Professional Documents
Culture Documents
In this intriguing New York Times article, Pam Belluck reports on a Purdue University
study just published in Science comparing how well college students learned using four
different methods to study a non-fiction passage:
- Studying it once students read the passage for about five minutes;
- Repeatedly studying students re-read the passage four times in consecutive
5-minute sessions;
- Concept mapping students referred to the passage to construct bubble
diagrams showing connections among the facts and concepts;
- Retrieval practice students put the passage aside and wrote down everything
they could remember, then looked back at the original passage, put it aside
again, and wrote as much as they could remember.
A week later, students were given a short-answer test on their recall of the facts in the
passage and their ability to draw logical conclusions from those facts. Students who had
used retrieval practice did significantly better than the other three groups on the facts
and on inferential reasoning.
In second place was repeated study, concept mapping was third, and studying once a
distant fourth. In a second experiment in which researchers compared just concept
mapping and retrieval practice, students who used retrieval practice outperformed the
concept mapping group even at drawing concept maps.
Why does retrieval practice work so well? Researchers speculate that being challenged
to remember information helps us organize it in our brains and creates cues and
connections, reinforcing it in memory so that its easier to access later on. With a
computer, when we retrieve information, nothing inside changes its simple playback.
But with the human brain, when we retrieve information, we strengthen access
pathways to that information. What we recall becomes more recallable in the future,
says UCLA psychologist Robert Bjork. In a sense, you are practicing what you are going
to need to do later.
Learning is about retrieving, says Jeffrey Karpicke, the lead researcher in the Purdue
study. So it is important to make retrieval practice an integral part of the learning
process.
Karpicke has been struck by the fact that during his studies, students dont think
retrieval practice is going to be the most effective strategy. With a passage right in front
of them, students think they know it better than they actually do. It may be surprising
to realize that there is such a disconnect between what students think will afford good
learning and what is actually best, says Karpicke. We, as educators, need to keep this
in mind as we create learning tools and evaluate educational practices.
Part of whats going on here is that retrieval practice seems harder and stirs up more
self-doubt and frustration as people use it. The struggle helps you learn, but it makes
you feel like youre not learning, says Williams College psychologist Nate Kornell. I
dont know it that well. This is hard and Im having trouble coming up with this
information. Students using repeated studying and concept mapping were
overconfident as they studied; they had the illusion that they knew the material better
than they actually did.
Kornell believes retrieval practice should be used extensively in classrooms to make
important information stick in students minds. Its going to last for the rest of their
schooling, he says, and potentially for the rest of their lives.
Take a Test to Really Learn, Research Suggests by Pam Belluck in The New York Times, Jan. 21, 2011 (p.
A12)
http://www.nytimes.com/2011/01/21/science/21memory.html?_r=1&scp=2&sq=
Pam%20Belluck&st=cse
The full article, Retrieval practice produces more learning than elaborative studying with concept mapping
by Purdue University professor Jeffrey Karpicke and Janell Blunt, was published online in Science, January
20, 2011, available at
http://www.sciencedaily.com/releases/2011/01/110121111216.htm
Dfinition
Le constructivisme est une thorie de l'apprentissage fonde sur l'ide que la connaissance est
construite par l'apprenant sur la base d'une activit mentale. Les tudiants sont considrs
comme des organismes actifs cherchant du sens, des significations. Le constructivisme est
bas sur l'hypothse que, en rflchissant sur nos expriences, nous construisons notre propre
vision du monde dans lequel nous vivons. Chacun de nous produit ses propres rgles et
modles mentaux , que nous utilisons pour donner un sens nos expriences. Apprendre
est donc simplement un processus d'ajustement de nos modles mentaux pour s'adapter de
nouvelles expriences. Les constructions du sens peuvent au dbut ne soutenir que peu le
rapport avec la ralit (comme dans les thories naves des enfants), mais elles deviendront de
plus en plus complexes, diffrencies et ralistes au fil du temps.
Historique
Le constructivisme est issu, entre autres, des travaux de Jean Piaget (1964) qui met la thorie
quun individu confront une situation donne va mobiliser un certain nombre de structures
cognitives, quil nomme schmes opratoires. Lapprentissage ou la sophistication des
schmes opratoires se fait travers deux processus complmentaires :
Pour que le mcanisme daccommodation opre, il faut, dans un premier temps, quune
tentative dassimilation ait lieu afin que les structures cognitives soient dj mobilises.
Lassimilation cre une perturbation au sein des structures cognitives que Piaget nomme
conflit cognitif qui est elle-mme rgule afin darriver une nouvelle forme dquilibre.
On considre ce niveau quil y a eu une progression au sein des stades ou des sous-stades de
dveloppement.
Constructivisme trivial
La racine de toutes les autres nuances de constructivisme dcrites ci-dessous, est le
constructivisme trivial (von Glasersfeld, 1990), ou constructivisme personnel . Dans cette
perspective, la connaissance est activement construite par l'tudiant, et non passivement
reue de l'environnement. (Plus de dtails...)
Constructivisme radical
Le constructivisme radical ajoute un seconde principe au constructivisme trivial (von
Glasersfeld, 1990) :Apprendre est un processus d'adaptation dynamique vers des
interprtations d'exprience plausibles . L'apprenant ne construit pas ncessairement la
connaissance partir d'un vrai monde . (Plus de dtails...)
Socio-constructivisme
Le monde social d'un apprenant inclut les gens qui affectent directement cette personne, y
compris les professeurs, amis, tudiants, administrateurs, et les participants toutes formes
d'activit . Il faut donc tenir compte de la nature sociale des processus locaux dans
l'apprentissage collaboratif et dans la discussion d'une collaboration sociale plus large
pour un sujet donn, par exemple la science. (Plus de dtails...)
Constructivisme culturel
Au-del de l'environnement social immdiat d'une situation d'tude on trouve le contexte plus
large des influences culturelles, y compris la coutume, la religion, la biologie, les outils et
la langue . Par exemple, le format des livres peut affecter l'apprentissage, en favorisant des
ides sur l'organisation, l'accessibilit et le statut de l'information qu'ils contiennent. (Plus de
dtails...)
Constructivisme critique
Le constructivisme critique considre le constructivisme dans un environnement social et
culturel, mais y ajoute une dimension critique qui a pour objectif de reformer ces
environnements afin d'amliorer le succs du constructivisme appliqu comme rfrent .
(Plus de dtails...)
Constructionnisme
Le constructionnisme affirme que l'approche constructiviste est particulirement efficace
lorsque l'apprenants est engag dans une construction destine d'autres. (Plus de dtails...)
Conclusions
Le constructivisme est une manire de penser le savoir, une rfrence pour construire des
modles de l'enseignement, de l'apprentissage et des programmes d'tudes (Tobin et Tippin,
1993). Dans ce sens, c'est une philosophie.
Le constructivisme peut galement tre utilis pour fonder une thorie de la communication.
Quand vous envoyez un message en disant quelque chose ou en fournissant des informations,
et que vous n'avez aucune connaissance du rcepteur, alors vous n'avez aucune ide de la
manire dont le message a t reu, et vous ne pouvez pas clairement interprter la rponse.
Dans cette perspective, l'enseignement devient l'tablissement et le maintien d'une langue et
de moyens de communication entre le professeur et les apprenants, aussi bien qu'entre les
apprenants eux-mmes. Prsenter simplement le matriel, annoncer les problmes et recevoir
des rponses n'est pas un processus de communication assez raffin pour un apprentissage
efficace.
Certains principes du constructivisme en termes pdagogiques :
Dans une perspective constructiviste, les tudiants sont actifs dans la recherche du sens.
L'enseignement cherche ce que les tudiants peuvent analyser, tudier, et comment ils peuvent
collaborer, partager, construire et gnrer du sens sur la base de ce qu'ils savent dj, plutt
que quels faits, qualifications ou processus ils peuvent singer. Pour y parvenir efficacement,
un enseignant doit tre un tudiant et un chercheur. Il doit essayer d'obtenir une plus grande
conscience des environnements et des participants dans une situation d'enseignement donne.
Ainsi, pour engager les tudiants dans l'apprentissage, il pourra s'ajuster continuellement
leurs actions et s'appuyer sur le constructivisme comme une rfrence.
Voir galement
socio-constructivisme, dcouverte apprenant, WebQuest...
Rfrences
College Of Education :
http://www.usask.ca/education/coursework/802papers/Skaalid/definition.html
Costa, A. & Liebmann, R. (1995). Process is as important as content. Educational Leadership.
52(6), pp 23-24.
Dougiamas, M. (1998). A journey into Constructivism,
http://dougiamas.com/writing/constructivism.html
Ellis, C. (1996). Evocative Autoethnography: Writing Emotionally about our lives. In: W.G.
