisions must be made.
Programmed decisions can be made only with structured
problems, where the elements of the problem and their
relations are well understood by the decision maker. By
contrast, non-programmed decisions cannot make use of
standard procedures, either because similar problems
Bounded Rationality and Learning B 483
have not arisen in the past, or because of their impre-
cise and complex nature. Such problems are called
unstructured, and neither the elements nor their rela- B
tionships are understood by the problem solver. Simon
readily accepts that real-life decisions cannot be neatly
classified into these two categories, which really repre-
sent the two extremes of a continuum. As a corollary,
many problems must be seen as semi-structured, that is,
some elements and relations are understood by the
problem solver, but others are not. Again, we can see
the importance of bounded rationality in this classifi-
cation. Programmed decisions can be made fairly easily
because previous solutions, as well as the cues indicat-
ing their appropriate application, have been stored in
long-term memory; this luxury is not possible with
non-programmed decisions, and then a relatively slow
and serial problem solving behavior must be used.
The thesis put forward by Simon is that bounded
rationality is inevitable due to the limits imposed on
human thinking by limits of attention, learning, and
memory capacity. As seen above, a major way to
alleviate these limits is learning, that is, to use
long-term memory as an extension of short-term
memory. In extreme cases, learning will lead to
high levels of expertise. Rather than storing
a handful of items in short-term memory, one can
store a large amount of information by using chunks
and schemata. Rather than carrying attention to
a single action, one can carry out several actions in
parallel without using much attention, assuming that
each action has been learned to the point that it has
become automatized.
Simon has developed the link between learning and
bounded rationality in the chunking theory, which
assumes that learning involves the acquisition of large
numbers of perceptual ▶ chunks and actions linked to
these chunks. This large reservoir of prior experience
injects flexibility and efficiency into a system that
would otherwise be rigid. Simon estimated that
a typical human ▶ expert will have acquired between
10,000 and 100,000 chunks over a training period of
10 years. This knowledge, together with the presence
of efficient heuristics, reduces the need for search, as
previous searches have essentially been compiled in
long-term memory. Efficient access and immediate
response give rise to the intuitive ease with which
domain experts rapidly diagnose and accurately
respond to problems in their area of expertise.
484 B Bracketing
Important Scientific Research and
Open Questions
The concept of bounded rationality is amply supported
by empirical evidence on probabilistic reasoning, logi-
cal reasoning, and game playing, to cite just a few
domains outside organizational behavior. While fairly
well accepted in organizational science and psychology,
Simon’s ideas are still disputed in economics,
where (neo-)classical theories are still dominant. It
should also be recognized that the idea of rationality,
in the sense of maximizing expected utility, can still be
found in fields such as psychology, most notably
in Anderson’s framework of rational analysis
(Anderson 1990).
Recent research on bounded rationality has
followed several avenues. The extent to which heuris-
tics and algorithms help alleviate the constraints
imposed by humans’ cognitive limitations has received
particular attention (e.g., Gigerenzer 2000).
A counterintuitive result from this line of research is
that simple heuristics sometimes lead to decisions that
are better than optimal decision algorithms, such as
linear regression. Another line of research has led to the
development of formal models – either mathematical
or computational – to account for the way decisions are
made (e.g., Rubinstein 1998).
As noted above, learning is one of the two main ways
humans have expanded their bounded rationality, and
there is also important research on this topic. Two lines
were directly started by Simon himself. The first line, on
expertise, tries to investigate the limits of human
cognition and to understand the training mechanisms
that allow experts to push these limits much further
than novices. Interestingly, while the assumption of full
rationality is at variance with research documenting the
development of expertise – differences in expertise are
meaningless in this framework, because even “novices”
should have unlimited computational resources and
access to information – bounded rationality offers
a natural theoretical account to understand expertise
and its development (Campitelli and Gobet 2010).
The second line has developed computer models of
the way humans acquire knowledge, and has focused
on the acquisition of perceptual chunks. One of
Simon’s key insights is that acquired knowledge
enables humans to take excellent and rapid decisions
by intuition, thus sidestepping the limits of their
limited rationality. The computer model ▶ CHREST
(Gobet et al. 2001) contains mechanisms explaining
how this can happen.
Cross-References
▶ Chunking Mechanisms and Learning
▶ Development of Expertise
▶ Learning in the CHREST Cognitive Architecture
▶ Schema
References
Anderson, J. R. (1990). The adaptive character of thought. Hillsdale:
Lawrence Erlbaum.
Campitelli, G., & Gobet, F. (2010). Herbert Simon’s decision-making
approach: Investigation of cognitive processes in experts. Review
of General Psychology, 14, 354–364.
Gobet, F., Lane, P. C. R., Croker, S., Cheng, P. C-H., Jones, G., Oliver,
I., Pine, J. M. (2001). Chunking mechanisms in human learning.
TRENDS in Cognitive Sciences, 5, 236–243.
Gigerenzer, G. (2000). Adaptive thinking: rationality in the real world.
New York: Oxford University Press.
Rubinstein, A. (1998). Modeling bounded rationality. Cambridge, MA:
MIT Press.
Simon, H. A. (1947). Administrative behavior. New York: Macmillan.
Simon, H. A. (1997). An empirically based microeconomics. Cam-
bridge: Cambridge University Press.
