Daniel T. Levin & Bonnie K. Angelone - The Visual Metacognition Questionnaire. A Measure of Intuitions About Vision

The Visual Metacognition Questionnaire: A Measure of Intuitions about Vision
Author(s): Daniel T. Levin and Bonnie L. Angelone

Source: The American Journal of Psychology, Vol. 121, No. 3 (Fall, 2008), pp. 451-472
Published by: University of Illinois Press
Stable URL: http://www.jstor.org/stable/20445476
Accessed: 21-01-2016 09:30 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/20445476?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
University of Illinois Press is collaborating with JSTOR to digitize, preserve and extend access to The American Journal of
Psychology.
http://www.jstor.org
This content downloaded from 209.175.73.10 on Thu, 21 Jan 2016 09:30:44 UTC
All use subject to JSTOR Terms and Conditions
The Visual Metacognition

Questionnaire: A measure of
intuitionsabout vision
DANIEL T. LEVIN
VanderbiltUniversity
BONNIE L. ANGELONE
RowanUniversity
Recent research has revealed a series of striking limits to visual perception. One
importantaspect of thesedemonstrations is thedegree towhich theyconflictwith
intuition;people often believe that theywill be able to see things thatexperiments
demonstrate theycannot see. This metacognitive error has been explored with
reference to a few specific visual limits,but no studyhas yet explored people's
intuitions about vision more generally. In this article we present the results of
a broad surveyof these intuitions.Results replicate previous overestimates and
underestimates of visual performance and document new misestimates of per
formance in tasks that assess inattention blindness and visual knowledge. We
also completed an initial exploratory factor analysis of the items and found that
estimates of visual performance forwell-structured information tend to covary.
These results represent an important initial step in organizing the intuitions that
may prove important in a varietyof settings, including performance of complex
visual tasks,evaluation of others people's visual experience, and even the teach
ing of psychology.
Recent research exploring visual perception has revealed a range of strik

ing failures of visual awareness inwhich salient events often go entirely
unnoticed (for a review, see Rensink, 2002). A key element of these failures
is the degree
towhich
own visual capabilities.
they conflict with people's intuitions about their

a range of experiments have docu
For example,
mented thatpeople have considerabledifficulty

detectingbetween-view
visualchanges.This phenomenon,called changeblindness,isextremely
counterintuitive: Not only do people express disbelief when told about
change blindness, but when they are asked to predict their ability to de
tect visual changes, they grossly overpredict their performance in a wide
range of circumstances
(see Levin & Beck, 2004, for a review). These
mispredictions have a number of implications ranging from poor control
of visual performance tomisconstruals of other people's capabilities and
AMERICANJOURNAL OF PSYCHOLOGY
Fall 2008,Vol. 121,No. 3, pp. 451-472
i 2008 by theBoard ofTrusteesof theUniversityof Illinois
LEVIN
452
experience.
Although much
of our research
&
ANGELONE
in this area has focused on
about change detection, little research has explored people's

beliefs about visual perception more generally. In this article we broaden
our exploration of intuitions about vision and report the results of an
initial application of a broad questionnaire,
the Visual Metacognition
predictions
Questionnaire
(VMQ),
of visual phenomena.
Visual
One
assessing people's
limits and misunderstandings

of the main
tion has been
organizing
the exploration
understanding
of a wide range
of limits
principles for research on visual percep

of visual limits. From the start of scientific
psychology, researchers have noted that we are not aware of all possible
that the act of at
stimuli in the external world and have hypothesized
tending to a subset of that information is crucial to becoming aware of
it.Recently, this kind of research has explored the perceptual limits im
posed by attention in a particularly compelling set of tasks,many ofwhich
take advantage of recent advances in digital imaging thatmake possible
the use of naturalistic stimuli (Hollingworth, 2003; Rensink, O'Regan, &
Clark, 1997; Simons & Chabris, 1999). Not only do these developments
allow for new tests of the role of attention in realistic settings, but they
also bring research on attention into the realm of familiar experience and
have the potential
to directly challenge
intuitions about vision. It is quite
difficult to connect naive beliefs about vision to rarefied stimuli and tasks
that subjects have little experience with and few expectations about. On
the other hand, as stimuli and tasks become more realistic, people begin
to have expectations about their performance, and these expectations
might end up being wrong. Here, we review recent research exploring
that demonstrate
visual limits, with particular emphasis on paradigms
these limitswithout the use of complex tasks and unfamiliar stimuli, and
called the VMQ.
present a broad measure of visual metacognitions
have demonstrated
the limits
Several recently explored phenomena
imposed
by visual attention. Change
blindness,
inattention blindness,
repetition blindness, and the attentional blink all constitute large failures
of visual awareness
(Raymond, Shapiro, & Arnell, 1992; Simons & Levin,

1997; Rensink, 2002; Mack & Rock, 1998). We focus on change blindness
and inattention blindness here because these are themost accessible to
intuition. In the case of change blindness, subjects have great difficulty
detecting the difference between two versions of a scene. Change blind
ness has been observed in a wide variety of situations, ranging from still
images tomovies to real-world interactions (for a review, see Simons &
Levin, 1997, and Rensink, 2002). It can even occur while people are at
tending to the changing object. For example, Levin and Simons (1997)
created short videos inwhich the sole actor changed into another person
VISUAL
METACOGNITION
QUESTIONNAIRE
453
(wearing different clothing) across shots and found that up to two thirds of
subjectsmissed thechange (see alsoAngelone, Levin,& Simons,2003).

A closely relatedphenomenon, inattentionblindness, occurswhen
subjects'attentionisdirected towardone location,object,or eventwhile
anotherstimulusispresented.The first
blindness
exampleof inattention
invisionwasNeisser andBecklen (1975),who showedsubjectsshortfilms
and asked them to attend to a group of people playing a basketball-passing
game. While subjects were attending to the game, they failed to detect a
woman
carrying an umbrella walking among the players (formore recent

examples, see Simons & Chabris, 1999; Wayand, Levin, & Varakin, 2005).
In another well-known paradigm, Mack and Rock (1998) asked subjects to
discriminate
the relative lengths of the two bars in a cross.While
theywere
focused on this task, another stimulus appeared on one of the quadrants

defined by the cross. Subjects often missed this stimulus, even when itap
right at the fixation point (so they fixated gaze on the center of the
screen while they attended to the cross, which was offset from center).
peared
Both inattention
blindnessand changeblindnessrepresentsituationsin
which attentional selection of a subset of objects within a scene, or a subset
of features within an object, can leave us almost completely oblivious to
other aspectsof thescene.Although generallyconsistent

with thescien
tificunderstandingof attention,thesedemonstrationsare particularly
compelling, and a large part of their value may be the degree towhich
they conflict with strong intuitions (Simons & Levin, 2003; Reisberg &
McLean,
1985). Indeed, we have consistently observed that they vastly
overpredictchange detection,whether theyare predictingtheirown or
others' performance in a wide variety of situations (for a review, see Levin

& Beck, 2004). For example, 90% of subjects predicted theywould see
of a large red scarf thatwas actually detected by 0% of
the disappearance
subjectsinLevin and Simons's (1997) originalexperiment.
Although little research has explored adults' beliefs about vision (for
an exception, see Harley, Carlsen, & Loftus, 2004), a number of studies
have explored children's knowledge about their perceptual abilities (for a
review,see Flavell,2004).Although thisresearchsuggestsa solid founda

tion for adult reasoning about vision, other studies suggest that children
about vision that are retained into adulthood (Winer
have misconceptions
& Cottrell, 2004),
broad
leaving open the possibility that adults misconstrue

range of visual and cognitive phenomena.
Summaryand hypotheses
To get a more general sense of people's intuitions about vision, we have
theVMQ. Three basic questions drove the development of
been developing
theVMQ.
First,we wanted
to know whether
itwould be possible
to observe
LEVIN
454
overestimates of visual performance
&
ANGELONE
similar to those observed for change
detection, especially for situations involving natural real-world stimuli.