Tierney and Y.S. Lincoln (Eds) Reframing the Narrative Voice. Funderstanding :
http://www.funderstanding.com/constructivism.cfm
McBrien, J.L. & Brandt, R.S.(1997). From The Language of Learning: A Guide to Education
Terms. Alexandria, VA: Association for Supervision and Curriculum
Development.http://www.ascd.org/portal/site/ascd/menuitem.d36b986168f3f8cddeb3ffdb621
08a0c/
Piaget, J. (1964). Six Etudes de Psychologie. Genve: Editions Gonthier.>
Tobin, K. & Tippins, D (1993) Constructivism as a Referent for Teaching and Learning. In:
K. Tobin (Ed) The Practice of Constructivism in Science Education, pp 3-21, LawrenceErlbaum, Hillsdale, NJ.
Von Glasersfeld, E. (1990) An exposition of constructivism: Why some like it radical. In R.B.
Davis, C.A. Maher and N. Noddings (Eds), Constructivist views on the teaching and learning
of mathematics (pp 19-29). Reston, Virginia: National Council of Teachers of Mathematics.
Wood, T., Cobb, P. & Yackel, E. (1995). Reflections on learning and teaching mathematics in
elementary school. In L. P. Steffe & J.Gale (Eds) Constructivism in education (pp 401-422).
Hillsdale, New Jersey: Lawrence Erlbaum.
Jeffrey D. Karpicke
Purdue University
Abstract
Retrieval is the key process for understanding learning and for promoting learning, yet retrieval is not often granted the central
role it deserves. Learning is typically identified with the encoding or construction of knowledge, and retrieval is considered
merely the assessment of learning that occurred in a prior experience. The retrieval-based learning perspective outlined
here is grounded in the fact that all expressions of knowledge involve retrieval and depend on the retrieval cues available in
a given context. Further, every time a person retrieves knowledge, that knowledge is changed, because retrieving knowledge
improves ones ability to retrieve it again in the future. Practicing retrieval does not merely produce rote, transient learning;
it produces meaningful, long-term learning. Yet retrieval practice is a tool many students lack metacognitive awareness of
and do not use as often as they should. Active retrieval is an effective but undervalued strategy for promoting meaningful
learning.
Keywords
retrieval processes, learning, education, metacognition, meaningful learning
If you know something, or if you have stored information about an event from the distant past, and never use
that information, never think of it, your brain is functionally equivalent to that of an otherwise identical brain
that does not contain that information.
Endel Tulving (1991)
To understand learning, it is essential to understand the processes involved in retrieving and reconstructing knowledge. We
may think we know something, that our minds contain or possess some knowledge, but the only way to assess knowledge is
by engaging in an act of retrieval. Differences in the ability to
recover knowledge may not stem from what is stored in our
minds but rather from differences in the retrieval cues available
in particular contexts. Given the fundamental importance of
retrieval for understanding the process of learning, it is surprising that retrieval processes have not received more attention in
educational research. Consider that over the past decade, many
influential National Research Council books on how people
learn have contained no mention of retrieval (National Research
Council, 2000, 2005a, 2005b).
It is essential to consider retrieval processes not only
because they are central to understanding learning but also
because the act of retrieval itself is a powerful tool for enhancing learning. Moreover, active retrieval does not merely
produce rote, transient learning; it produces meaningful, longterm learning. The idea that retrieval is the centerpiece for
158
and instructional practices have placed a premium on identifying the best ways to encode knowledge and experiences.
Retrieval processes, the processes involved in using available
cues to actively reconstruct knowledge, have received less
attention. There seems to be a tacit assumption that successful
encoding or construction of knowledge, in itself, is sufficient
to ensure learning.
Basic research on learning and memory, however, has
emphasized that retrieval must be considered in any analysis
of learning. In part, this is because people do not store static,
exact copies of experiences that are reproduced verbatim at
retrieval. Instead, knowledge is actively reconstructed on the
basis of the present context and available retrieval cues
(Bartlett, 1932; Neisser, 1967; see too Moscovitch, 2007;
Roediger, 2000). The reconstructive nature of mind is revealed
in the systematic errors people make in retrieving knowledge,
errors that verbatim recording devices would not make. The
past never occurs again in its exact form, so a mental storehouse of copies of past experiences would be of little use.
People instead have the ability to use the past to meet the
demands of the present by reconstructing knowledge rather
than reproducing it exactly.
What people express when they reconstruct knowledge
depends on the retrieval cues available in a given context. Ultimately, knowledge reconstruction depends on the diagnostic
value of cues, the degree to which cues help people recover
particular target information to the exclusion of competing
candidates (Nairne, 2002; Raaijmakers & Shiffrin, 1981). We
may wish to examine what a person has constructed or stored
in mind, but it is impossible to directly assess the contents of
storage, per se. We can only ever examine what a person
reconstructs given the available cues and context (Roediger,
2000; Roediger & Guynn, 1996; Tulving & Pearlstone, 1966).
Thus, it is essential to consider retrieval processes in any analysis of learning.
The second crucial reason retrieval is important for learning is that learning is altered by the act of retrieval itself. Every
time a person retrieves knowledge, that knowledge is changed,
because retrieving knowledge improves ones ability to
retrieve it again in the future (Karpicke & Roediger, 2007,
2008; Karpicke & Zaromb, 2010). This is a feature of a functional learning and memory system. Our minds are sensitive to
the likelihood that we will need knowledge at a future time,
and if we retrieve something in the present, there is a good
chance we will need to recover it again. The process of retrieval
itself alters knowledge in anticipation of demands we may
encounter in the future. Retrieval is therefore not only a tool
for assessing learning but also a tool for enhancing learning
(Roediger & Karpicke, 2006a).
Karpicke
would you want to do next? You have three options: You can
(a) go back and restudy either all of the material or parts of it,
(b) try to recall the material without restudying afterward, or
(c) do something else. Which would you choose?
We (Karpicke, Butler, & Roediger, 2009) gave this question to a large group of college students. Most students (57%)
said they would reread their notes or textbook, and 21% said
they would do something else. Only 18% said they would
attempt to recall material after reading it.1 The decision to
repeatedly read makes sense if we identify learning with processes of encoding and constructing knowledge and consider
retrieval to be only a way to assess prior learning. It stands to
reason that more studying (i.e., more encoding and knowledge
construction) should produce more learning, whereas retrieval
should measure learning but not produce it.
Would students be better off repeatedly reading than engaging in retrieval? We conducted an experiment with a design
that mirrored the question asked in the survey (Roediger &
Karpicke, 2006b). Students read educational texts and recalled
them under one of three conditions. One group of students
spent time repeatedly studying a text in four study periods. A
second group read a text in three study periods and then
recalled it in one retrieval period (labeled SSSR), in which the
students wrote down as many ideas from the text as they could
recall. A third group read the text during one study period and
then practiced recalling it during three consecutive repeated
retrieval periods. Students did not reread the text or receive
any feedback after any of the recall periods; they only practiced actively retrieving material.
At the end of the learning phase, the students made a judgment of learning: a prediction of how well they would remember the material in the future. Then, one week later, students
recalled the material again to see how much they actually
retained in the long term.
Figure 1b shows students judgments of learning. The more
times students repeatedly read the material, the better they
believed they had learned it. However, Figure 1a shows that
students actual learning exhibited the opposite pattern. The
more times students practiced actively retrieving the material,
the better they retained it in the long term. Students spent the
same amount of time experiencing the material in all three
conditions, and students in the repeated-retrieval condition
only recalled and did not restudy the text, yet active retrieval
produced the best long-term retention (for further discussion
of metacognitive awareness of the effects of retrieval practice,
see Karpicke & Grimaldi, 2012).
Returning to the survey of student learning strategies
(Karpicke et al., 2009), one might think the results would
change if we reworded the survey question. Namely, students
might choose to engage in active retrieval if they could reread
after attempting retrieval. In a second version of the survey,
we asked students the exact question described above, but
changed option (b) to say, try to recall the material, and then
go back and restudy the text. Students choices did change,
but not as dramatically as one might expect: Forty-two percent
159
Retrieval-Based Learning
Actual Recall
Predicted Recall
.7
4.8
Judgment of Learning
5.0
.6
.5
.4
.3
.2
SSSS
SSSR
SRRR
4.6
4.4
4.2
4.0
3.8
3.6
SSSS
SSSR
SRRR
Fig. 1. Final recall (a) after repeatedly studying a text in four study periods (SSSS condition), reading a text
in three study periods and then recalling it in one retrieval period (SSSR condition), or reading a text in
one study period and then repeatedly recalling it in three retrieval periods (SRRR condition). Judgments of
learning (b) were made on a 7-point scale, where 7 indicated that students believed they would remember
material very well. The data presented in these graphs are adapted from Experiment 2 of Roediger and
Karpicke (2006b). The pattern of students metacognitive judgments of learning (predicted recall) was
exactly the opposite of the pattern of students actual long-term retention.
.9
.8
.7
.6
.5
.4
.3
.2
.1
.0
Study
Retrieve
Once
Repeated
Retrieval
160
Karpicke
Verbatim Questions
Inference Questions
.8
.8
.7
.7
Proportion Correct
Proportion Correct
a
.6
.5
.4
.3
.2
.1
.0
.6
.5
.4
.3
.2
.1
.0
Metacognitive Predictions
Judgment of Learning
.8
.7
.6
.5
.4
.3
.2
.1
.0
Fig. 3.Proportion correct on final short-answer verbatim questions (a) and inference questions (b) 1 week after learning,
and metacognitive judgments of learning (predicted proportion of items correct) made during the initial learning phase (c).