Bracketing
▶ Learning Through the Breach: Language
Socialization
Brain-Based Visual Learning
▶ Visual Communication and Learning
Brainstorming and Learning
AYTAC GOGUS
Center for Individual and Academic Development,
Sabanci University CIAD, Istanbul, Turkey
Synonyms
Consensus learning; Idea generation

Definition
Brainstorming means using the brain to storm
a creative solution for a problem. Brainstorming is
a method of generating ideas and solving problems,
therefore there is a strong connection between brain-
storming productivity and domain learning. Brain-
storming is a group activity in which all the members
of a group suggest ideas and then discuss them as
a brainstorming session. Brainstorming is an approach
to active learning that allows learners to participate in
learning activities, to take the responsibility for their
own learning, and to establish connections between
ideas by analyzing, synthesizing, and evaluating.
Theoretical Background
Alex Osborn (1953, rev. 1957, rev. 1963), an advertising
executive, suggested brainstorming as a method for
improving the creativity of groups. Osborn (1963)
designed a brainstorming technique with the following
four specific brainstorming instructional components
to prevent inappropriate and untimely evaluation of
ideas and stimulate more creative ideas:
1. Criticism is ruled out. Adverse judgment of ideas
must be withheld until later.
2. “Free-wheeling” is welcomed. The wilder the idea, the
better; it is easier to tame down than to think up.
3. Quantity is wanted. The greater the number of
ideas, the more the likelihood of useful ideas
[generated].
4. Combination and improvement are sought. In addi-
tion to contributing ideas of their own, participants
should suggest how ideas of others can be turned
into better ideas; or how two or more ideas can be
joined into still another idea (Osborn 1963, p. 156).
According to Baruah and Paulus (2008), “the brain-
storming technique was first popularized by Alex
Osborn in 1957 as a method to generate new ideas by
asking group members to freely generate as many ideas
as they can on a given topic following four specific
rules: (a) express any ideas that come to mind, (b) do
not criticize any ideas, (c) generate as many ideas as you
can, and (d) combine and improve these ideas to come
up with newer and better ideas.” (p. 523). Osborn
(1963) claimed that “interactive groups (face-to-face
groups that verbally exchange ideas) should perform
better than the nominal groups (same number of indi-
vidual brainstormers who work alone on the same
Brainstorming and Learning B 485
problem) if they follow the above rules in
a brainstorming session.” (Baruah and Paulus 2008,
p. 523). Members of interactive groups, who engage in B
verbal brainstorming, should be able to benefit from the
diverse perspectives and abilities of their group mem-
bers by cognitively stimulating each other to generate
many ideas (Baruah and Paulus 2008; Osborn 1963).
According to Baruah and Paulus (2008), many research
concluded that
" Face-to-face groups actually perform more poorly
when compared to the same number of persons work-
ing alone . . . This productivity loss appears to be due to
the inability to express ideas (production blocking) as
they occur in a group setting. Apprehension about
evaluation of ideas and social loafing may also contrib-
ute to the lowered group performance. Social compar-
ison processes may lead individuals in the group to
develop low standards for performance and to move
their performance in the direction of low performers.
(p. 524)
During group brainstorming, group members
should generate many ideas, think of uncommon
ideas, combine, evaluate and improve ideas, and
avoid from untimely and unappropriate criticism.
Learning may result from the brainstorming process,
as it provides a momentum to engage in constructing
ideas and self-explanations.
Social and cognitive factors that lead to increased
production gains in group brainstorming are; increased
accountability, competition, upward comparison, and
social and cognitive stimulation (Baruah and Paulus
2008; Dugosh and Paulus 2005). Baruah and Paulus
(2008) suggest the use of facilitators that employ vari-
ous social and cognitive procedures to keep the brain-
stormers on task and encourage all participants to
contribute the generation of ideas in groups, although
the efforts of a trained facilitator are costly.
Rossiter and Lilien (1994) present six new princi-
ples emerging from academic and industry researches
on the generation of high-quality creative ideas by
“brainstorming” as stated below:
(a) Brainstorming instructions are essential and
should emphasize, paradoxically, number and not
quality of ideas.
(b) A specific, difficult target should be set for the
number of ideas.
486 B Brainstorming and Learning
(c) Individuals, not groups, should generate the initial
ideas.
(d) Groups should then be used to amalgamate and
refine the ideas.
(e) Individuals should provide the final ratings to
select the best ideas, which will increase commit-
ment to the ideas selected.
(f) The time required for successful brainstorming
should be kept remarkably short (Rossiter and
Lilien 1994, p. 61).
Rossiter and Lilien (1994) state that, brainstorming
can more dependably produce high-quality creative
results by following these six principles. Rossiter and
Lilien (1994) suggest using the I-G-I (Individual-
Group-Individual) procedure for producing, refining,
and evaluating creative ideas by following six steps:
1. Chairperson announces problem and gives brain-
storming instructions to five to seven individuals
seated around a table in the same room. (Rationale:
this “silent groups” format preserves individuality
but introduces a possible social facilitation effect
from the presence of others.)
2. Individuals, with no talking, write down or key into
personal computers as many ideas as they can in the
specified time period, usually 15 min. (Rationale:
immediate recording of ideas helps to remove the
“production blocking” problem whereby mental
rehearsal of initial ideas blocks the production of
further ideas.)