Therefore, we asked subjects to predict their performance in situations
similar to several well-known
attention and memory.
experiments
that illustrate limits in visual
It is important to note that these performance
compari
predictioncomparisonsnecessarilyinvolvebetween-experiment
sons that do not equate specificmethods
or subject populations. Therefore,
we generally focus on large differences (in the range of 30-80%

between
performance
and predictions).
we were
Second,
divergence
interested in
correlations between estimates of visual and nonvisual performance.
This
ismotivated by two hypotheses. The firstintercorrelation hypoth

esis is that overestimates of visual performance occur in part because the
question
meaningful
structure inherent to natural events, scenes, and objects leads
subjects to believe
that theywill be able to countenance
of visual information. Therefore,

well-structured
overestimates
a large amount
should be strongest for
stimuli, and intercorrelations would
be strongest among
scenarios describing structured stimuli. A key assumption underlying this

hypothesis isone's definition of structure. For individual scenes, we follow
research exploring scene perception and define this structure in terms of
meaningful interobject and scene-object relationships. So themeaningful
thematic and causal relationships between objects constitute part of this
definition, and scene-object relationships such as support, interposition,
and familiar location, familiar size, and semantic scene-object
relation
ships constitute another part (see Biederman, Mezzanotte, & Rabinowitz,
1982). In addition, we will be exploring people's beliefs about their ability
to remember sets of scenes, so our notion of structure should include the
contrast between sets of scenes that are less structured because they are
unrelated and sets that are more structured because they illustrate a single
coherent narrative.
The second intercorrelation hypothesis is that these intercorrelations
it is important
should be limited to questions about visual phenomena;
to know whether one set of beliefs drives reasoning about all structured
information or whether these beliefs are specific to vision. Finally, we
interested inwhether beliefs about visual capacity correlated with
were
other kinds of visual knowledge
and misconceptions.
We
therefore asked
subjects about depth perception, the basic relationship between light and
seeing, and visual search. In the section that follows, we describe each
section of the VMQ.
The VMQ
Items on theVMQ ask subjects to predict their own visual performance
in 10 different task categories (see Appendix). Selection of these categories
was based on recent research in visual attention and awareness and on
VISUAL
METACOGNITION
QUESTIONNAIRE
455
texts in cognition and perception. One critical criterion for

selecting topics was that the experiments demonstrating them had to be
straightforward and easily explained. In addition, we avoided experiments
undergraduate
that produced
highly unnatural perceptual
that the situations
in the VMQ
experiences.
are all common
rather that they do not require subjects tomake

ences when making
predictions
about
This isnot to say
or entirely natural, but

a large number of infer
their performance
in the rarified
sometimes necessary for perception experiments. Many of

these questions are very closely based on actual experiments thatdocument
limits in visual perception and visual memory, but others are less directly
circumstances
based on specific research findings and are focused on more general no

tions such as attentional limits.We should therefore be clear thatwe do not
consider
theVMQ
to be an exhaustive
should be considered
test of visual intuitions. Instead, it
a broad sampling of visual intuition for situations in
which subjects might reasonably base their responses on everyday experi
ence.
Note
that the order of questions
order of questions
1. Change
in theVMQ
does not follow the precise
in theAppendix.
detection.
Five questions
ask subjects whether theywould

detect unexpected visual changes to complex scenes. Four of these ques
tions ask about change detection in natural scenes, and a fifthasks about
change detection for an array of individual objects. Three of these ques
tions are associated with a known empirical base rate of actual change
detection, and all are illustrated with color images depicting the changes.
For each question, subjects first read a brief description of the circum
the change, then they see
stances under which theywould encounter
the change illustrated on the next page. For three of the questions (1.1,
1.2, and 1.3), subjects are told to imagine that they are watching a movie
inwhich a change occurs between cuts (see Figure 1 for an example).
Subjects are also told to assume that they are not on the lookout for
changes. Subjects indicate on a 4-point scale whether they definitely or
notice the change or definitely or probably would not
notice the change. Three additional questions are included in this sec
tion. One
(1.4) asks whether subjects think theywill see a change to an
object array in incidental task (as is the case for the first three), and two
probably would
(1.5 and 1.6) ask whether subjects will detect a change for one normal
and one jumbled scene under intentional circumstances
(i.e., they are
told to imagine they are looking for a change). One key feature of these
is that they are designed to ask
questions, and most others on the VMQ
about cognitive limitswithout allowing the subject to actually experience
them. In this case subjects are told about the specific visual changes they
will be encountering and therefore cannot actually experience change
blindness.
456
LEVIN
&
ANGELONE
Figure 1.Example change detection question. Each panel representsone of three

successivelyviewed slides
2. Inattention blindness. The questions in Section 2 assess subjects'
beliefs about the degree towhich people can detect specific objects and
events when their attention isnot focused on them.Question 2.1 describes
the Simons and Chabris (1999) experiment and asks subjects whether they
would be likely to notice the unexpected gorilla. Questions
2.2 and 2.3
ask subjects whether theywould notice a pedestrian sign while they are
driving and engaged in a complex verbal task (2.2) or a complex visual
task (2.3). Finally,Question 2.4 asks what proportion of pilots would see an
airplane on the runway they are about to land on while using a heads-up
display (based on Haines, 1991, who showed that pilots sometimes miss
such stimuli). This question departs from our usual technique of asking
subjects whether they themselves would detect something because we
assumed thatmost of our subjects are not pilots.
3. Visual attention. Three questions ask subjects about visual atten
tion. The first two are based on developmental
research arguing that
young children falsely believe that looking at one thing in one location
also allows them to see other things in other locations (Flavell, Green, &
Flavell, 1995). Flavell et al. argue that young children believe attention
ismore like a lamp than a spotlight in that orienting one's attention to
a scene illuminates all of it instead of some small part of it. In contrast
to the implicit assumption that adults really do understand the spotlight

model of visual attention, however, we have observed variability among
adults in their commitmnent to the idea that looking at one thing precludes
seeing another thing (Levin, 2001). The first visual attention question
(3.1) asks subjects to imagine they are looking at a painting in a museum,
then to indicate whether theywould see the painting's frame while they
are looking at the painting. The second (3.2) asks subjects whether they
would see a fire hydrant 10 feet away from a friend they are looking at
from across the street.A third visual attention question (3.3) asks subjects
what proportion of the objects in a scene they typically attend to.
Subjects are also asked whether attention isgenerally necessary todetect
changes (3.4) and for remembering information (3.5). These questions
are based on research suggesting that children do not fully comprehend
VISUAL
METACOGNITION
QUESTIONNAIRE
457
thenecessity
ofattendingto information
forthepurposesof remembering
it (Miller & Weiss, 1982), and again we suspect that there may be diversity
among adults despite theassumption in thedevelopmental literature