Practicing retrieval enhanced long-term learning relative to elaborative studying with concept mapping, yet this benefit was largely
unanticipated by students. Reprinted from Retrieval Practice Produces More Learning Than Elaborative Studying With Concept
Mapping, by J. D. Karpicke and J. R. Blunt, 2011, Science, 331, pp. 772775. Copyright 2011 by the American Association for the
Advancement of Science. Reprinted with permission.
161
Retrieval-Based Learning
Actual Performance
.5
.4
.3
.2
b
Judgment of Learning
Proportion Correct
Concept
Mapping
Retrieval
Practice
Metacognitive Predictions
.5
.4
.3
.2
Concept
Mapping
Retrieval
Practice
Fig. 4. Proportion correct on a final concept-map test (a) and metacognitive judgments of learning
(predicted proportion of items correct) during the initial learning phase (b). The data presented
in these graphs are adapted from Experiment 2 of Karpicke and Blunt (2011b). Even though the
criterial test involved creating a concept map, practicing retrieval during original learning enhanced
performance relative to elaborative studying with concept mapping, despite the fact that students
believed elaborative concept mapping would produce better learning.
162
learning of the material. Yet once again, the students generally
believed they would do better after elaborative studying than
after practicing retrieval, as shown in Figure 4b. Most students
did not expect active retrieval to produce more learning than
elaborative studying with concept mapping, but in fact it did.
Karpicke
Roediger, H. L. (2000). (See References). An in-depth discussion of
why retrieval is the key process for understanding learning.
Roediger, H. L., & Karpicke, J. D. (2006a). (See References). A comprehensive and historical overview of research on the effects of
active retrieval on learning.
Tulving, E. (1991). (See References). Essential reading for any
student of learning and memory: an interview in which Tulving explains why retrieval processes must be considered in any
analysis of learning.
Acknowledgments
The author would like to thank Stephanie Karpicke for assistance
with preparing the manuscript.
Funding
The writing of this paper was supported by National Science
Foundation Grant DUE-0941170 and by the U.S. Department of
Educations Institute of Education Sciences Grant R305A110903.
The opinions expressed here are those of the author and do not represent views of the Institute of Education Sciences or the U.S.
Department of Education.
Note
1. We were unable to score ambiguous responses given by 4% of
students.
2. There is some evidence that retrieving a portion of some material
can impair recovery of nonpracticed parts of the material (Storm,
2011), but with meaningful and integrated educational materials,
practicing retrieval of part of the materials typically enhances learning of nonpracticed parts of the materials (Chan, 2009; Chan,
McDermott, & Roediger, 2006).
References
Bartlett, F. C. (1932). Remembering: A study in experimental and
social psychology. Cambridge, England: Cambridge University
Press.
Butler, A. C. (2010). Repeated testing produces superior transfer
of learning relative to repeated studying. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 36, 11181133.
Chan, J. C. K. (2009). When does retrieval induce forgetting and
when does it induce facilitation? Implications for retrieval inhibition, testing effect, and text processing. Journal of Memory and
Language, 61, 153170.
Chan, J. C. K., McDermott, K. B., & Roediger, H. L. (2006).
Retrieval-induced facilitation: Initially non-tested material can
benefit from prior testing of related material. Journal of Experimental Psychology: General, 135, 553571.
Congleton, A., & Rajaram, S. (in press). The origin of the interaction
between learning method and delay in the testing effect: The roles
of processing and conceptual retrieval organization. Memory &
Cognition.
Retrieval-Based Learning
Grimaldi, P. J., & Karpicke, J. D. (in press). When and why do retrieval
attempts enhance subsequent encoding? Memory & Cognition.
Hinze, S. R., & Wiley, J. (2011). Testing the limits of testing effects
using completion tests. Memory, 19, 290304.
Johnson, C. I., & Mayer, R. E. (2009). A testing effect with multimedia learning. Journal of Educational Psychology, 101, 621
629.
Karpicke, J. D. (2009). Metacognitive control and strategy selection:
Deciding to practice retrieval during learning. Journal of Experimental Psychology: General, 138, 469486.
Karpicke, J. D., & Blunt, J. R. (2011a). Response to comment on
Retrieval practice produces more learning than elaborative
studying with concept mapping. Science, 334, 453.
Karpicke, J. D., & Blunt, J. R. (2011b). Retrieval practice produces
more learning than elaborative studying with concept mapping.
Science, 331, 772775.
Karpicke, J. D., Butler, A. C., & Roediger, H. L. (2009). Metacognitive strategies in student learning: Do students practice retrieval
when they study on their own? Memory, 17, 471479.
Karpicke, J. D., & Grimaldi, P. J. (2012). Retrieval-based learning:
A perspective for enhancing meaningful learning. Unpublished
manuscript, Purdue University, West Lafayette, IN.
Karpicke, J. D., & Roediger, H. L. (2007). Repeated retrieval during
learning is the key to long-term retention. Journal of Memory and
Language, 57, 151162.
Karpicke, J. D., & Roediger, H. L. (2008). The critical importance of
retrieval for learning. Science, 319, 966968.
Karpicke, J. D., & Roediger, H. L. (2010). Is expanding retrieval a
superior method for learning text materials? Memory & Cognition, 38, 116124.
Karpicke, J. D., & Zaromb, F. M. (2010). Retrieval mode distinguishes the testing effect from the generation effect. Journal of
Memory and Language, 62, 227239.
Kornell, N., Hays, M. J., & Bjork, R. A. (2009). Unsuccessful retrieval
attempts enhance subsequent learning. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 35, 989998.
Mayer, R. E. (2008). Learning and instruction (2nd ed.). Upper Saddle River, NJ: Pearson.
McDaniel, M. A., Howard, D. C., & Einstein, G. O. (2009). The readrecite-review study strategy: Effective and portable. Psychological Science, 20, 516522.
Moscovitch, M. (2007). Memory: Why the engram is elusive. In
H. L. Roediger, Y. Dudai, & S. M. Fitzpatrick (Eds.), Science of
memory: Concepts (pp. 1721). New York, NY: Oxford University Press.
Nairne, J. S. (2002). The myth of the encoding-retrieval match. Memory, 10, 389395.
163
National Research Council. (2000). How people learn: Brain, mind,
experience, and school. J. D. Bransford, A. L. Brown, & R. R.
Cocking (Eds.). Washington, DC: National Academies Press.
National Research Council. (2005a). How students learn: Science in
the classroom. M. S. Donovan & J. D. Bransford (Eds.). Washington, DC: National Academies Press.
National Research Council. (2005b). Taking science to school:
Learning and teaching science in grades K-8. R. A. Duschl,
H. A. Schweingruber, & A. W. Shouse (Eds.). Washington, DC:
National Academies Press.
Neisser, U. (1967). Cognitive psychology. New York, NY: AppletonCentury-Crofts.
Novak, J. D., & Gowin, D. B. (1984). Learning how to learn. Cambridge, England: Cambridge University Press.
Raaijmakers, J. G. W., & Shiffrin, R. M. (1981). Search of associative
memory. Psychological Review, 88, 93134.
Roediger, H. L. (1980). Memory metaphors in cognitive psychology.
Memory & Cognition, 8, 231246.
Roediger, H. L. (2000). Why retrieval is the key process in understanding human memory. In E. Tulving (Ed.), Memory, consciousness, and the brain: The Tallinn conference (pp. 5275).
Philadelphia, PA: Psychology Press.
Roediger, H. L., & Guynn, M. J. (1996). Retrieval processes. In E. L.
Bjork & R. A. Bjork (Eds.), Memory (pp. 197236). San Diego,
CA: Academic Press.
Roediger, H. L., & Karpicke, J. D. (2006a). The power of testing
memory: Basic research and implications for educational practice. Perspectives on Psychological Science, 1, 181210.
Roediger, H. L., & Karpicke, J. D. (2006b). Test enhanced learning:
Taking memory tests improves long-term retention. Psychological Science, 17, 249255.
Rohrer, D., Taylor, K., & Sholar, B. (2010). Tests enhance the transfer of learning. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 36, 233239.
Storm, B. C. (2011). The benefit of forgetting in thinking and remembering. Current Directions in Psychological Science, 20, 291295.
Tulving, E. (1991). Interview. Journal of Cognitive Neuroscience, 3,
8994.
Tulving, E., & Pearlstone, Z. (1966). Availability vs. accessibility of
information in memory for words. Journal of Verbal Learning
and Verbal Behavior, 5, 381391.
Wissman, K. T., Rawson, K. A., & Pyc, M. A. (2011). The interim test
effect: Testing prior material can facilitate the learning of new
material. Psychonomic Bulletin & Review, 18, 11401147.
Zaromb, F. M., & Roediger, H. L. (2010). The testing effect in free
recall is associated with enhanced organizational processes.
Memory & Cognition, 38, 9951008
From the competition of these pre-paradigmatic schools, one paradigm emerges - "To be accepted as
a paradigm, a theory must seem better than its competitors, but it need not, and in fact never does,
explain all the facts with which it can be confronted", thus making research possible. As a paradigm
grows in strength and in the number of advocates, the other pre-paradigmatic schools or the previous
paradigm fade.