3. Chairperson records individuals’ ideas, in rotation,
one idea per person per rotation, on a group-visible
flip chart or electronic screen. (Rationale: the rota-
tion procedure removes some of the anonymity of
a “talk in any order” group while at the same time
producing a list of ideas that are recorded without
authorship.)
4. Group clarifies and discusses ideas, combining or
refining them as it sees fit. Ideas are taken one at
a time, and each individual is asked for reasons of
agreement or disagreement as well as to make con-
structive suggestions for improvement. (Rationale:
groups are efficient and usually superior for com-
bining and refining ideas.)
5. The revised ideas are then recorded by the chair-
person in a group-visible final list. (Rationale:
memory reliance is again minimized and also
a degree of democratic anonymity is reinstated.)
6. Revised ideas are rated or ranked by individuals
privately, with no discussion. Best idea or ideas cho-
sen by pooled individual votes. (Rationale: demo-
cratic voting increases commitment and pooled
individuals’ judgments usually provide more accu-
rate prediction.) (Rossiter and Lilien 1994, p. 67)
According to Rossiter and Lilien (1994), creative
idea production is fundamental to planning and man-
agement and is vital to the generation of concepts for
new products and for the advertising of new and
established products. Using brainstorming technique
is vital in many areas. A significant correlation between
idea generation productivity and learning from brain-
storming is reported in experimental studies (Rossiter
and Lilien 1994; Brown and Paulus 2002).
Important Scientific Research and
Open Questions
Brown and Paulus (2002) have a study on making
group brainstorming more effective from an associate
memory perspective. Brown and Paulus (2002) state
that much literature on group brainstorming finds it to
be less effective than individual brainstorming; however,
a cognitive perspective suggests that group brainstorming
can be an effective technique for generating creative ideas:
" A cognitive perspective points to methods that can be
used so that group exchange of ideas enhances idea
generation. Groups of individuals with diverse sets of
knowledge are most likely to benefit from the social
exchange of ideas. Although face-to-face interaction is
seen as a natural modality for group interaction, using
writing or computers can enhance the exchange of
ideas. The interaction should be structured to ensure
careful attention to the shared ideas. Alternating
between individual and group ideation is helpful
because it allows for careful reflection on and processing
of shared ideas. (Brown and Paulus 2002, p. 211)
Brown and Paulus (2002) emphasize that computer
simulations of an associative memory model of idea
generation in groups suggest that groups have the
potential to generate ideas which are less likely to be
generated through individual brainstorming alone.
Brown and Paulus (2002) conclude that exchanging
ideas and retrieving information from one’s long-
term conceptual memory by means of writing or
using semantic network programs, alternating solitary

and group brainstorming, and using heterogeneous
groups appear to be useful approaches for enhancing
group brainstorming.
Empirical findings from brainstorming research
suggest that loss of coordination and motivation in
a team can hamper the effectiveness of brainstorming,
and also brainstorming in interdisciplinary team and
social interaction may not always facilitate the genera-
tion of creative ideas (McGlynn et al. 2004). Moreover,
McGlynn et al. (2004) argue that, the usefulness of
brainstorming as a process facilitating creative ideas is
limited to the earlier stages of task performance in
groups. McGlynn et al. (2004) state that “group brain-
storming is usually considered a task of divergent
thinking, and the ideas produced in most research on
brainstorming are counted and scored for creativity but
put to no further use” (p. 75). McGlynn et al. (2004)
have a study on brainstorming in which they embed it
in a rule induction task and suggest that brainstorming
initially requires divergent thinking and thereby
increasingly requires convergent thinking as evidence
accumulates across their trials. McGlynn et al. (2004)
conduct an experimental design with two groups: nom-
inal and interacting groups to find answers to the
following research questions: (1) How effective is
group brainstorming in producing ideas in
a problem-solving context? (2) Does group brain-
storming affect task performance? As a result of their
study, McGlynn et al. (2004) found out that “brain-
storming did not generally affect performance on the
induction task . . . Group brainstorming was perceived
as more effective than individual brainstorming by
both interacting and nominal group members,
a finding that extends the illusion of group productiv-
ity in brainstorming to tasks of convergent thinking.”
(p. 75). McGlynn et al. (2004) conclude that, for face-
to-face groups generating hypotheses in the service of
inductive problem-solving, brainstorming can have
some facilitating effects on task performance.
Osborn (1963) believed that the effectiveness of
group brainstorming for generating creative ideas in
organizations could be attributed to both cognitive and
social processes. Based on this, Dugosh and Paulus
(2005) present that there are social and cognitive fac-
tors influencing group brain storming:
" Many studies have verified that groups generate more
ideas when they use Osborn’s brainstorming rules than
Brainstorming and Learning B 487
when they do not . . . However, when the performance
of interactive brainstorming groups is compared to
the pooled performance of the same number of indi- B
viduals brainstorming alone (nominal groups), nominal
groups outperform interactive groups in both the
quantity and quality of ideas generated . . . Several
social and procedural factors have been identified as
potential causes for this productivity gap, including
evaluation apprehension . . ., social loafing and free-
riding . . ., production blocking . . ., and downward per-
formance matching . . . Most brainstorming research
has focused on social factors in the productivity gap
between interactive and nominal groups . . . However,
researchers have recently begun to investigate
cognitive factors as well, in particular the extent to
which idea exchange influences idea generation.