that theyunderstand this.The question about the impactof attention
on change detection
is a generalization
of this, and in the past we have
observedvariabilityin intuitions
on thispoint (Levin,Drivdahl,Momen,
& Beck, 2002; Levin,Momen, Drivdahl,& Simons,2000).
4. Auditory
generally
attention. We
about
included
a pair of questions
the influence of information outside
asking more
the focus of at
tentionby assessing intuitionsabout auditoryattention.Although the

focus of the VMQ
is clearly visual, we have included
this scenario
to al
lowmore general theoreticalunderstandingofwhether some beliefs

about attention are specifically visual or are more
describe
a cocktail party inwhich
general. These
items
subjects imagine they are conversing
with one personwhile ignoringother nearby conversations.Subjects

indicatewhether theywould understand themeaning of the ignored
conversation
(4.1) and whether
theywould
mentioned in theother conversation(4.2).

5. Visual memory. These
questions
hear
their name
ask subjects about
if itwere
their ability to
remember visual information. One
set of three questions
is based on re
that people
are surprisingly good
at recognizing
search demonstrating
largenumbersof picturesof real-world

objectsand scenes.This research
used a typical recognition memory protocol inwhich subjects were first
shown an inspection set consisting of a wide variety of pictures ranging
from individual objects to natural scenes, then were shown a test set in
which half of the pictures were repeats from the inspection set and half
were new. Several experiments converge in demonstrating that recognition
memory is good for inspection sets of up to 10,000 pictures (Standing,
1973; Standing,Conezio, & Haber, 1970). In theVMQ picturememory
scenario, the task is described to subjects very concretely, and they are
shown 3-item sample inspection and test sets. Subjects respond to three
questions asking them to predict their performance with 20-, 50-, and
1,000-iteminspectionssets (5.1,5.2, and 5.3). Three additionalscenarios

testbeliefs about visual memory. Two of these ask subjects to estimate their
success on a picture memory task characterized by a more organized in
spection set: They are asked what proportion of the scenes in a movie they
if theywere (5.5) or were not (5.4) intentionally trying
the information. A final question asks subjects to imagine
that they had witnessed an automobile accident and that theywere asked
1month laterwhether some specific person had also been at the scene of
would
remember
to remember
theaccident (5.6).
6. Visual knowledge. This question assesses subjects' beliefs about the

towhich they could reproduce in a drawing the details from a
degree
LEVIN
458
familiar object. Based
on the work of Nickerson
&
and Adams
ANGELONE
(1979), who
the details of
showed that subjects have difficulty accurately reproducing

a Lincoln penny, these questions ask subjects to imagine that theywere
asked to draw a penny and to predict whether
theywould
remember and
correcfly place a series of features on their drawing. We predict that subjects

will overestimate their success. This is similar to overestimates of change
detection
inwhich well-structured visual information may lead subjects to
overestimate
anecdotal
the depth of their representation,
All questions are accompanied

placement
and there is at least one
report that thiswill be the case (Keil, Rozenblit, & Mills, 2004).
by a picture of a penny showing the correct
of the features. Question
6.1 asks subjects to predict whether
the head facing in the correct direction, Question

6.2
asks whether theywould remember to put the word Liberty on the coin,
Question 6.3 asks whether theywould remember to include and correctly
theywould
place
place
the phrase United States ofAmerica, and Question
6.4 asks whether
theywould place the date in the correct location.

7. Icon and subitization. These questions are based on research demon
strating that briefly presented visual information can persist momentarily
in an afterimage, or "icon." In themost well-known demonstrations
phenomenon,
subjects view arrays containing
9-12
of this
letters for 50 ms and
these conditions,
then report as many of the letters as possible. Under
subjects can generally report only 4 or 5 letters. However, researchers no
ticed that subjects sometimes commented that theywould have been able
to report more letterswere itnot for the time it took to verbalize the first
few; they could "see" all of the letters for a moment, and ifonly they could
get the information into a more durable form, theywould have been able
to report many of them, a claim that has been empirically verified (see
Averbach & Sperling, 1968, for a review). The icon is a fairly compelling
phenomenal
experience, and early authors such asJames (1892) not only
described itbut also intuited its use as a short-term storage system.
Therefore, we have asked subjects to predict how theywould perform
under these circumstances. Question 7.1 asks subjects to estimate the num
ber of items they could see in a 50-ms display, Question 7.2 asks whether
they could briefly "see" the array after ithad disappeared from the screen,
and Question 7.3 asks subjects to predict whether they "can retain more
information" than they could report.
8. Visual search. As children begin to understand other people's visual
they begin to understand at least some of the factors that
things easy and difficult to locate in a visual search task (Miller &
Bigi, 1977). To explore these intuitions in adults, we included a series of
questions asking subjects about the conditions thatmight produce a serial
or parallel visual search (Treisman & Gelade, 1980). For the firstof these
questions, subjects are asked to imagine a search task inwhich theirjob is
experience,
make
VISUAL
METACOGNITION
459
QUESTIONNAIRE
to locate a specific target (a green vertical line) among a number of other

objects (red horizontal and vertical lines). In this case the target isunique
in terms of a basic visual feature: color. Therefore,
visual search should
be very efficient, and the time necessary to locate the target should not
increase if additional distractors are added to the display. In contrast, a
second question
asks about a more difficult search. In this case, the same
target (red vertical line) must now be located among a mixed set of red
horizontal and green horizontal and vertical lines. Color is no longer
diagnostic
of the target, and subjects must search for a conjunction of

color and line orientation. In this situation, the search ismore difficult,
and reaction time to locate the target increases with added distractors. In
both cases, subjects are firstasked tojudge whether the search will be "Very
fast," "Somewhat fast," "Somewhat slow," or "Very slow" when the target
is hidden among
fivedistractors (Question
8.2 for the conjunction

difficult itwill be when
8.1 for the feature search and
search). Then, they are asked how much more

the target ishidden among 300 distractors (Ques
tion 8.3 for the feature search and 8.4 for the conjunction
search).
9. Extramissionbeliefs.Research byWiner and colleaguesdemonstrat
ed that people
can misunderstand
for seeing (for a review, seeWiner
one of the most basic requirements