A paradigm transforms a group into a profession or, at least, a discipline. And from this follow the
formation of specialised journals, foundation of professional bodies and a claim to a special place in
academe. There is a promulgation of scholarly articles intended for and "addressed only to
professional colleagues, [those] whose knowledge of a shared paradigm can be assumed and who
prove to be the only ones able to read the papers addressed to them".
III - The Nature of Normal Science.
If a paradigm consists of basic and incontrovertible assumptions about the nature of the discipline,
what questions are left to ask?
When they first appear, paradigms are limited in scope and in precision. But more successful does not
mean completely successful with a single problem or notably successful with any large number.
Initially, a paradigm offers the promise of success. Normal science consists in the actualisation of that
promise. This is achieved by extending the knowledge of those facts that the paradigm displays as
particularly revealing, increasing the extent of the match between those facts and the paradigm's
predictions, and further articulation of the paradigm itself.
In other words, there is a good deal of mopping-up to be done. Mop-up operations are what engage
most scientists throughout their careers. Mopping-up is what normal science is all about! This
paradigm-based research is "an attempt to force nature into the pre-formed and relatively inflexible
box that the paradigm supplies". No effort is made to call forth new sorts of phenomena, no effort to
discover anomalies. When anomalies pop up, they are usually discarded or ignored. Anomalies are
usually not even noticed and no effort is made to invent a new theory (and theres no tolerance for
those who try). Those restrictions, born from confidence in a paradigm, turn out to be essential to the
development of science. By focusing attention on a small range of relatively esoteric problems, the
paradigm forces scientists to investigate some part of nature in a detail and depth that would
otherwise be unimaginable" and, when the paradigm ceases to function properly, scientists begin to
behave differently and the nature of their research problems changes.
IV - Normal Science as Puzzle-solving.
Doing research is essentially like solving a puzzle. Puzzles have rules. Puzzles generally have
predetermined solutions.
A striking feature of doing research is that the aim is to discover what is known in advance. This in
spite of the fact that the range of anticipated results is small compared to the possible results. When
the outcome of a research project does not fall into this anticipated result range, it is generally
considered a failure.
So why do research? Results add to the scope and precision with which a paradigm can be applied.
The way to obtain the results usually remains very much in doubt - this is the challenge of the puzzle.
Solving the puzzle can be fun, and expert puzzle-solvers make a very nice living. To classify as a
puzzle (as a genuine research question), a problem must be characterised by more than the assured
solution, but at the same time solutions should be consistent with paradigmatic assumptions.
Despite the fact that novelty is not sought and that accepted belief is generally not challenged, the
scientific enterprise can and does bring about unexpected results.
V - The Priority of Paradigms.
The paradigms of a mature scientific community can be determined with relative ease. The "rules"
used by scientists who share a paradigm are not so easily determined. Some reasons for this are that
scientists can disagree on the interpretation of a paradigm. The existence of a paradigm need not
imply that any full set of rules exist. Also, scientists are often guided by tacit knowledge - knowledge
acquired through practice and that cannot be articulated explicitly. Further, the attributes shared by a
paradigm are not always readily apparent.
Paradigms can determine normal science without the intervention of discoverable rules or shared
assumptions. In part, this is because it is very difficult to discover the rules that guide particular
normal-science traditions. Scientists never learn concepts, laws, and theories in the abstract and by
themselves. They generally learn these with and through their applications. New theory is taught in
tandem with its application to a concrete range of phenomena.
Sub-specialties are differently educated and focus on different applications for their research findings.
A paradigm can determine several traditions of normal science that overlap without being coextensive.
Consequently, changes in a paradigm affect different sub-specialties differently. "A revolution
produced within one of these traditions will not necessarily extend to the others as well".
When scientists disagree about whether the fundamental problems of their field have been solved, the
search for rules gains a function that it does not ordinarily possess .
VI - Anomaly and the Emergence of Scientific Discoveries.
If normal science is so rigid and if scientific communities are so close-knit, how can a paradigm
change take place? Paradigm changes can result from discovery brought about by encounters with
anomaly.
Normal science does not aim at novelties of fact or theory and, when successful, finds none.
Nonetheless, new and unsuspected phenomena are repeatedly uncovered by scientific research, and
radical new theories have again and again been invented by scientists . Fundamental novelties of fact
and theory bring about paradigm change. So how does paradigm change come about? There are two
ways: through discovery - novelty of fact - or by invention novelty of theory. Discovery begins with
the awareness of anomaly - the recognition that nature has violated the paradigm-induced
expectations that govern normal science. The area of the anomaly is then explored. The paradigm
change is complete when the paradigm has been adjusted so that the anomalous become the
expected. The result is that the scientist is able "to see nature in a different way".. How paradigms
change as a result of invention is discussed in greater detail in the following chapter.
Although normal science is a pursuit not directed to novelties and tending at first to suppress them, it
is nonetheless very effective in causing them to arise. Why? An initial paradigm accounts quite
successfully for most of the observations and experiments readily accessible to that science's
practitioners. Research results in the construction of elaborate equipment, development of an esoteric
and shared vocabulary, refinement of concepts that increasingly lessens their resemblance to their
usual common-sense prototypes. This professionalisation leads to immense restriction of the
scientist's vision, rigid science, resistance to paradigm change, and a detail of information and
precision of the observation-theory match that can be achieved in no other way. New and refined
methods and instruments result in greater precision and understanding of the paradigm. Only when
researchers know with precision what to expect from an experiment can they recognise that something
has gone wrong.
Consequently, anomaly appears only against the background provided by the paradigm . The more
precise and far-reaching the paradigm, the more sensitive it is to detecting an anomaly and inducing
change. By resisting change, a paradigm guarantees that anomalies that lead to paradigm change will
penetrate existing knowledge to the core.
VII - Crisis and the Emergence of Scientific Theories.
As is the case with discovery, a change in an existing theory that results in the invention of a new
theory is also brought about by the awareness of anomaly. The emergence of a new theory is
generated by the persistent failure of the puzzles of normal science to be solved as they should.
Failure of existing rules is the prelude to a search for new ones . These failures can be brought about
by observed discrepancies between theory and fact or changes in social/cultural climates Such
failures are generally long recognised, which is why crises are seldom surprising. Neither problems
nor puzzles yield often to the first attack . Recall that paradigm and theory resist change and are
extremely resilient. Philosophers of science have repeatedly demonstrated that more than one
theoretical construction can always be placed upon a given collection of data . In early stages of a
paradigm, such theoretical alternatives are easily invented. Once a paradigm is entrenched (and the
tools of the paradigm prove useful to solve the problems the paradigm defines), theoretical
alternatives are strongly resisted. As in manufacture so in science--retooling is an extravagance to be
reserved for the occasion that demands it . Crises provide the opportunity to retool.
VIII - The Response to Crisis.
The awareness and acknowledgement that a crisis exists loosens theoretical stereotypes and provides
the incremental data necessary for a fundamental paradigm shift. Normal science does and must
continually strive to bring theory and fact into closer agreement. The recognition and
acknowledgement of anomalies result in crises that are a necessary precondition for the emergence of
novel theories and for paradigm change. Crisis is the essential tension implicit in scientific research.
There is no such thing as research without counterinstances. These counterinstances create tension
and crisis. Crisis is always implicit in research because every problem that normal science sees as a
puzzle can be seen, from another viewpoint, as a counterinstance and thus as a source of crisis .
In responding to these crises, scientists generally do not renounce the paradigm that has led them into
crisis. Rather, they usually devise numerous articulations and ad hoc modifications of their theory in
order to eliminate any apparent conflict. Some, unable to tolerate the crisis, leave the profession. As a
rule, persistent and recognised anomaly does not induce crisis . Failure to achieve the expected
solution to a puzzle discredits only the scientist and not the theory To evoke a crisis, an anomaly must
usually be more than just an anomaly. Scientists who paused and examined every anomaly would not
get much accomplished. An anomaly must come to be seen as more than just another puzzle of
normal science.
All crises begin with the blurring of a paradigm and the consequent loosening of the rules for normal
research. As this process develops, the anomaly comes to be more generally recognised as such,
more attention is devoted to it by more of the field's eminent authorities. The field begins to look quite
different: scientists express explicit discontent, competing articulations of the paradigm proliferate and
scholars view a resolution as the subject matter of their discipline. To this end, they first isolate the
anomaly more precisely and give it structure. They push the rules of normal science harder than ever
to see, in the area of difficulty, just where and how far they can be made to work.