(p. 313)
Dugosh and Paulus (2005) provide evidence for
both social and cognitive factors in brainstorming:
" The social aspect . . . there may be a mutual influence
process during group brainstorming in which the pro-
ductivity of one brainstormer affects that of another. In
novel task situations, participants may be uncertain
about appropriate performance standards. This may
lead them to imitate or socially compare themselves
with co-performers . . . The cognitive aspect . . . sug-
gests that exposure to ideas from others can stimulate
associations that lead to the generation of additional
ideas . . . Ideas from others can stimulate concepts in
long-term memory, which are connected by means of
a semantic network. This can lead to an associational
chain of ideas . . . Group members initially generate the
ideas whose associations are most accessible in mem-
ory, then proceed to generate ideas whose associa-
tions are less accessible, until they run out of ideas. In
an interactive context, group members may provide
external cues that activate ideas that are low in acces-
sibility for a particular individual . . . The stimulating
impact of ideas from others depends on the extent to
which people attend to these ideas and retain them in
memory during the brainstorming session. (p. 314)
According to Dugosh and Paulus (2005), during
brainstorming session, attention to shared ideas of
a group is important for cognitive stimulation of addi-
tional creative ideas and for enhancing the impact of
the diverse perspectives being shared in a group.
488 B Breaches

Cross-References
▶ Active Learning Bruner, Jerome S (1915–)
▶ Problem Solving
NORBERT M. SEEL
References Department of Education, University of Freiburg,
Baruah, J., & Paulus, P. B. (2008). Effects of training on idea- Freiburg, Germany
generation in groups. Small Group Research, 39, 523–541.
Brown, V. R., & Paulus, P. B. (2002). Making group brainstorming more
effective: recommendations from an associative memory perspec- Life Dates
tive. Current Directions in Psychological Science, 11, 208–212. Jerome Bruner was born in New York City on October 1,
Dugosh, K. L., & Paulus, P. B. (2005). Cognitive and social compar- 1915. He received his B.A. degree from Duke University
ison processes in brainstorming. Journal of Experimental Social
in 1937 and his Ph.D. from Harvard in 1947, where he
Psychology, 41, 313–320.
McGlynn, R. P., McGurk, D., Effland, V. S., Johll, N. J., & Harding, was a member of the faculty at the Department of
D. J. (2004). Brainstorming and task performance in groups Psychology from 1952 to 1972. Furthermore, he was
constrained by evidence. Organizational Behavior and Human the cofounder and director of the Center for Cognitive
Decision Processes, 93, 75–87. Studies at Harvard. In 1972, Bruner left Harvard to teach
Osborn, A. F. (1953, rev. 1957, 1963). Applied imagination: Principles
at the University of Oxford, where he remained until
and procedures of creative problem-solving. New York: Charles
Scribner’s Sons.
1979. Then he returned to Harvard as a visiting profes-
Rossiter, J. R., & Lilien, G. L. (1994). New “brainstorming” principles. sor, before joining the faculty of the new School for
Australian Journal of Management, 19(1), 61–72. Social Research in New York City 2 years later.
Jerome S. Bruner is one of the best known and
most influential psychologists of the twentieth century.
He was one of the key figures in the so-called cognitive
revolution – but the field that was influenced most by
Breaches his work is doubtlessly education, in which he was
engaged from the late 1950s on.
▶ Learning Through the Breach: Language
Socialization
Theoretical Background
" It was, we thought, an all-out effort to establish mean-
ing as the central concept of psychology . . . It focused
upon the symbolic activities that human beings
Breaks in Frames employed in constructing and in making sense not
only of the world, but of themselves. (Bruner 1990, p. 2)
▶ Learning Through the Breach: Language
Socialization With these words described Bruner the paradigm
shift from behaviorism to cognitive psychology in the
late 1950s, which still giving substance to contempo-
rary psychology today. His theoretical framework is
based on the notion that learners actively construct
Broad Science Education new concepts and ideas by relating them to existing
knowledge structures. Therefore, Bruner is sometimes
▶ Integrated, Multidisciplinary, and Technology-
considered as a cofounder of constructivist learning
Enhanced Science Education
theories.
However, Bruner began in the 1940s (along with
Leo Postman) with research on perception from
a functional orientation, with a particular emphasis
Broadcast on the role of needs, motivations, and expectations.
In addition, Bruner investigated the role of strategies
▶ Streaming Media in human categorization and its effects on the cognitive

development of children (Bruner et al. 1956). Clearly,
Bruner’s interest in the development of human cogni-
tion and modes of mental representations was widely
influenced by Piaget. Indeed, Bruner agreed with most
of Piaget’s thinking about cognitive development and
semiotic functions – especially with regard to language.
However, like Vygotsky he also stressed the importance
of social setting in the acquisition of language and
cognitive development. Bruner was one of the first
American psychologists who made Vygotsky popular
in the United States.
Bruner’s early research in the 1940s focused not
only on perception but also on the role of strategies in
categorization as a means of human cognition. In con-
trast to the dominating behavioristic paradigm of this
time, Bruner considered children as active problem
solvers who are able to explore even difficult subjects
of interest. He believed that even young children are
capable of grasping the structure of knowledge when
engaged in problem solving and discovery learning.
Bruner emphasized the role of structure in learning
and its role for categorization. Like Cronbach (see
entry), he also focused on learning in response to
instruction and teaching.