& Cottrell, 2004). These
experiments
consistently
demonstratedthatapproximately30% of adultsendorse an
extramissionist account of vision, assuming that something (such as "rays")
comes out of the eyes to allow seeing. Therefore, the VMQ includes an
extramission question modeled on thework ofWiner, Cottrell, Karefilaki,
and Chronister (1996). This itemaskswhether people see by emitting

something from their eyes, by having something enter their eyes, or by
both processes (9.1). This verbal description is followed by a simple il
lustration of these processes very similar to that used byWiner et al.
10.Depth perception.Althoughdepthperceptionusuallyisconsidered
to be automatic, recent research demonstrates that in at least some cases,

accurate judgments of the size of distant objects depend on children's
ability to verbalize the relationship between apparent size and distance
the VMQ includes several questions that
(Granrud, 2004). Therefore,
assess people's understanding of depth. In one (10.2), subjects are asked
why the moon
looks larger on the horizon, and in another (10.1), they

are asked how theywould make a fish they caught look larger in a photo
graph. Finally, in another question (10.3) observers are asked to place a
block on a picture of a set of railroad tracks that converge
such that it appears very large.
in the distance
Familiarit questions
Because we tested subjects in general psychology classes, and because
much of the research included in theVMQ has received a fair amount of
LEVIN
460
publicity, we asked subjects whether
they had heard about
&
ANGELONE
research on
each of the topic areas tested in class, on television, or elsewhere.
In some
cases, a topic area was not tested because either there was no research
testing the relevant hypothesis or because the questions were general and
not closelyrelated to any specificresearchfinding.The VMQ included

with researchon changedetection,inattention
questionsabout familiarity
blindness,auditoryattention,visualmemory,visual knowledge, iconic
memory,visual search,extramission,
and depth perception.
measures
Demographic and general intelligence
In thefinalsectionof theVMQ we collectedbasic demographic infor
mation
such as age, sex, and
level of education.
In addition, we asked
academic testscores,includingSAT andACT scores,

subjectstoself-report
and academic
performance
in the form of current grade point averages.
Finally, we asked subjects to list their current major and to list the psychol
ogy courses
they have taken.
EXPERIMENT
METHOD
Subjects
A totalof 108 Kent State University students (68 women, 40 men) completed
theVMQ in exchange for course credit in theirgeneral psychology class. Their
mean age was 21.48 years (range 18-44).
Procedure
The VMQwas administered on computer as a PowerPoint presentation to small
groups of subjects. The experimenter read each question and gave subjects time
to respond on Scantron bubble sheets before proceeding to thenext slide. It took
approximately 45 minutes to complete the questionnaire, afterwhich subjects
were debriefed.
RESULTS
This section isdivided into threeparts,correspondingto the specific
hypotheses presented earlier. First,we describe a factor analysis of sections
in theVMQ thatassesses the intercorrelation

hypotheses.As discussed
wewere interestedinwhetherresponsesregarding
well-structured
earlier,
about vision per se

stimuli would cohere and whether metacognitions
would cohere. Then, we discuss results from each of the sections on the
VMQ, organized by the groupings suggested in the factor analysis. Within
these sections, we pay particular attention to situations inwhich subjects
appear
to have overestimated
or underestimated
their performance.
VISUAL
461
QUESTIONNAIRE
METACOGNITION
Exploratoryfactoranalysis
To
towhich
test the degree
different kinds of metacognitions
were
VMQ
interrelated,11 summaryscoreswere generated for thedifferent
sections.Two of theserepresentedthevisualmemory section,and the
other sections were represented by one score each. We chose
to use sum
mary scoresbecause enteringeach of the individualquestions into the

analysis would
have produced
too many
items relative to the number of
subjectswho completed thequestionnaire (seeBryant& Yarnold, 1995).

The first
summaryscore,labeled "visual
memory1" includedresponsesto
memory forunstructured
Questions 5.1-5.3 and was intendedto reflect
materials. The other summary score, labeled "visual memory 2," included
memoryforwell-structured,
Questions 5.4-5.6 andwas intendedtoreflect
meaningful materials. The other nine scores reflected themean

on each of the other nine sections.
The
results of this analysis provide
some preliminary
response
support for the
bothof theintercorrelation
hypotheses;responsesabout structuredstim
uli correlate, as do responses about visual stimuli. The analysis extracted
threefactors
usingprincipalcomponentanalysisthatcollectively
explained
48.75% ofvariance.
The firstfactor included five scores to explain 24.5% of variance. (Scores
to factors when they had a loading of .4 or above. In cases
were assigned
a score's loading was above .4 for rmore than one factor, itwas as
signed to the factor it loaded most highly orn.) It included change detection,
visual memory 2 (structured), visual attention, inattentional blindness,
and visual knowledge. This factor seems co represent subjects' estimates of
where
visual capacity for structured materials. Almost all of the tasks it includes
involve meaningful real-world stimuli, and they are all visual. Tasks that
involvedlessstructured
visualmaterials (such as thosedescribed in the
visual memory 1 scenarios) and those describing nonvisual attention (e.g.,

the cocktail party scenarios) did not load most strongly onto this factor.
The second and third factors were more difficult to interpret. The sec
ond factor explained 13.9% of variance by including responses to icon,
extramission, and depth perception
items, and therefore it appears to
more qualitativeknowledgeabout vision.The lastfactorincluded

reflect
1 (unstructured), auditory attention, and visual search to

explain a total of 10.3% of variance (see Table 1 for the component matrix
visual memory
forthesethreefactors).
Item results: Factor
Change detection.As inpreviousresearch,subjectsoverestimatedtheir

ability to detect between-view visual changes.
In the situation described
inQuestion 1.1 (jerseyand ball change), 12.3% of subjectsdetected the

change
(Angelone,
Levin, & Simons,
2003), whereas
87% claimed
they
LEVIN
462
&
ANGELONE
Table 1.Component matrix for exploratory factor analysis

Component
Variable
Visual memory 1 (unstructured)
Change detection
Visual memory 2 (structured)
Visual attention
Auditory attention
Icon and subitization
Visual search
Extramission
Inattention blindness
Visual knowledge
Depth perception
1
.379
-.701a
.657a
.692a
.252
.376
-.416
_.399
.568a
.539a
-.060
2
.417
.038
.150
.193
.367
.562a
.363
.431a
-.319
-.235
.606a
3
-.616a
.026
.046
.227
.475a
.136
.451a
-.027
.350
-.266
-.245
aScore assigned to factor represented in thatcolumn.

would
definitely or probably detect the change, X2(2) = 87.81, p < .001.

1.2 (actor change) and 1.3 (plate change) the known base
For questions
(Levin & Simons, 1997), as compared with 58.3% and 37.9%

success in the VMQ
X2(2) = 12.51, p< .001 and X2(2) = 5.81,
p < .025, respectively. There is no performance baseline for the stimulus
1.4 (array change) but Beck, Levin, and Angelone
in Question
(2007)
observed approximately 10% change detection in an incidental taskwith
10-object scenes. This compares with 67.6% predicted success. There are
rate is 0%
predicting
no baselines for the scenes inQuestions 1.5 and 1.6 (normal andjumbled
scenes), but 87.9% and 83.3% of subjects predicted success, respectively.
However, comparing the mean ratings on these questions, subjects pre
dicted significantly less success on the jumbled scene, M= 3.157, than on
the normal scenes, M = 3.444, t(107) = -4.120, p < .001, suggesting that
they account for scene structure when it ismade obvious.
Inattention blindness. As in the change detection questions, subjects
overpredicted success in a situation inwhich they are attending to one
thing while another unusual event occurs. Simons and Chabris (1999)
found that 42% of subjects detected the unexpected gorilla in approxi
mately the conditions we described, as compared with 87.9% of subjects
predicting success in this experiment, X2(2) = 16.67, p < .001.When think
ing about verbal and spatial referents, respectively, 60.1% and 47.2%
of subjects indicated that theywould detect the pedestrian sign, giving
mean ratings of 2.75 for the verbal condition and 2.50 for the spatial
condition, t(107) = 4.73, p < .001. The pilot scenario (Question 2.4) pro
duced a range of responses, 63.8% of which were 15% or less. In fact, 2
of 9 (22%) pilots failed to see the plane, so it appears thatmost subject
overestimate. However, because there were so few subjects in the original
VISUAL
METACOGNITION
QUESTIONNAIRE
463
experiment,a testcomparingsubjects'estimates
withbaselinewould be
underpowered.
Visual attention. The two attentional breadth questions produced a
range of intuitions. On Question 3.1, 62% of subjects indicated theywould
see the frame a painting is in, and 38.9% indicated theywould see the fire
hydrant(Question 3.2). On Question 3.3, subjectsindicatedthey