All crises close in one of three ways. (i) Normal science proves able to handle the crisis-provoking
problem and all returns to "normal." (ii) The problem resists and is labelled, but it is perceived as
resulting from the field's failure to possess the necessary tools with which to solve it, and so scientists
set it aside for a future generation with more developed tools. (iii) A new candidate for paradigm
emerges, and a battle over its acceptance ensues. Once it has achieved the status of paradigm, a
paradigm is declared invalid only if an alternate candidate is available to take its place . Because there
is no such thing as research in the absence of a paradigm, to reject one paradigm without
simultaneously substituting another is to reject science itself. To declare a paradigm invalid will require
more than the falsification of the paradigm by direct comparison with nature. The judgement leading to
this decision involves the comparison of the existing paradigm with nature and with the alternate
candidate. Transition from a paradigm in crisis to a new one from which a new tradition of normal
science can emerge is not a cumulative process. It is a reconstruction of the field from new
fundamentals. This reconstruction changes some of the field's foundational theoretical generalisations.
It changes methods and applications. It alters the rules.
How do new paradigms finally emerge? Some emerge all at once, sometimes in the middle of the
night, in the mind of a man deeply immersed in crisis. Those who achieve fundamental inventions of a
new paradigm have generally been either very young or very new to the field whose paradigm they
changed. Much of this process is inscrutable and may be permanently so.
IX - The Nature and Necessity of Scientific Revolutions.
Why should a paradigm change be called a revolution? What are the functions of scientific revolutions
in the development of science?
A scientific revolution is a non-cumulative developmental episode in which an older paradigm is
replaced in whole or in part by an incompatible new one . A scientific revolution that results in
paradigm change is analogous to a political revolution. Political revolutions begin with a growing sense
by members of the community that existing institutions have ceased adequately to meet the problems
posed by an environment that they have in part created. The dissatisfaction with existing institutions is
generally restricted to a segment of the political community. Political revolutions aim to change political
institutions in ways that those institutions themselves prohibit. As crisis deepens, individuals commit
themselves to some concrete proposal for the reconstruction of society in a new institutional
framework. Competing camps and parties form. One camp seeks to defend the old institutional
constellation. One (or more) camps seek to institute a new political order. As polarisation occurs,
political recourse fails. Parties to a revolutionary conflict finally resort to the techniques of mass
persuasion.
Like the choice between competing political institutions, that between competing paradigms proves to
be a choice between fundamentally incompatible modes of community life. Paradigmatic differences
cannot be reconciled. When paradigms enter into a debate about fundamental questions and
paradigm choice, each group uses its own paradigm to argue in that paradigm's defence The result is
a circularity and inability to share a universe of discourse. A successful new paradigm permits
predictions that are different from those derived from its predecessor . That difference could not occur
if the two were logically compatible. In the process of being assimilated, the second must displace the
first.
Consequently, the assimilation of either a new sort of phenomenon or a new scientific theory must
demand the rejection of an older paradigm . If this were not so, scientific development would be
genuinely cumulative. Normal research is cumulative, but not scientific revolution. New paradigms
arise with destructive changes in beliefs about nature.
Consequently, "the normal-scientific tradition that emerges from a scientific revolution is not only
incompatible but often actually incommensurable with that which has gone before". In the circular
argument that results from this conversation, each paradigm will satisfy more or less the criteria that it
dictates for itself, and fall short of a few of those dictated by its opponent. Since no two paradigms
leave all the same problems unsolved, paradigm debates always involve the question: Which
problems is it more significant to have solved? In the final analysis, this involves a question of values
that lie outside of normal science altogether. It is this recourse to external criteria that most obviously
makes paradigm debates revolutionary..
X - Revolutions as Changes of World View.
During scientific revolutions, scientists see new and different things when looking with familiar
instruments in places they have looked before. Familiar objects are seen in a different light and joined
by unfamiliar ones as well. Scientists see new things when looking at old objects. In a sense, after a
revolution, scientists are responding to a different world.
Why does a shift in view occur? Genius? Flashes of intuition? Sure. Because different scientists
interpret their observations differently? No. Observations are themselves nearly always different.
Observations are conducted within a paradigmatic framework, so the interpretative enterprise can only
articulate a paradigm, not correct it. Because of factors embedded in the nature of human perception
and retinal impression? No doubt, but our knowledge is simply not yet advanced enough on this
matter. Changes in definitional conventions? No. Because the existing paradigm fails to fit? Always.
Because of a change in the relation between the scientist's manipulations and the paradigm or
between the manipulations and their concrete results? You bet. It is hard to make nature fit a
paradigm.
XI - The Invisibility of Revolutions.
Because paradigm shifts are generally viewed not as revolutions but as additions to scientific
knowledge, and because the history of the field is represented in the new textbooks that accompany a
new paradigm, a scientific revolution seems invisible.
The image of creative scientific activity is largely created by a field's textbooks. Textbooks are the
pedagogic vehicles for the perpetuation of normal science. These texts become the authoritative
source of the history of science. Both the layman's and the practitioner's knowledge of science is
based on textbooks. A field's texts must be rewritten in the aftermath of a scientific revolution. Once
rewritten, they inevitably disguise not only the role but the existence and significance of the revolutions
that produced them. The resulting textbooks truncate the scientist's sense of his discipline's history
and supply a substitute for what they eliminate. More often than not, they contain very little history at
all. In the rewrite, earlier scientists are represented as having worked on the same set of fixed
problems and in accordance with the same set of fixed canons that the most recent revolution and
method has made seem scientific. Why dignify what science's best and most persistent efforts have
made it possible to discard?
The historical reconstruction of previous paradigms and theorists in scientific textbooks make the
history of science look linear or cumulative, a tendency that even affects scientists looking back at
their own research . These misconstructions render revolutions invisible. They also work to deny
revolutions as a function. Science textbooks present the inaccurate view that science has reached its
present state by a series of individual discoveries and inventions that, when gathered together,
constitute the modern body of technical knowledge - the addition of bricks to a building. This
piecemeal-discovered facts approach of a textbook presentation illustrates the pattern of historical
mistakes that misleads both students and laymen about the nature of the scientific enterprise. More
than any other single aspect of science, the textbook has determined our image of the nature of
science and of the role of discovery and invention in its advance.
XII - The Resolution of Revolutions.
How do the proponents of a competing paradigm convert the entire profession or the relevant
subgroup to their way of seeing science and the world? What causes a group to abandon one tradition
of normal research in favour of another?
Scientific revolutions come about when one paradigm displaces another after a period of paradigmtesting that occurs only after persistent failure to solve a noteworthy puzzle has given rise to crisis.
This process is analogous to natural selection: one theory becomes the most viable among the actual
alternatives in a particular historical situation.
What is the process by which a new candidate for paradigm replaces its predecessor? At the start, a
new candidate for paradigm may have few supporters (and the motives of the supporters may be
suspect). If the supporters are competent, they will improve the paradigm, explore its possibilities, and
show what it would be like to belong to the community guided by it. For the paradigm destined to win,
the number and strength of the persuasive arguments in its favour will increase. As more and more
scientists are converted, exploration increases. The number of experiments, instruments, articles, and
books based on the paradigm will multiply. More scientists, convinced of the new view's fruitfulness,
will adopt the new mode of practising normal science, until only a few elderly hold-outs remain. And
we cannot say that they are (or were) wrong. Perhaps the scientist who continues to resist after the
whole profession has been converted has ipso facto ceased to be a scientist.
XIII - Progress Through Revolutions.
In the face of the arguments previously made, why does science progress, how does it progress, and
what is the nature of its progress?
To a very great extent, the term science is reserved for fields that do progress in obvious ways. But
does a field make progress because it is a science, or is it a science because it makes progress?
Normal science progresses because the enterprise shares certain salient characteristics, Members of
a mature scientific community work from a single paradigm or from a closely related set. Very rarely do
different scientific communities investigate the same problems. The result of successful creative work
is progress.
Even if we argue that a field does not make progress, that does not mean that an individual school or
discipline within that field does not. The man who argues that philosophy has made no progress
emphasises that there are still Aristotelians, not that Aristotelianism has failed to progress. It is only
during periods of normal science that progress seems both obvious and assured. In part, this progress
is in the eye of the beholder. The absence of competing paradigms that question each other's aims
and standards makes the progress of a normal-scientific community far easier to see. The acceptance
of a paradigm frees the community from the need to constantly re-examine its first principles and
foundational assumptions. Members of the community can concentrate on the subtlest and most
esoteric of the phenomena that concern it. Because scientists work only for an audience of
colleagues, an audience that shares values and beliefs, a single set of standards can be taken for
granted. Unlike in other disciplines, the scientist need not select problems because they urgently need
solution and without regard for the tools available to solve them. The social scientists tend to defend
their choice of a research problem chiefly in terms of the social importance of achieving a solution.
Which group would one then expect to solve problems at a more rapid rate? .
We may have to relinquish the notion, explicit or implicit, that changes of paradigm carry scientists and
those who learn from them closer and closer to the truth . The developmental process described by
Kuhn is a process of evolution from primitive beginnings. It is a process whose successive stages are
characterised by an increasingly detailed and refined understanding of nature. This is not a process of
evolution toward anything. Important questions arise. Must there be a goal set by nature in advance?
Does it really help to imagine that there is some one full, objective, true account of nature? Is the
proper measure of scientific achievement the extent to which it brings us closer to an ultimate goal?