Indeed, Bruner’s interest in education and teaching
was – along with his interest in the development of
human cognition – the pervasive feature of his scien-
tific work. He considered instruction as a major factor
for assisting or shaping cognitive growth. His books
The Process of Education (Bruner 1960) and Towards
a Theory of Instruction (Bruner 1966) are widely recog-
nized as educational classics of the twentieth century.
Here, Bruner introduced his ideas of “readiness for
learning” and the spiral curriculum, according to
which basic ideas should be repeated again and again
until a complete understanding and mastery of a subject
is achieved. Basically, Bruner argued that any subject
could be taught at any stage of development in such
a way that its cognitive abilities are met. He believed that
schooling and curricula should be designed to encour-
age and reward intuitive and analytical thinking and
should aim at the development of intuitive skills and
mental leaps in the process of problem solving. In this
context, Bruner also focused on the role of motivation
for learning, and he argued that interest in a subject
matter is the best incentive for learning.
Based on these ideas, Bruner described a theory of
instruction that necessarily involves some major
Bruner, Jerome S (1915–) B 489
features designed to assist students in their cognitive
growth (Bruner 1960). The first feature is the human
predisposition to learn. Motivational, cultural, and per- B
sonal factors contribute to this feature, as does the
ability of parents and teachers to maintain and direct
a child’s spontaneous explorations of the environment.
Bruner argues that children are intrinsically motivated
to learn. The second feature of effective instruction in
Bruner’s theory of instruction has to do with the struc-
ture of knowledge. Though it is possible to structure
knowledge in a way which enables learners to immedi-
ately comprehend the information, there are many
different ways to structure knowledge and learners
also have many different preferences in how they struc-
ture knowledge. Nevertheless, understanding the struc-
ture of a subject facilitates its understanding. In
Bruner’s view, categorization is a basic and effective
process for structuring knowledge effectively because
it is easier to remember structured patterns of infor-
mation than isolated parts. The third feature of effec-
tive instruction is related to the different modes of
mental representation. Whereas Piaget (1959) had dis-
tinguished between seven so-called semiotic functions
(“imitation” or “symbolic play,” “delayed imitation,”
“drawings,” “painting,” “modeling,” the use of “internal
images,” and finally “verbal language” or the use of
linguistic signs), Bruner distinguished between
enactive, iconic, and symbolic modes of mental represen-
tation. Basically, this distinction between three modes
of representation can still be found in cognitive psy-
chology today (see the entry on ▶ Semiotics and Learn-
ing). The fourth feature of effective instruction is the
effective sequencing of learning tasks and activities. As
with the structure of knowledge, sequencing is not an
absolute but rather a relativistic feature because no one
sequencing of tasks will fit every learner. Nevertheless,
sequencing, or the lack of it, can facilitate or impair
learning. Generally, it seems reasonable to operate
along the dimensions of increasing complexity and
difficulty of learning tasks. Additionally, the form and
pacing of reinforcement play an important role for
adjusting instruction to learners.
The implications of Bruner’s theory of instruction
for teaching and curricular development have been
described by Farnham-Diggory (1972) in her compre-
hensive book Cognitive Processes in Education, which is
still one of the best overviews of cognitive psychology
and its applications for education today.
490 B Bruner, Jerome S (1915–)
Contribution(s) to the Field of
Learning
As Gardner (2001) remarks, Bruner has had a profound
influence on our understanding of education and
instruction. Beyond the big picture of education that
he drew (Bruner 1996), however, his most important
contribution to the field of learning was his emphasis
on discovery learning. Inspired by John Dewey’s (1916)
Democracy and Education and Piaget’s (1954) philoso-
phy of education, Bruner argues that “practice in dis-
covering for oneself teaches one to acquire information
in a way that makes that information more readily
viable in problem solving” (Bruner 1961, p. 26). This
article can be considered as the starting point of the
discovery learning movement of the 1960s and 1970s,
which produced an abundance of empirical studies on
discovery learning in general and on guided discovery
learning in particular. Stated briefly, discovery learning
is a method of learning that encourages students to ask
questions, formulate their own tentative answers, and
derive general principles from practical examples or
experiences. Discovery learning occurs when students
use their own mental processes to discover a meaning
or how to do something for themselves. The past
decades have seen the development of numerous
approaches and conceptions designed to systematically
promote problem-solving behavior in the classroom.
They may be positioned on a scale ranging from inde-
pendent discovery learning to guided discovery learn-
ing and controlled-insightful learning. In accordance
with Bruner, educational psychologists argue that stu-
dents acquire special procedures and strategies for
problem solving during discovery learning, that they
develop an intrinsic motivation for subsequent learn-
ing and problem-solving processes, and that they learn
in time to find the path of discovery on their own, thus
freeing themselves increasingly from their dependency
on confirmation from the teacher.
Straka and Macke (1979) described the fundamen-
tal characteristics of controlled-insightful learning and
guided discovery learning and determined that
controlled-insightful learning consists in large part in
an assimilation of new content into existing cognitive
structures. In contrast to controlled-insightful learn-
ing, the approach of guided discovery learning (as
developed by Bruner) emphasizes the independent
mental processes of problem solving and the discovery
of new relations and principles. The focus is on
enabling the learner to analyze problems cognitively.
By providing suitable learning tasks and materials, the
teacher enables the learners to solve problems and learn
further on their own in later life. According to Bruner,
the main external factor which can influence discovery
learning is the prestructuring of problem situations.