would
look at an average of 58.1%
On Question
3.4,
to
see chang
65.7% of subjects indicated that visual attention "helps a lot"
es, and 23.1%
indicated
of objects
(SD=
19.5%).
that it "helps a little." On Question
3.5, 29.6%
indicatedthatattentionis"absolutely
necessary"torememberthings,
and
62.9% indicated that it "helps a lot." Generally, subjects rated attention
as being more necessary formemory than for perception, t(107) = 6.086,
p < .001. Even so, 40.9% of subjects indicated that attention was just as
important formemory as for perception, and 9.3% indicated that atten
important for change detection than formemory.
Visual memory 2 (structured). In contrast to the large underestimates
of performance observed for unstructured picture sets, subjects appear to
tion was more
havepredictedmore successwithmovies,indicatingthey
would remember
an average of 57.2%
of scenes when
theywere not trying to remember
toremember(Question
were trying
(Question5.4) and 72.6%when they
5.5), t(l07) = 9.991, p < .001. In addition, only 16.6% of subjects indicated
that theywould remember equivalent amounts of information whether
intendingto rememberor not.On Question 5.6, 34.3% of subjectsindi

cated theywould
recognize a person
in a crowd.
Visual knowledge.Subjectsgenerallyoverestimatedthedegree towhich

they could reproduce the features of a penny on a recall test.On these
questions, 85.2% of subjects predicted they could correctly reproduce
the direction of Lincoln's head, whereas only 50% of Nickerson and Ad
ams's (1979) subjects got this correct, X2(2) = 12.90, p < .001. Furthermore,
39.8% indicated they could correctly reproduce Liberty (compared with
5% baseline), X2(2) = 9.066, p < .01; 71.3% indicated they could correctly
reproduce the phrase United States ofAmerica (compared with 20% base
line), X2(2) = 19.10, p < .001; and 77.7% indicated theywould place the
date correctly (compared with 60% baseline),
2(2) = 2.844, p < .10.
Item results: Factor 2
Icon and subitization. Subjects predicted they could see an average
of 3.17 letters flashed on the screen for 1/20 s. This compares with an
average of 4.5 letters for trained subjects (Averbach & Sperling, 1968).
Furthermore,69.4% of subjectsindicatedthey
would see an afterimage,
and 61.1% indicated that theymight be unable to report all the informa
tion they could see.
Extramission. Although only a few subjects gave the extramission-only
LEVIN
464
&
ANGELONE
(3.7%), whereby seeing involves the eyes producing rays,many

(48.1%) indicated that seeing involves both the intromission and the extra
mission of rays from the eye. This is consistent with the results of Cottrell
(1994), who observed 4.2% endorsing the extramission-only
and Winer
response
response and 34% endorsing

Depth
perception.
themixed
response.
Subjects varied widely in the degree
towhich
they
of depth per
consistent with a correct understanding
ception. Most subjects (82.4%) correctly indicated that an object would
appear larger ifheld closer to a camera (Question 10.1). The other two
questions were more difficult. Only 45.4% of subjects correctly indicated
gave responses
it appears to be
that the moon appears large on the horizon because
near distant objects (Question 10.2), and only 48.1% indicated that the
object would appear the largest if itwas placed at the top of the railroad
track picture
(Question
10.3).
Item results: Factor 3

attention. More subjects indicated that itwould be possible
in a neighboring conversation than themeaning of
the conversation. However, 49% indicated that theywould probably or
definitely hear themeaning of the other group's conversation. Clearly,
Auditory
to hear one's name
incorrect, but on some interpretations of the question,

ifone assumes that in a typical setting one might
briefly stop attending to the current conversation to quickly sample an
this is nominally
itmight be reasonable
other.
Visual memory 1 (unstructured). As in previous experiments, subjects

their ability to remember large sets of unrelated
greatly underestimated
pictures. Ifone assumes that 90% correct is a reasonable estimate based on
previous research (the closest comparison isNickerson, 1965), then 78.7%
their performance for 20 pictures (Question
of subjects underestimated
their performance for 50 pictures (Question
5.1), 94.4% underestimated
for 1,000 pictures
their performance
5.2), and 94.4% underestimated
(Question 5.3). In fact, 51.8% of subjects essentially indicated theywould
be guessing for 1,000 pictures.
Visual search. Most subjects recognized that the feature search would be
easy and that the conjunction search would be more difficult: 88.9% indi
cated that the feature search would be "very fast," compared with 29.9%
thought so for the conjunction search. It is important to note that
the 64.8% of subjects who indicated that the conjunction search would be
"somewhat fast"were not necessarily incorrect because the example they
were responding to had only 5 distractors. The key is thatmost subjects
who
knew to differentiate the two. Furthermore, 78.7% of subjects indicated

that the conjunction search would be slowed by distractors more than the
feature search, and 0% of subjects indicated that the feature search would
VISUAL
METACOGNITION
465
QUESTIONNAIRE
be more slowed. In addition,many subjects (67.5%) indicated thatthe

feature search would
be the same or fasterwith the addition
of a large
numberof distractors,
and 87.9% indicatedthattheconjunctionsearch
be slower with more
would
distractors. However,
68.4%
of subjects who
indicatedthattheconjunctionsearchwould be slowedgave the"somewhat

slower"response.
Familiarityresponses
On average, 53.5% of subjects reported having heard about the relevant
questions
research(range,18-76%). Overall, responsesto thefamiliarity
were not associatedwith responsesto themetacognitivequestions.In one
case, there was a trend for responses
toQuestion
7.3 to be associated with
havingheard about researchexploringperceptionof brieflypresented

more
displays.Subjectswho had heard about this(72.2%) were slightly
likely to report that theywould retain more
report, t(106) = 1.801, p= .075.
information than they could
withdemographicquestions
Global correlations
To testwhether global achievement or general experience in psychol
ogy affected responses on the VMQ, we ran a series of regressions using
self-reported achievement test scores, grade point average, and number
of psychology courses as predictors for each of the 11 section scores (sum
marized
in Table 2). This analysis suggests that psychology classes reduced
degree. In
changedetectionoveroptimismtoamarginallynonsignificant
addition,
subjects with high grade point averages and achievement
test
scores endorsed less expansive views about visual attention. High achieve
ment also predicted more optimistic estimates of visual knowledge and
More surprisingly,
response.
lowerlikelihoodofgivingan extramissionist
high achievement scores were associated with less accuracy on the visual
search questions. Finally, subjects who had taken more psychology courses
were more likely to endorse limits on auditory attention.
DISCUSSION
This initial administration of theVMQ has produced a range of findings,
and to focus the discussion we return to the three hypotheses mentioned
earlier. First,we were interested in exploring whether itwould be possible
to observe other large overestimates of visual performance in addition to
overestimates of change detection. In thiswe were successful for the inat
tention blindness and visual knowledge items. In the case of inattention
blindness, the difference between estimated performance and the base
rate was 45.9%, an overestimate of a magnitude
similar to that observed
here for change detection (which ranged from 37.9% to 74.7%). In the
LEVIN
466
&
ANGELONE
Table 2. Regressions using achievement, GPA, and coursework to predict VMQ