The analogy that relates the evolution of organisms to the evolution of scientific ideas "is nearly
perfect" . The resolution of revolutions is the selection by conflict within the scientific community of the
fittest way to practice future science. The net result of a sequence of such revolutionary selections,
separated by period of normal research, is the wonderfully adapted set of instruments we call modern
scientific knowledge. Successive stages in that developmental process are marked by an increase in
articulation and specialisation. The process occurs without benefit of a set goal and without benefit of
any permanent fixed scientific truth. What must the world be like in order than man may know it?
processes are gone through to disprove an existing theory, and also the technique needed to
develop new theories which may eventually be accepted by the scientific community.
Kuhn starts off with a brief introduction of the role for history, which he then relates to a term
called normal science which will be defined later in this summary. Next, Kuhn explains the
difference between paradigms, normal science, and scientific discoveries. The scientific
discoveries lead to scientific theories which can cause crises within the scientific community.
These crises can lead to a scientific revolution which will potentially change the way the
world views a particular subject - referred to by Kuhn as a paradigm shift.
The following is a brief summary of each of the chapters in The Structure of Scientific
Revolutions, but we highly recommend that you take it upon yourself to read and enjoy the
book yourself as we did.
Contents
This chapter explores the relationship between rules, paradigms, and normal science. Kuhn
uses an example throughout the chapter of the terms chair, leaf, and game and
describes them in the scientific method described throughout the chapter.
He then moves on to discuss four reasons that paradigms should be looked at as more then
just a theoretical aspect of normal science.
These conversions occur despite the resistance of some of the scientific community.
Lisa K. Son
Nate Kornell
Williams College, USA,
Bridgid Finn
Washington University, USA
Memory research has a 125-year history dating back to the seminal work of
Ebbinghaus (1885/1964). In this time, the field has matured into a science with
reliable data, firm principles, and well-established theories. In recent years, cognitive
psychologists have increasingly shifted their focus towards identifying principles of
learning and memory that can be used to make recommendations about enhancing
educational practices (e.g. Metcalfe, Kornell, & Son, 2007; Roediger, 2009; Roediger
& Karpicke, 2006a; Rohrer & Pashler, 2010; Willingham, 2009). While the approach
has stemmed from cognitive science, educational psychologists have long grappled
with similar objectives and issues (see Groninger, 1971; Richardson, 1998). Here, we
attempt to organize -- in a style that is succinct and useful for both teachers and
students -- the findings into four principles of memory improvement that can be
applied to real-word educational problems. The principles can be defined broadly as
follows: 1) process material actively, 2) practice retrieval 3) use distributed practice,
and 4) use metamemory.
FOUR PRINCIPLES
Figure 1. Roediger and Karpicke (2006a) Showed That Retrieval Practice Has Long-term
Benefits for Learning at Longer Retention Intervals. SSSS = All Studying, SSS =
Three Study Sessions Followed by 1 Testing. STTT = One Study Session,
Followed by Three Testing Trials.
Are there situations where it is not advisable to test oneself? For example, it might
seem pointless to take a test on something one has not yet started to learnparticularly because generating incorrect answers might seem like it would have harmful
effects. Yet research indicates that even an unsuccessful retrieval attempt, if it is
followed by feedback, can be more effective than an opportunity to study information
without being tested (Finn & Metcalfe, 2010; Kornell, Hays, & Bjork, 2009; Richland,
Kornell, & Kao, 2009). That is, even in a situation in which retrieval practice seems
as if it might be counterproductive, it is actually effective.
Recent studies have shown that people do not necessarily recognize the benefits of
tests (e.g., Karpicke, 2009). In one experiment, participants judged restudying to be
more effective than testing, even though they actually learned more via testing
(Kornell & Son, 2009). Fortunately, when given the choice of how to study, they
preferred self-testing. They chose to test themselves, apparently, to see how well they
were doing. Thus, in this case, people chose to self-test for a good reason, but not
because they realized that self-testing is itself effective for learning (Karpicke, 2009).
These recent findings go against an intuitive and deep-seated belief, among students
and educators, in the value of errorless learning (but see Wilson, 2009, for the
advantages of errorless learning in amnesic patients). However, it should be noted
that in some situations, errors that are made on a test, especially if they are uncorrected, can persist into the future (see Richardson, 1998). The fact that self-testing is
10
unintuitive makes it all the more important to communicate the moral of this research:
test thyself!
USE DISTRIBUTED PRACTICE
Distributed practice is learning that is spread out across relatively long periods of time
rather than massed all at once. Beginning with Ebbinghaus (1885/1964), hundreds of
studies have demonstrated the benefits of distributed learningthe so-called spacing
effect (see Cepeda, Pashler, Vul, Wixted, & Rohrer, 2006; see Dempster, 1988). Spacing effects have been demonstrated in motor learning, verbal learning, mathematics
learning, and in other educational settings (as described by Cepeda et al., 2006;
Kornell, 2009). Distributed practice may be the most powerful method by which
people can improve their memory without changing the amount of time spent
studying (Baddeley & Longman, 1978; Bjork, 1988; Kornell, 2009).
In a recent study that involved a practical application of distributed practice,
Kornell(2009) asked participants to study difficult word-synonym pairs using a computerized flashcard paradigm. On each trial, the computer displayed a relatively
difficult word (e.g., effulgent); when the participant pressed a button the words
synonym appeared (e.g., brilliant). Participants studied each pair 8 timeshalf of the
participants learned by massing; the other half by distributing trials over four days. At
final test, cued recall was up to twice as accurate in the distributed condition as in the
massed condition (see Figure 2). The advantage of distribution over massing was present in over 90% of the students. However, over 70% of participants judged massing
as more effective than distributing at the outset of the experiment. Thus, the effects of
distributed practice are strong, consistent, but unfortunately, counterintuitive.
FOUR PRINCIPLES
11
1980). Even for children, though, enforcing distributed study sessionseven when it
goes against their decision to masscan improve learning (Son, 2010; Vlach,
Sandhofer, & Kornell, 2008).
For many students, the alternative to distributed learning is to procrastinate and
then cram just before a test. And while cramming can work in the short term
(Kornell, 2009), it does not produce long-term gains. It may be unrealistic to expect
students, even those who know about distributed practice, to stop relying on cramming. However, educators may help their students by assigning topics in a way that
lead to distributed practice (e.g., Taylor & Rohrer, 2010). To summarize, distributing
one's study may take effort and resolve, but the person who does so will greatly
improve their learning efficiency.
USE METAMEMORY
Metamemory refers to judgments and decisions we make about our own learning and
memory (Karpicke, 2009; Metcalfe, 2009; Son, 2010). Beliefs and judgments are
referred to as memory monitoring; decisions that we make based on monitoring are
referred to as control. For example, students monitor their learning while they study,
which allows them to make decisions about how to study, when to study, what
resources to rely on, how much to study, and so forth. We will briefly outline the promise of metamemory for improving study habits and the pitfalls to avoid when
making judgments and decisions about ones learning.
Judgments of learning (henceforth, JOLs) are predictions of future remembering
made at the time of learning (Nelson & Leonesio, 1988). Implicit JOLs (e.g., am I
ready for the test tomorrow?) can play an important role in peoples study decisions.
But those study decisions will tend to be flawed if they are based on inaccurate JOLs.
Fortunately, testing oneselfnot immediately after studying, but after a meaningful
delaygreatly increases the accuracy of JOLs (Dunlosky & Nelson, 1992). In this
context, tests are beneficial because they are an accurate way of diagnosing ones
memory. However, tests have multiple advantages: retrieval directly enhances
learning, and it increases monitoring accuracy.
JOLs are driven by heuristics, and like many other heuristic-based judgments, they
are subject to systematic biases. One near-universal bias is overconfidence (e.g. Son
& Kornell, 2010). Unfortunately, overconfidence is exacerbated by a human tendency
to overlook a fundamental aspect of memory: forgetting. For example, Koriat, Bjork,
Sheffer, and Bar (2004) asked one group of people how they would do on a test in
five minutes and another how they would do on a test on one week. The groups made
almost identical predictionsthat is, participants acted like they would not forget
anything over the course of a week. In reality, of course, they forgot a great deal, and
so their initial judgments were highly overconfident. Overconfidence is problematic
to learning because a student who is overconfident may stop studying prematurely.
Even if judgments are accurate, they are only a first step toward improving learning; making effective study decisions is an additional challenge. One adaptive
strategy is based on Kornell and Metcalfe's(2006) Region of Proximal Learning
model, which states that efficient learning occurs when people study the easiest items
that have not yet been mastered (also see Metcalfe et al., 2007). These items are most
likely to make the crucial transition from being unknown to being memorable later.
12
Consider how this strategy might apply to a student studying for an exam. The
student may begin studying for an exam, only to have her study session interfered
with by any number of factors (a party, a family emergency, etc). If the student had
focused on the most difficult items, she might not have mastered anything, thus
leading to poor test performance (and low learning efficiency). But if she had prioritized learning the easier items, she would be likely to remember those items the
next day.
To illustrate this point, Son and Kornell (2009) asked participants to make JOLs on
GRE word-synonym pairs (e.g., ignominious shameful). Participants then had the
opportunity to choose which items to study, and for how long. In support of the proximal learning model, participants focused on the easier items first, studying them in
advance of the more difficult items. However, participants shifted their focus from the
easy items to the difficult items and studied the latter more often (see Figure 3).