Concretely, he calls for learning material that provokes,
promotes, and supports active and independent infor-
mation processing, that enables the learner to organize
knowledge structures through generalization (by
constructing schemata), stabilizes attitudes and
motives in support of discovery learning, and effects
a transition from extrinsic to intrinsic motivation. In
order to support the process of discovery and the
necessary transfer of knowledge, Bruner recommended
special learning aids which can either be process- or
product-oriented depending on whether they aid the
learner in comprehending the problem, collecting nec-
essary information, forming a hypothesis, or searching
for solution alternatives or whether they only place
relevant information, useful previous knowledge,
given relationships, or sub-solutions at the learner’s
disposal.
Today, discovery learning is an inquiry-based, con-
structivist learning theory that refers to problem-
solving situations in which the learners have to discover
solutions to problems by referring to past experiences
and existing knowledge ([Hammer 1997]; see also the
entry on ▶ Discovery Learning). Since the 1960s when
Bruner introduced the idea of discovery learning, an
abundance of studies on this inquiry-based theory of
learning has been carried out with heterogeneous
results (see, e.g., Hermann 1969; Kirschner et al. 2006).
Cross-References
▶ Constructivist Learning
▶ Discovery Learning
▶ Guided Discovery Learning
▶ History of the Science of Learning
▶ Problem-Based Learning
▶ Problem Solving
▶ Semiotics and Learning
References
Bruner, J. S. (1960). The process of education. Cambridge, MA: Har-
vard University Press.
Bruner, J. S. (1961). The act of discovery. Harvard Educational Review,
31(1), 21–32.

Bruner, J. S. (1966). Toward a theory of instruction. Cambridge, MA:
Belkapp Press.
Bruner, J. (1990). Acts of meaning. Cambridge, MA: Harvard Univer-
sity Press.
Bruner, J. (1996). The culture of education. Cambridge, MA: Harvard
University Press.
Bruner, J., Goodnow, J., & Austin, A. (1956). A study of thinking.
New York: Wiley.
Dewey, J. (1916). Democracy and education. New York: Simon and
Schuster.
Farnham-Diggory, S. (1972). Cognitive processes in education:
A psychological preparation for teaching and curriculum develop-
ment. New York: Harper & Row.
Gardner, H. (2001). Jerome S. Bruner. In J. A. Palmer (Ed.), Fifty
modern thinkers on education. From Piaget to the present. London:
Routledge.
Hammer, D. (1997). Discovery learning and discovery teaching.
Cognition and Instruction, 15(4), 485–529.
Hermann, G. (1969). Learning by discovery: a critical review of
studies. Journal of Experimental Education, 38(1), 58–72.
Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance
during instruction does not work: An analysis of the failure of
constructivist, discovery, problem-based, experiential, and
inquiry-based teaching. Educational Psychologist, 41(2), 75–86.
Piaget, J. (1954). To understand is to invent. New York: Grossman.
Piaget, J. (1959). La naissance de l’intelligence chez l’enfant. Neuchâtel,
Suisse: Delachaux et Niestlé.
Straka, G. A., & Macke, G. (1979). Lehren und Lernen in der Schule.
Stuttgart: Kohlhammer (Teaching and learning in school).
Internet Sources
http://www.guardian.co.uk/education/2007/mar/27/academicexperts.
higher educationprofile
http://www.youtube.com/watch?v=r2H_swMUlOg
http://www.youtube.com/watch?v=yFYBY_YUH5I&feature=related
Building Experience
▶ Individual Learning
Bullying
▶ Learned Aggression in Humans
Buoyancy
▶ Resilience and Learning
Burnout in Teaching and Learning B 491
Burnout in Teaching and
Learning B
AYALA M. PINES
Department of Business Administration, Ben-Gurion
University of the Negev, Beer-Sheva, Israel
Synonyms
Existential perspective on burnout; Learners, learning
and burnout; Psychoanalytic perspective on burnout;
Stress and burnout; Teachers and burnout; The role of
discipline problems in teacher burnout
Definition
Burnout characterizes people who entered their careers
with high hopes, ideals, and ego involvement. It is
the end result of a process of attrition that is experi-
enced as a state of physical, emotional, and mental
exhaustion, lowered sense of accomplishment, and
depersonalization.
Theoretical Background
Since the introduction of burnout to the scientific
community in the mid-1970s, teacher burnout has
received extensive and continuous research attention
(e.g., Montgomery and Rupp 2005). It has even been
argued that teachers are the largest homogenous occu-
pational group that is investigated in burnout research
(22% of all samples, according to Wilmar Schaufeli in
a 1998 talk). Much of this research documented the
existence of teacher burnout and attempted to identify
job-related variables that cause it. The reason for the
emphasis on job-related variables has been the assump-
tion that burnout is related to organizational more
than to personal variables, as well as research showing
that teacher burnout is better explained by organiza-
tional than by personal predictors.
One example of an organizational variable identified
as causing teacher burnout is class size, or more accu-
rately, student–teacher ratio. A number of studies
showed that in large classes, teachers spend more time
restraining and disciplining their students and less time
educating and encouraging them. Other organizational
variables identified as causing teacher burnout were role
ambiguity, role conflict, overload, lack of voice in deci-
sion making, and lack of feedback from colleagues and
492 B Burnout in Teaching and Learning
administration. In addition, studies documented such
causes of teacher burnout as difficulties in managing
disruptive children, incompetent administrators, lack
of administrative support in dealing with discipline
problems, as well as poor salaries, lack of job mobility,
involuntary transfers, public pressure, budget cuts,
demanding parents, excessive paperwork, and excessive
testing.