subsection scores (n = 94)
Ach Test
Subsection
Change detection
Visual memory 2 (structured)
Visual attention
Inattention blindness
Visual knowledge
Icon and subitization
Extramission
Depth perception
Visual memory 1 (unstructured)
Auditory attention
Visual search
GPA
NCourses
-.184#
-.255*
-.213#
.217*
-.241 *
-.243*
-.261*
Note. Entries are beta values for each significantor near-significantpredictor in

multiple linear regressions predicting responses on each of the 11VMQ sections.
GPA = grade point average; VMQ = Visual Metacognition Questionnaire.
*p <
.05. #p <
.10.
case ofmemory
for a common
object, subjects again overestimated
by up
to 50%.
The factor analysis produced only one strong factor but was interesting
for two reasons. The first factor demonstrates a moderate
nonetheless
relationship between the questions asking subjects about their ability to
process well-structured visual information. It is interesting to note that this
factor includes responses to questions about memory for the structured
information inmovies (in which all of the individual scenes are related by
a narrative) but not the questions about memory for the less structured
information. Thus it appears that the difference between processing of
structured and unstructured information ismore salient to subjects than
the difference between long-term memory and themore immediate kinds
of memory inherent to change detection. In addition, the fact that the
inattention blindness and visual attention scenarios also loaded on this
factor suggests that these beliefs are broad and might be summarized as
can effectively process and monitor meaningful visual
information. This is consistent with previous findings (Levin et al., 2002)
that subjects predict change detection success even when the prechange
a belief that people
and postchange displays are described as being separated in time by up to

an hour. One would think that such a scenario would clearly invoke the
need formemory for visual properties, but it did not. Subjects not only
demonstrated overestimates of change detection equivalent to no-delay
the concept of memory in response
scenarios but also rarelymentioned
justifications (in contrast to a scenario asking about memory
for unrelated
digits).
VISUAL
METACOGNITION
QUESTIONNAIRE
467
Also consistentwith thehypothesisthatsubjectsdraw on heuristics

specificallyrelated tovision is that thepredictionsabout nonvisual at
tention(e.g., thecocktailpartyscenario) did not correlatestrongly
with
Factor 1.Although
is unclear,
the degree
this is a potentially
towhich a cocktail party is trulystructured

important finding because
it suggests that
intuitions
about visionare distinctandmay exclude intuitionsrelatedto
other senses. This
is important because many explanations of the coun

rest on arguments about the im
terintuitive nature of change blindness

mediate
accessibility of the visual world
as opposed
overestimate of awareness across all modalities
to a more
general
(see Levin & Beck, 2004;
Rensink,2000). It isalso important

because itconstitutes
further
evidence
that overestimates of visual awareness cannot be attributed solely tomore
general cognitive heuristics such as a hindsight bias.
Limits and potential uses for theVMQ
It is important to note that at this point theVMQ
should be considered
an experimental
survey, not a fully validated instrument. We have not

tested the degree towhich ithas external validity and have not tested its
test-retest reliability (in fact,we know of no case inwhich this has been
testedformetacognitivejudgmentsinadults).
Another
important issue is thatwe tested theVMQ
on a population
that
has had some academic experience with psychology. Therefore, parts of

our current administration might reflect a combination of formal learning
and intuition. However, it is important to note that reported psychology
class experience and familiarity with specific sections had only a small
impact on a few of the subsections. Therefore,
intuitions about visual
experience may resist change. For example, it has proven quite difficult
to eliminate extramissionist responses with anything but themost direct
and concrete educational interventions (seeWiner & Cottrell, 2004, for a
review), and at least some other misestimates
are not affected bywarnings

(see Harley et al., 2004).
Our undergraduate population may not represent the general popula
tion in other ways as well. For example, we tested predominantly young
adults, who may be more optimistic about their abilities than older adults.
However, we collected some metacognitive data recently from a sample of a
range of hospital employees. Most of these were not full-time students, and
theywere much older than the students (mean age, 42; n = 28). Although
the specific testconditions were different,we did ask them variants of three
change detection questions thatwere on the VMQ, and their responses
were very similar to those of the college students in the current sample:
85% of the hospital employees
indicated
theywould detect the shirt and
basketball change (compared with 87% of the current college-age subjects),

58% indicated theywould detect the actor change (compared with 53% of
LEVIN
468
current subjects), and 32% indicated
&
ANGELONE
theywould detect the plate change
(compared with 38% of current subjects).

Potential
applications
for theVMQ
ranging from research on the link be

to research on the
and visual performance,
tool. As a tool for assessing
beliefs of jurors, to its use as a pedagogical
We
envision uses for the VMQ
tween visual metacognition
and performance,
the relationship between metacognition
provide an initial, broad test for a series ofmisunderstandings
theVMQ can
about vision
and visual attention thatmight reveal performance-limiting blind spots. In

addition, itmight be possible to reveal individual limits on performance if
theVMQ proves to tap a reasonably stable set of intuitions and knowledge.
We suspect that theVMQprobably will reveal performance-metacognition
links for highly complex visual tasks requiring deliberation and those for
which calibration of performance based on short-term experience is dif
ficult. For example, many visual tasks require monitoring a large amount
of visual information for infrequent, vaguely defined
targets. In other situ
ations, visual tasks are preformed only rarely,with little opportunity for
practice. In such cases, a person's beliefs about the difficulty of detecting
the particular target, or about detecting targets in general, could play a
role in allocating visual resources.
However, there are more reasons
to understand metacognition
than
searching for its impact on visual performance. At the beginning of this
article we reviewed many situations that directly tap intuitions about hu
man capabilities. In general, any time it is necessary to predict what an
other person will do, we need to understand not only general principles
about representational states but also more specific limits on his or her
ability to extract information from the visual world. For example, many
legal cases, especially lawsuits, rest on assumptions about the capabilities
of a reasonable person. So when a shopper sues a mall after breaking his
or her ankle in a construction area, legal fact finders must determine
a "reasonable
person" would have seen thewarning signs posted

This inevitably involves ajudgment about the
by themall's management.
reasonable person's visual and cognitive capabilities, and ifpeople have a
whether
mistaken impression of these, then poor assignment of blame may result