Figure 3. Data from Son and Kornell (2009). Participants Studied 16 Total Items. Depending
on Group, They Could Then Re-study 8, 16, or 24 Times. The Mean Number of
Times An Item Was Studied is Plotted as a Function of Whether the Participant
Gave the Item a High or Low JOL. Items Deemed More Difficult by the Participants
Were Studied More Often; the Data Also Showed, However, That among Items
Selected for Study, Easier Items Tended to be Selected Relatively Early.
In addition to which items to study and for how long, students also face crucial
decisions about how to study. The present article provides guidance for such decisions: People should focus on meaning, practice retrieval, distribute practice, and use
metamemory. Teachers can encourage these techniques using homework assignments
and in-class guidance. Likewise, students can use them when they are deciding for
themselves how to study.
FOUR PRINCIPLES
13
14
FOUR PRINCIPLES
15
On peut concevoir de nouvelles thories (non issues directement de l'exprience) et en tirer par raisonnement logique
et mathmatique des consquences qui pourront alors tre testes exprimentalement, ces expriences rpondant la
demande de vrification de la thorie. Chronologiquement dans ce cas, l'exprience suite la thorie.
Ex : caractre ondulatoire des corpuscules matriels prdit par de Broglie (" = h/mv") en 1924 qu'on ne "voit"
exprimentalement que plus tard (1927) : exprience de Davisson et Germer de diffraction d'un faisceau d'lectrons par
un rseau cristallin (cf. diffraction d'un faisceau de rayons X par un rseau cristallin).
-1-
Dans son ouvrage "La Logique de la dcouverte scientifique", Popper analyse de manire
particulirement critique et rigoureuse cette question de "vrification exprimentale"
ou plus prcisment de "confrontation l'exprience". Il propose aussi de modifier le
schma classique de la dcouverte scientifique.
Critique de Popper
Popper bannit l'usage du principe d'induction.
Il considre comme insurmontables les difficults attaches la logique inductive. En
effet, suivant Popper, d'une part, ce principe d'induction mne des incohrences
logiques que l'on ne peut viter que difficilement (et encore quand cela est possible).
D'autre part, il doit tre lui-mme un nonc universel. Sa vrit doit donc tre connue
par exprience (en suivant la logique positiviste). Il rencontre donc un problme
identique ceux pour la solution desquels il a t introduit ("cercle vicieux de
l'induction"). Soit le principe d'induction drive d'un principe suprieur qui son tour
doit rpondre aux mmes objections et on arrive ainsi une rgression l'infini. Soit on
considre le principe d'induction comme a priori vrai (ex. Kant) et il n'est donc plus
scientifique (voir plus loin le critre de dmarcation par lequel Popper qualifie un nonc
scientifique s'il est rfutable).
Principe de falsifiabilit
Par opposition "l'inductivisme", Popper prne "le dductivisme" ou mthode dductive
de contrle = conception selon laquelle une hypothse ne peut tre que soumise des
tests empiriques (confrontation avec l'exprience) et seulement aprs avoir t
avance.
Une interfrence est dite dductive si elle passe d'un niveau suprieur (nonc
universel) un niveau infrieur (nonc singulier).
Exemple :
Tous les hommes sont mortels (nonc universel). Donc : Socrate est mortel
(nonc singulier).
Comment s'opre la falsification :
Pour la mise l'preuve (testing), Popper prne la dmarche suivante. Le point de
dpart est un nonc2 hypothtique nouveau qui n'a pas encore ce stade reu de
justification (on peut encore dire qu'il est avanc titre d'essai).
Par
raisonnement mathmatique ou logique, on dduit de cet nonc des conclusions
ou prdictions qui sont des noncs singuliers.
Ou une hypothse ou une proposition ou une thorie, par exemple un ensemble d'axiomes.
-2-
Ce sont ces conclusions qui vont tre testes3 avec d'autres noncs singuliers
exprimant les rsultats d'exprimentation (mise l'preuve proprement dite). Si
la confrontation est positive (on n'a pas trouv de contradiction), on n'aura pas
trouv, ce stade ( ce jour) de raison d'carter la thorie, elle sera accepte ;
Popper dit qu'elle est corrobore. Par contre, en cas de dcision ngative, c'est-dire si on trouve au moins un dsaccord (par exemple, on teste "tous les cygnes
sont blancs" et il existe un cygne noir au zoo d'Anvers) une conclusion est
falsifie, cette falsification falsifie aussi la thorie dont elle avait logiquement
t dduite.
Un nonc sera donc accept tant qu'il n'aura pas t falsifi, rfut.
Ce qu'on admet avoir ainsi "vrifi" n'est qu'une tentative de rfutation ngative
(la thorie a subi les tests positivement) : la thorie n'a pas t rfute,
momentanment du moins, mais elle n'a pas t prouve. La "vrit" (abus de
langage) n'est donc que non falsification (non-fausset).
Provisoire et relatif
La corroboration est donc provisoire, puisqu'un test ultrieur pourrait rfuter l'nonc
considr. Un nonc corrobor aujourd'hui peut en effet se voir rfut demain. (On
pensera par exemple l'apparition de nouveaux virus, tel le sida, modifiant la situation
exprimentale dans ce domaine). Par contre, un nonc faux car rfut (donc test)
restera toujours faux (le test falsifiant pouvant toujours tre refait dans les mmes
conditions (pourrait tre critiqu : si les conditions changent), on ne peut voquer
des situations exprimentales qui n'existaient pas puisqu'il y a eu test).
On dira qu'une thorie a fait ses preuves tant qu'elle a rsist provisoirement
des tests systmatiques et rigoureux.
Cette notion a clairement un caractre relatif et provisoire qu'on opposera au caractre
absolu et intemporel de "vrai" et "faux", un nonc corrobor sous-entendant par quels
tests et quand. C'est dans ce sens qu'on peut dire qu'il n'existe pas de vrit
absolue. On fait ainsi un abus de langage, couramment rencontr en appelant vrit un
nonc non rfut.
La mise l'preuve doit tre systmatique et rigoureuse.
Le caractre systmatique a dj t voqu quand nous avons expliqu comment toutes
les prdictions dduites de l'nonc seraient testes, entre elles, et avec des
prdictions issues d'autres thories pralablement admises.
Les prdictions sont aussi testes logiquement entre elles si il y a lieu afin de tester leur cohrence, avec l'nonc de
dpart (contradiction? Cf. les dmonstrations par l'absurde en mathmatique), avec des noncs pralablement admis
issus d'autres thories relatives au problme vis par l'nonc de dpart.
-3-
Objective
On s'assurera de l'objectivit de cette mise l'preuve en soumettant les noncs
(prdictions) considrs des tests intersubjectifs (c'est--dire l'issue desquels les
chercheurs du domaine incrimin sont d'accord de cette manire, on est assur que le
rsultat retenu des tests ne dpend pas de la personne qui l'a ralis.
Reproductible
La mise l'preuve doit tre reproductible.
On ne prendra pas en considration des vnements exceptionnels non reproductibles.
Dans le mme ordre d'ides, on convient de dfinir un effet physique comme un effet
que n'importe qui peut reproduire rgulirement s'il excute l'exprience de la manire
prescrite.
Rle de l'exprience
Chez Popper, le recours l'exprience n'intervient donc pour tester des noncs
(thories) avancs titre d'essai, dcider de leur corroboration ou de leur rfutation.
Il suffit d'une exprience (ici et pour un) falsifiante pour falsifier un nonc universel ;
Exemple :
Tous les cygnes sont blancs (nonc = loi universelle = thorie).
Il y a un cygne noir au zoo d'Anvers (exprience singulire falsifiante).
D'un point de vue logique
On remarquera qu'intervient dans la "vrification" exprimentale des deux dmarches
scientifiques que nous avons examines (inductivisme et rationalisme critique de Popper)
l'asymtrie qui existe entre vrificabilit et falsifiabilit, noncs universels et
noncs singuliers.
Pour tre dclar vrai, un nonc universel devrait tre test un nombre infini de fois
(observer la couleur tous les cygnes de part le monde) alors que pour qu'un nonc
universel soit dclar faux, il suffit de trouver une seule situation exprimentale o il
s'avre faux (c'est--dire un nonc singulier falsifiant, par exemple observer un cygne
noir quelque part).
D'un point de vue logique, l'attitude de Popper est rigoureuse. La dmarche inductive
par contre est plus intuitive (est-ce "scientifique" ?), plus proche du monde de pense
de monsieur tout-le-monde.
[On ne peut pas montrer qu'un nonc universel est vrai mais on peut le falsifier
(exemple : tous les cygnes sont blancs). Un nonc singulier est vrai mais ne peut tre
falsifi (exemple : il y a un cygne noir au zoo d'Anvers).
En rsum, on peut donc montrer qu'un nonc est faux mais on ne peut montrer qu'il
est vrai.
On ne peut donc prouver que la fausset d'un nonc.