However, managing disruptive students, student
violence and apathy, and poor relationship between
teacher and students were identified as the best predic-
tor of teacher burnout, followed by administrative
insensitivity and lack of support, bureaucratic incom-
petence, and lack of voice in organizational decision
making (e.g., Farber 1991). The reason for the critical
role disruptive students play in teacher burnout seems
clear: they make it impossible for teachers to teach.
Lack of interest in learning on the part of students lets
teachers know that they have failed as educators. These
“shuttered dreams of impeccable performance” are the
major cause of teacher burnout (Friedman 2000).
“I’ve had it”, “I can’t take it anymore” are the most
common phrases one hears from burned out educa-
tors. Teacher burnout was found to be related to
a number of physical and psychological symptoms.
The most common symptoms are physical and emo-
tional exhaustion and anxiety. Some teachers are con-
stantly fearful and hypervigilant about their personal
safety (Farber 1991). Research also demonstrated the
effect of burnout on teachers’ mental health, somatic
complaints (including insomnia, headaches, and ulcers
as well as abdominal pain, nausea, difficulty in breath-
ing, dizziness, loss of appetite, muscle tightening, cold
sweats, back pains, and occupational injuries, e.g.,
Belcastro 1982). And burnout was found to be related
to a desire to leave teaching and to teacher turnover.
Burnout effects teachers’ performance as well.
Burned out teachers exert less effort than they once
did, they are no longer motivated, patient or optimistic
and they look for ways to reduce their involvement
with students (Farber 1991). Clearly, burnout is very
costly for teachers, students, schools, and society at
large.
A number of theoretical models and perspectives
have been offered in an attempt to explain teacher
burnout. Isaac Friedman, for example, suggested that
self-efficacy and a sense of accomplishment are the
critical psychological mechanisms underlying burnout.
When educators’ need to feel respected and fulfilled at
work is thwarted, burnout is likely. Wilmar Schaufeli,
using a social exchange model, suggested that teacher
burnout is the result of a lack of reciprocity in relation-
ships with students, colleagues, and administrators.
When teachers invest more in these relationships than
they receive in return, burnout is likely. On the basis of
critical theory, Barry Farber suggested that burnout is
not merely a psychological state, but the subjective
experience of a predominantly social problem, the
result of a dynamic interaction between an individual
teacher and a social world.
An alternative explanation was suggested by me
(Pines 2002) based on the assumption that the root
cause of burnout lies in people’s need to believe that
their life is meaningful, that the things they do – and
consequently they themselves – are important and sig-
nificant. Victor Frankl, in his “Man’s Search for Mean-
ing” wrote that “the striving to find meaning in one’s
life is the primary motivational force in man.” And
Ernest Becker in “The Denial of Death,” argued that
people’s need to believe that the things they do are
meaningful is their way of coping with the angst caused
by facing their mortality. In order to be able to deny
death, people need to feel heroic, to know that they
matter in the larger “cosmic” scheme of things. In
previous eras, religion most often gave people the
answer to their existential quest. Today, for many peo-
ple, religion is no longer adequate. For them, the most
frequently chosen alternative is work. People who
choose this alternative try to derive from their work
a sense of meaning for their entire life.
The relevance of the existential perspective to the
case of teacher burnout is supported by research show-
ing teacher burnout to be related to such higher order
needs as a need for self-actualization, which includes
the need for success, achievement, and working at one’s
full potential. (e.g., Malanowski and Wood 1984). In
addition, the existential perspective has the advantage
of incorporating the other conceptualizations of the
dynamic underlying teacher burnout described earlier.
What are the self-efficacy and sense of accomplishment
that Friedman wrote about, if not the feeling that your
actions, and consequently you yourself, matter in the
larger scheme of things? What is the lack of reciprocity
in relationships with students, colleagues, and admin-
istrators Schaufeli wrote about, if not a denial of the
significance of your efforts? As Farber noted, burnout is

the subjective experience of failure in a socially defined
hero system (to use Becker’s terminology).
The existential perspective can also explain key
findings in teacher burnout research. Why, for exam-
ple, is managing disruptive students so often the top-
ranking item in teacher stress surveys if not, as most
teachers indeed note, because discipline problems
impair their teaching effectiveness? In other words,
disruptive students make it impossible for teachers to
derive a sense of significance from their work. The
reason large classes are difficult for teachers is that in
such classes, teachers spend too much time restraining
and disciplining their students and not enough time
educating them – the activity that contributes most to
their sense of significance.
Other organizational variables identified as causing
teacher burnout can also be explained by the existential
perspective: Administrative insensitivity, bureaucratic
incompetence, lack of participation in organizational
decision making, and lack of organizational support
also let teachers know that they are insignificant in the
larger – organizational – scheme of things. Overload
causes burnout because teachers feel that they can’t do
their work the way it should be done. Poor salaries and
budget cuts imply to teachers that society doesn’t value
their work. Excessive paperwork and excessive testing
make teachers feel that they are spending their time
satisfying administrative demands, instead of teaching.