(for a review, see Rachlinski, 2004). It is also possible that the VMQ will
prove to be a valid and reliable testof individual differences in understand
ings about visual limits. if this is the case, then itmay be useful to use it as
a guide for juror selection and as a way of testing for particularly strong
misconceptions
among different populations who must make vision-based
judgments.
Finally, results from theVMQ, in conjunction with other research on vi
sual metacognition, may lay the foundations for expert testimony in visual
VISUAL
METACOGNITION
QUESTIONNAIRE
469
cognition. Traditionally, expert testimony is allowed only when an expert

can provide information that isbeyond the ken of the ordinaryjuror, and
we know of at least one case inwhich a psychologist was not allowed to
testifybecause the judge determined that change blindness was consistent
with intuition. The data reported here provide a sound refutation to this
argument. More generally, theVMQ might be used as an educational tool.
It isoften difficult to convince people that their intuitions are wrong, and
theVMQ might be administered, then reviewed as a means of confronting
people with their own incorrectjudgments and perhaps convincing them
of the need to formally test hypotheses about visual functioning instead
of simply considering the degree towhich something is clearly visible to
them.
Appendix.
List of questions
on theVMQ
1. Change detection
1.1 Jerseyand ball
1.2 Person (actor)
1.3 Plate
1.4 Array:Monkey
1.5 Intent scene
1.6 Intentjumble scene
2. Inattentionblindness
2.1 Gorilla
2.2 Think words
2.3 Think space
2.4 Pilot
3. Visual attention
3.1 Frame (breadth)
3.2 Friend (breadth)
3.3 Percentage look at (countenance)
3.4 To see change (necessity)
3.5 To remember (necessity)
4. Auditoryattention
4.1Meaning
4.2 Name
5. Visual memory
5.1-5.3 Inspection set: 20, 50, and 1,000
5.4 Percentage incidental (movie memory)
5.5 Percentage intent (moviememory)
5.6 Face memory
6. Visual knowledge
6.1 Head
6.2 Liberty
6.3 United States ofAmerica
6.4 Date
LEVIN
470
&
ANGELONE
7. Icon and subitization

7.1 Number of letters
7.2 Afterimage
7.3 Partial report
8. Visual search
8.1 Speed feature
8.2 Speed conjunction
8.3 Slope feature
8.4 Slope conjunction
9. Extramission
10. Depth perception
10.1 Fish
10.2Moon (horizon)
10.3 Tracks (horizon)
Notes
This
on work
is based
material
under
no.
grant
SES-0214969,
supported
by the National
to Daniel
T. Levin.
awarded
Science
Foundation
Thanks toGerald Winer, JimArlington, and Stephen Killingsworth for reading
on
of this manuscript.
versions
commenting
previous
to Daniel
about
this article should be addressed
T Levin, De
Correspondence
of
and
Human
Vanderbilt
partment
Development,
University,
Psychology
Peabody
and
College #512,230Appleton Place, Nashville, TN 37203-5701 (e-mail:daniel.tlevin?

for publication
Received
vanderbilt.edu).
2007.
May
5, 2006;
revision
received
March
17,
References
B. L., Levin,
The
roles of representation
(2003).
D.J.
in change
blindness.
Perception, 32, 947-962.
comparison
in vision.
G. (1968).
Short-term
E., 8c Sperling,
Averbach,
storage of information
In R. N. Haber
research and theory in visual perception
(Ed.), Contemporary
(pp.
Angelone,
and
New
202-214).
Beck,
M.
D. T,
8c Simons,
failures
R., Levin,
Beliefs
about
York:
D. T,
Holt,
&
8cWinston.
Rinehart,
Angelone,
the roles of intention
B. L.
and
(2007).
scene
and Cognition,
16, 31-51.
R. J., 8c Rabinowitz,
Biederman,
I., Mezzanotte,
and judging
Detecting
objects
undergoing
14, 143-177.
Psychology,
Change
complexity
blindness
in
change
blindness:
detection.
Consciousness
Scene
(1982).
J. C.
perception:
relational
violations.
Cognitive
8cYarnold,
P. R. (1995).
and explor
Principal-components
analysis,
In
factor
L.
G.
Grimm
P.
R.
Yarnold
and
8c
(Eds.),
atory
analysis.
confirmatory
and understanding
multivariate
statistics (pp. 99-136).
DC:
Reading
Washington,
American
Association.
Psychological
Bryant,
Cottrell,
F. B.,
J. E.,
ception:
218-228.
8cWiner,
The
G. A.
decline
(1994).
Development
of extramission
beliefs.
in the understanding
Developmental
of per
Psychology,
30,
VISUAL
H.
Flavell,J.
(2004).
and
Thinking
of
Development
seeing: Visual
F. L.,
Green,
J. H.,
8c Flavell,
attentional
about
about
vision.
in adults
and
knowledge
metacognition
Cambridge, MA: MIT Press.

Flavell,
471
QUESTIONNAIRE
METACOGNITION
E. R.
(1995).
The
In D. T. Levin
children
of children's
development
focus.
(Ed.),
13-36).
(pp.
706-712.
Developmental
Psychology,
Visual
and the development
of size con
(2004).
metacognition
and seeing: Visual metacognition
in adults
In D. T. Levin
(Ed.), Thinking
knowledge
C. E.
Granrud,
stancy.
31(4),
and children(pp. 121-144). Cambridge, MA: MIT Press.
In
A breakdown
of simultaneous
information
(1991).
processing.
to
8c L. W. Stark
research: From molecular
(Eds.), Presbyopia
biology
visual adaptation. New York: Plenum.
K. A., 8c Loftus, G. R. (2004).
The
effect:
E. M., Carlsen,
"saw-it-all-along"
Harley,
R.
Haines,
F.
G. Obrecht
of visual
Demonstrations
and
Memory,
A.
(2003).
Learning,
Hollingworth,
detection
Cognition,
of retrieval
Failures
scenes.
in natural
Journal
and Performance,
29, 388-403.
F. C,
L.,
Rozenblit,
the
ing
and
of Experimental
constrain
comparison
ofExperimental
Psychology:
Psychology: Human
change
Perception
course.New York: Holt.

A briefer
(1892). Psychology:
James,W.
Keil,
bias. Journal
960-968.
30,
hindsight
C. M.
8cMills,
limits of understanding.
(2004). What
In D. T. Levin
lies beneath:
(Ed.),
Understand
and
Thinking
seeing:
Visual metacognition in adults and children (pp. 227-250). Cambridge, MA:
MIT
Press.
D.
T. (2001).
Visual metacognitions
underlying
change
at the Vision
estimates ofpicture memory. Poster presented
FL.
Sarasota,
Levin,
D. T., 8c Beck, M. R.
Levin,
between
(2004).
failure
metacognitive
blindness
blindness
Sciences
and
conference,
about seeing: Spanning

the difference
Thinking
In D. T. Levin
and success.
(Ed.), Thinking and
seeing:Visual metacognitionin adults and children(pp. 121-144). Cambridge,
MA:
Press.
MIT
D. T, Drivdahl,
Levin,
S. B., Momen,
N. M.,
8c Beck, M. R.
about
the detectability
of visual changes:
of attended
and the continuity
memory,
Consciousness
blindness.
D. T., Momen,
Levin,
ness
role of beliefs
objects
in causing
507-527.
N. M.,
error
in motion
Bulletin
Psychonomic
pictures.
I. (1998).
Inattentional
A., & Rock,
P. H.,
Miller,
sions
& Bigi, L. (1977).