-4-
Enoncs ngatifs
Ceci se retrouve dans la formulation de certaines lois universelles par des noncs
"ngatifs" : noncs de non-existence. (exemple : il n'existe pas de charges lectriques
fractionnaires (si on prend la charge de l'lectron comme unit de rfrence).
De telles formulations sont simplement la consquence de l'quivalence logique qui
existe entre un nonc universel (forme habituelle des lois de la nature) et un nonc
existentiel ngatif (nonc "il y a u pas").
Exemple :
Il existe un cygne pas blanc (nonc existentiel prohibitif)
Est quivalent
Tous les cygnes ne sont pas blancs (nonc universel).
Degr de corroboration
On peut associer une hypothse non rfute son degr de corroboration (une manire
de dire comment on la considre comme "vraie"). Le degr de corroboration (la qualit,
le "mrite") d'une hypothse est proportionnel aux possibilits de la rfuter (nombre,
diversit et svrit des tests).
Il est donc fonction de la falsification de l'nonc qui dpend par exemple :
- Du degr d'universalit de l'nonc (son champ d'application) un nonc
concernant les cygnes est moins universel qu'un nonc concernant les
palmipdes, un nonc concernant les palmipdes sera donc plus facilement
falsifiable qu'un nonc concernant les cygnes.
- Des falsificateurs virtuels de l'nonc (noncs de base incompatibles avec
l'nonc ex il y a un cygne noir, ou il existe une charge lectrique
fractionnaire (par rapport la charge de l'lectron).
Principe de dmarcation
Une autre question vient tout naturellement complter la rflexion de Popper au sujet
de la vrit, elle trouve d'ailleurs rponse l'usage de la mme notion de falsifiabilit.
Existe-t-il des critres permettant de dpartager la science de la non-science ou
mtaphysique, c'est--dire rpondre la question : "Est-ce scientifique," Quel critre
appliquer pour dterminer si une proposition (nonc) peut tre considre comme
scientifique ?
K. Popper tablit cet effet un critre de dmarcation ou critre de falsifiabilit qui
est au cur de son pistmologie.
-5-
4
Dire que les noncs scientifiques sont ceux qui peuvent tre soumis des tests ne signifie pas que chaque nonc ait
t soumis des tests mais qu'on exclut l'ide qu'il existe un nonc scientifique qui ne puisse tre soumis des tests et
qu'on doive l'accepter comme vrai.
5
Ex "Pile (P) = " est considr comme "la masse de ce mobile m = 343g".
-6-
auxquels les thories sont soumises (en vue d'une possible rfutation) visant la non prise
en compte d'vnements non reproductibles.
Il tait important pour Popper de traiter des noncs de probabilit qui se rencontrent
frquemment en physique o il existe mme des thories qui ont une formulation
probabiliste comme la mcanique quantique (exemple : hypothse de base :
interprtation de la fonction d'onde associe une particule quantique comme
l'amplitude de probabilit de prsence de trouver cette particule dans une rgion
donne de l'espace) et en physique statistique (qui dtermine des proprits
macroscopiques au dpart de proprits microscopiques et de lois de probabilit,
exemple en thermodynamique lois des gaz parfaits PV = nRT).
Et la science ?
Popper ne conoit donc pas la science comme un systme d'noncs certains. Il dnonce
ainsi l'idal d'une connaissance vraie et absolument dmontrable puisque toute
hypothse n'est corrobore que de manire provisoire et relative. Il ne nie pas le
concept de "vrit" mais montre qu'on ne peut scientifiquement pas l'atteindre.
Pour lui, la science est "la qute obstine et audacieusement critique de la vrit", c'est
d'ailleurs par cette caractristique qu'il dfinit le scientifique et non par la possession
de connaissances, d'irrfutables vrits.
Et suivant Popper, d'o "sort" alors la thorie ? (puisqu'elle ne sort pas de
l'exprience immdiate). Elle provient des amliorations successives et progressives de
la thorie rsultant des confrontations l'exprience. Chaque test ngatif, chaque
rfutation est ainsi source de progrs puisque l'hypothse rfute devra tre
remplace et la nouvelle thorie teste son tour (la difficult est de dterminer quelle
est l'hypothse rfute parmi l'ensemble des hypothses ou noncs ou lois
universelles qui constituent la thorie).
C'est ce type d'attitude qui seul permet d'tablir des nouvelles thories qui prdisent
des effets qui n'ont pas encore t observs. C'est ce qui s'est pass aux dbuts de la
mcanique quantique quand de Broglie a prdit un comportement ondulatoire pour des
corpuscules matriels (dualit onde-corpuscule). Un tel comportement n'avait t
jamais observ et ne le sera qu'aprs l'nonc de Broglie dans l'exprience de Davisson
et Germer ()
Popper modifie ainsi le schma classique de la recherche scientifique, puisqu'en
suivant ses rgles mthodologiques, c'est la thorie avance (nonce) qui "commande"
(dtermine) l'exprience qui pourrait la rfuter. C'est le chercheur qui formule les
questions poser la nature de manire obtenir un "oui" ou un "non". Le recours
l'exprience est ainsi actif et non plus passif ou fortuit (cf. la pomme de Newton).
-7-
-8-
Reprints
This copy is for your personal, noncommercial use only. You can order presentation-ready copies for
distribution to your colleagues, clients or customers here or use the "Reprints" tool that appears next to any
article. Visit www.nytreprints.com for samples and additional information. Order a reprint of this article now.
Taking a test is not just a passive mechanism for assessing how much people know, according to new
research. It actually helps people learn, and it works better than a number of other studying techniques.
The research, published online Thursday in the journal Science, found that students who read a passage,
then took a test asking them to recall what they had read, retained about 50 percent more of the
information a week later than students who used two other methods.
One of those methods repeatedly studying the material is familiar to legions of students who cram
before exams. The other having students draw detailed diagrams documenting what they are learning
is prized by many teachers because it forces students to make connections among facts.
These other methods not only are popular, the researchers reported; they also seem to give students the
illusion that they know material better than they do.
In the experiments, the students were asked to predict how much they would remember a week after using
one of the methods to learn the material. Those who took the test after reading the passage predicted they
would remember less than the other students predicted but the results were just the opposite.
I think that learning is all about retrieving, all about reconstructing our knowledge, said the lead author,
Jeffrey Karpicke, an assistant professor of psychology at Purdue University. I think that were tapping into
something fundamental about how the mind works when we talk about retrieval.
Several cognitive scientists and education experts said the results were striking.
The students who took the recall tests may recognize some gaps in their knowledge, said Marcia Linn, an
education professor at the University of California, Berkeley, and they might revisit the ideas in the back of
their mind or the front of their mind.
When they are later asked what they have learned, she went on, they can more easily retrieve it and
organize the knowledge that they have in a way that makes sense to them.
The researchers engaged 200 college students in two experiments, assigning them to read several
paragraphs about a scientific subject how the digestive system works, for example, or the different types
of vertebrate muscle tissue.
http://www.nytimes.com/2011/01/21/science/21memory.html?pagewanted=print[12/06/2013 9:53:49]
In the first experiment, the students were divided into four groups. One did nothing more than read the
text for five minutes. Another studied the passage in four consecutive five-minute sessions.
A third group engaged in concept mapping, in which, with the passage in front of them, they arranged
information from the passage into a kind of diagram, writing details and ideas in hand-drawn bubbles and
linking the bubbles in an organized way.
The final group took a retrieval practice test. Without the passage in front of them, they wrote what they
remembered in a free-form essay for 10 minutes. Then they reread the passage and took another retrieval
practice test.
A week later all four groups were given a short-answer test that assessed their ability to recall facts and
draw logical conclusions based on the facts.
The second experiment focused only on concept mapping and retrieval practice testing, with each student
doing an exercise using each method. In this initial phase, researchers reported, students who made
diagrams while consulting the passage included more detail than students asked to recall what they had
just read in an essay.
But when they were evaluated a week later, the students in the testing group did much better than the
concept mappers. They even did better when they were evaluated not with a short-answer test but with a
test requiring them to draw a concept map from memory.
Why retrieval testing helps is still unknown. Perhaps it is because by remembering information we are
organizing it and creating cues and connections that our brains later recognize.
When youre retrieving something out of a computers memory, you dont change anything its simple
playback, said Robert Bjork, a psychologist at the University of California, Los Angeles, who was not
involved with the study.
But when we use our memories by retrieving things, we change our access to that information, Dr. Bjork
said. What we recall becomes more recallable in the future. In a sense you are practicing what you are
going to need to do later.
It may also be that the struggle involved in recalling something helps reinforce it in our brains.
Maybe that is also why students who took retrieval practice tests were less confident about how they would
perform a week later.
The struggle helps you learn, but it makes you feel like youre not learning, said Nate Kornell, a
psychologist at Williams College. You feel like: I dont know it that well. This is hard and Im having
trouble coming up with this information.
By contrast, he said, when rereading texts and possibly even drawing diagrams, you say: Oh, this is easier.
http://www.nytimes.com/2011/01/21/science/21memory.html?pagewanted=print[12/06/2013 9:53:49]
http://www.nytimes.com/2011/01/21/science/21memory.html?pagewanted=print[12/06/2013 9:53:49]