If we accept the general premise that people today
are trying to derive a sense of existential significance
from their work, the next question we need to address is
why they choose to do it via the particular career that
they have chosen. Why does one person try to achieve
a sense meaning by being a teacher, another by being
a nurse, and a third by being a police officer? The choice
of a career is a complex and multifaceted process. Many
attempts have been made to identify the factors that
influence this process. Most of these attempts included
such factors as aptitudes, abilities, interests, ambitions,
resources, limitations, requirements of success as well
as opportunities. Psychoanalytic theory makes
a significant contribution to this body of research by
adding the dimension of unconscious career choices.
The unconscious determinants of any vocational
choice reflect the individual’s personal and familial
history. People choose an occupation that enables
them to replicate significant childhood experiences,
gratify needs that were ungratified in their childhood,
Burnout in Teaching and Learning B 493
and actualize occupational dreams and professional
expectations passed on to them by their familial heri-
tage. When the choice of a career involves such signif- B
icant issues, people enter it with very high hopes and
expectations, high ego involvement and passion. This is
why, when they feel that they failed, the result is devas-
tating, and a major cause of burnout.
A study that supported the existential perspective
on teacher burnout included several samples of Israeli
and American teachers as well as two comparison sam-
ples of nurses and police officers. The study explored
the goals and expectations teachers had when they
entered a teaching career, the sense of significance
they had in their work, the causes of their burnout,
and the level of their burnout (Pines 2002).
The findings revealed a negative correlation between
the sense of significance teachers derived from teaching
and their level of burnout: The greater the sense of
significance, the lower the level of burnout. In was fur-
ther found that teachers entered their profession with
similar goals and expectations (to teach, educate, and
inspire) and that the primary causes of their burnout
(discipline problems) were related to the frustration of
these goals and expectations. In addition, the Israeli
teachers (because of the threats to the very survival of
their nation, which enhanced their sense of significance)
reported lower levels of burnout than their American
counterparts. It was also found that religious teachers
reported lower levels of burnout than their nonreligious
counterparts, thus demonstrating the effect of religious
belief as a buffer against burnout – in further support of
the existential perspective. Finally, it was found that the
experience of teacher burnout was both similar to and
different from burnout of nurses and police officers
(e.g., while the highest item for all three groups was
“tired,” the teachers were more tired than the police
officers and less tired than the nurses). This suggests
that people in the human services share some common
characteristics in their burnout profile, while some char-
acteristics of the composite burnout score are profession
specific (Pines 2002).
Even if teaching appears to be a high burnout
occupation, learning should be the antithesis of burn-
out. Unfortunately, this is rarely the case when formal
learning is concerned. The following example demon-
strates the reasons. Years ago, when I was just starting
my research on burnout, I needed a comparison sample
for some of the professional groups I was studying
494 B Byte-Sized Learning
(such as teachers, nurses, and police officers), so I gave
the Burnout Measure to a group I expected to have very
low level of burnout – UC Berkeley students. These
were students who ranked at the top 10% of their
high school classes, they were studying in one of the
best institutions of higher learning in the world, in
a beautiful campus, they were young, and it was spring-
time. To my great surprise, I discovered that they had
very high levels of burnout. There were two primary
reasons for this sad state of affairs: (1) the students were
not studying topics they were interested in, but
required courses they had little interest in – which
prevented them from deriving a sense of significance
from their school work; (2) the competition for grades
was fierce (since the students knew that very few will be
accepted to graduate school, and acceptance was based
on grades) – which prevented them from using their
classmates as a social support network.
Important Scientific Research and
Open Questions
Unlike teacher burnout, very few studies addressed burn-
out in learning (and those few focused on the learning
environment), so research on the causes, correlates, and
consequences of burnout in learning is still needed.
What can be done to reduce or avoid burnout in
teaching and in learning?
In both cases, the recommendation is similar: every
effort should be made to increase the teacher’s or learner’s
sense of significance, because it is not stress that causes
burnout but the feeling that one’s effort is insignificant.
Cross-References
▶ Achievement Motivation and Learning
▶ Affective and Emotional Dispositions of/for
Learning
▶ Anxiety, Stress, and Learning
▶ Coping with Stress
▶ Flow Experience and Learning
▶ Humanistic Theory of Learning: Maslow
▶ Motivation and Learning: Modern Theories
▶ Motivation Enhancement
▶ Motivation to Learn
▶ Personality and Learning
▶ Personality Effects on Learning
▶ Self-Efficacy for Self-Regulated Learning
▶ Stress Management
References
Belcastro, P. A. (1982). Burnout and its relationship to teachers’
somatic complaints and illness. Psychological Reports, 50,
1045–1046.
Farber, B. A. (1991). Crisis in education: Stress and burnout in the
American teacher. San Francisco, CA: Jossey Bass.
Friedman, I. (2000). Burnout: Shattered dreams of impeccable
performance. Journal of Clinical Psychology, 56(5), 595–606.
Malanowski, J. R., & Wood, P. H. (1984). Burnout and self-
actualization in public school teachers. The Journal of Psychology,
117, 23–26.
Montgomery, C., & Rupp, A. A. (2005). A meta-analysis for exploring
the diverse causes and effects of stress in teachers. Canadian
Journal of Education, 28, 458–486.
Pines, A. M. (2002). Teacher burnout: A psychodynamic existential
perspective. Teachers and Teaching: Theory and Practice, 8,
121–140.
Byte-Sized Learning
▶ Microlearning