affect performance.
P. H.,
& Weiss,
variables
affect
Miller,
U.,
Neisser,
M.
G.
R.
?f Review,
blindness.
Cambridge,
of how
understanding
Child Development,
48, 1712-1715.
MA:
R.
S.
(1965).
blindness
MIT
stimulus
objects
Press.
dimen
and adults'
about what
knowledge
Child Development,
543-549.
53,
to
Selective
looking: Attending
visually specified
Children's
attention.
(1975).
events. CognitivePsychology,7(4), 480-494.
Nickerson,
attention,
change
to attended
changes
501-506.
4,
Children's
(1982).
selective
8c Becklen,
predictions
about
D. J. (2000). Change
blind
of overestimating
change-detection
metacognitive
397-412.
7(1-3),
ability. Visual Cognition,
to detect
Failure
Levin, D. T., 8c Simons, D. J. (1997).
Mack,
False
and Cognition,
11(4:),
S. B., 8c Simons,
Drivdahl,
The
blindness:
(2002).
The
Short-term
memory
for complex
meaningful
visual
con
LEVIN
472
A demonstration
figurations:
155-160.
R. S., & Adams,
Nickerson,
of capacity.
(1979).
M.J.
Canadian
Journal
of Psychology,
for a common
memory
Long-term
ANGELONE
&
19,
object.
CognitivePsychology,11, 287-307.
Rachlinski, J.J. (2004). Misunderstanding ability,misallocating responsibility. In
D. T. Levin
(Ed.),
and
Thinking
seeing: Visual
metacognition
in adults
and
children
(pp. 251-276). Cambridge, MA: MIT Press.

K
J. E., Shapiro,
Raymond,
8cArnell,
L.,
K. M.
(1992).
Temporary
of
suppression
visual processing in an RSVP task:An attentional blink?Journal ofExperimental
Psychology: Human
D., 8cMcLean,
Reisberg,
Perception
J. (1985).
and Performance,
Meta-attention:
849-860.
18(3),
Do we
know when
we
are
distracted?Journal ofGeneralPsychology,112,291 -306.

R. A.
Rensink,
Visual
(2000).
tentional
search
Visual
processing.
for change:
A
probe
7, 345-376.
into
being
the nature
of at
Cognition,
Rensink, R. A. (2002). Change detection. Annual Review ofPsychology,53, 245

277.
see or not to see: The need
R. A., O'Regan,
J. K., 8c Clark, J. J. (1997). To
to
scenes.
in
for attention
Science, 8, 368-373.
perceive
changes
Psychological
in our midst:
C. F. (1999). Gorillas
Sustained
inattentional
Simons, D. J., & Chabris,
events.
blindness
for dynamic
28, 1059-1074.
Perception,
blindness.
Trends in Cognitive Science,
Simons, D. J., & Levin, D. T. (1997).
Change
Rensink,
1, 261-267.
D.
blindness
J., 8c Levin, D. T. (2003). What makes
change
interesting?
E. Irwin 8c B. H. Ross
The psychology of learning and motivation
(Eds.),
advances
in research and theory: Cognitive
vision
New
(Vol. 42, pp. 295-322).
Simons,
In D.
Academic
Press.
Standing,
L.
Learning
Standing,
L.,
York:
(1973).
25, 207-222.
pictures:
73-74.
Conezio,
10,000
J., 8c Haber,
Single-trial
learning
pictures.
R. N.
of 2500
Journal
ofExperimental
Psychology,
and memory
(1970).
Perception
visual
stimuli. Psychonomic
Science,
for
19,
A feature
8c Gelade,
G.
(1980).
theory of attention.
integration
12, 97-136.
Psychology,
D. A. (2005).
Inattentional
blindness
for a
Levin, D. T., 8cVarakin,
Wayand,
J.W,
noxious
multimodal
stimulus. American
339-352.
118,
Journal
ofPsychology,
Treisman,
A.,
Cognitive
Winer, G. A., 8cCottrell, J. E. (2004). The odd belief that raysexit the eye during
vision.
In D. T. Levin
(Ed.),
Thinking
and
seeing: Visual
and children(pp. 97-120). Cambridge, MA: MIT Press.

Winer,
G. A.,
words,
metacognition
in adults
K. D., 8c Chronister,
M.
(1996).
J. E., Karefilaki,
Images,
Variables
that
influence
beliefs
in children
about
vision
questions:
Cottrell,
and
and adults.Journal ofExperimentalChildPsychology,63, 499-425.

Daniel T. Levin & Bonnie K. Angelone - The Visual Metacognition Questionnaire. A Measure of Intuitions About Vision

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Daniel T. Levin & Bonnie K. Angelone - The Visual Metacognition Questionnaire. A Measure of Intuitions About Vision

Uploaded by

Copyright:

Available Formats

The Visual Metacognition Questionnaire: A Measure of Intuitions about Vision

Author(s): Daniel T. Levin and Bonnie L. Angelone

The Visual Metacognition

Recent research exploring visual perception has revealed a range of strik

own visual capabilities.

they conflict with people's intuitions about their

mented thatpeople have considerabledifficulty

in this area has focused on

about change detection, little research has explored people's

limits and misunderstandings

tion has been

principles for research on visual percep

intuitions about vision. It is quite

by visual attention. Change

(Raymond, Shapiro, & Arnell, 1992; Simons & Levin,

subjectsmissed thechange (see alsoAngelone, Levin,& Simons,2003).

carrying an umbrella walking among the players (formore recent

the relative lengths of the two bars in a cross.While

focused on this task, another stimulus appeared on one of the quadrants

other aspectsof thescene.Although generallyconsistent

overpredictchange detection,whether theyare predictingtheirown or

others' performance in a wide variety of situations (for a review, see Levin

subjectsinLevin and Simons's (1997) originalexperiment.

review,see Flavell,2004).Although thisresearchsuggestsa solid founda

leaving open the possibility that adults misconstrue

overestimates of visual performance

similar to those observed for change

detection, especially for situations involving natural real-world stimuli.

that illustrate limits in visual

It is important to note that these performance

or subject populations. Therefore,

we generally focus on large differences (in the range of 30-80%

correlations between estimates of visual and nonvisual performance.

ismotivated by two hypotheses. The firstintercorrelation hypoth

structure inherent to natural events, scenes, and objects leads

that theywill be able to countenance

of visual information. Therefore,

should be strongest for

stimuli, and intercorrelations would

scenarios describing structured stimuli. A key assumption underlying this

other kinds of visual knowledge

texts in cognition and perception. One critical criterion for

highly unnatural perceptual

that the situations

are all common

rather that they do not require subjects tomake

This isnot to say

or entirely natural, but

sometimes necessary for perception experiments. Many of

based on specific research findings and are focused on more general no

test of visual intuitions. Instead, it

a broad sampling of visual intuition for situations in

which subjects might reasonably base their responses on everyday experi

that the order of questions

does not follow the precise

ask subjects whether theywould

Figure 1.Example change detection question. Each panel representsone of three

to the implicit assumption that adults really do understand the spotlight

among adults despite theassumption in thedevelopmental literature

of this, and in the past we have

the influence of information outside

tentionby assessing intuitionsabout auditoryattention.Although the

is clearly visual, we have included