Spatial versus object visualizers: A new characterization ...nmr.mgh.harvard.edu/mkozhevnlab/wp-content/uploads/pdfs/spatial_… · than men could. Moreover, the distinction between

Copyright 2005 Psychonomic Society Inc 710

Memory amp Cognition2005 33 (4) 710-726

A cognitive style is a psychological dimension that rep-resents consistencies in how an individual acquires andprocesses information (Ausburn amp Ausburn 1978 Mes-sick 1984) Researchers have proposed a wide variety ofcognitive style dimensions (eg Keefe 1979 Messick1976) Some styles are conceptualized as typical responsesto particular stimuli and other styles are regarded as cognitive principles that underlie complex behavior Al-though there have been many studies of cognitive stylesmost of them were not motivated by a theory or generalframework that specifies the dimensions upon which cog-nitive processing may vary As a consequence much ofthe previous work suffered from arbitrary distinctions andoverlapping dimensions Thus it is not surprising that onlya few of these dimensions continue to generate researchsuch as field dependencendashindependence (Witkin MooreGoodenough amp Cox 1977) reflectionndashimpulsivity(Messer 1976) and the visualizerndashverbalizer dimension(Kirby Moore amp Shofield 1988 Paivio 1971 A Richard-son 1977 Riding amp Cheema 1991) We focus on thislast dimension in the present article

Bartlett (1932) Paivio (1971) and A Richardson (1977)were among the first to propose that individuals can bereliably classified as visualizers versus verbalizers Ac-cording to this view visualizers (also called imagers) relyprimarily on imagery when attempting to perform cogni-tive tasks whereas verbalizers rely primarily on verbal-analytical strategies Some researchers have suggestedthat visualizers are expected to be more field indepen-dent and holistic whereas verbalizers are more field de-pendent and analytic (eg Kirby et al 1988) A majorchallenge for visualizerndashverbalizer cognitive style re-search has been to devise methods and instruments to as-sess the dimension accurately Paivio (1971) was the firstto design an individual differences questionnaire withwhich to evaluate the extent to which different people ha-bitually use imagery versus verbal thinking This ques-tionnaire asked participants to indicate whether or noteach of a list of statements such as ldquoI often use mentalpictures to solve the problemsrdquo described their habitualmethod of thinking However factor analyses on the re-sponses given to the individual items identified not onlyimagery and verbal factors but also a number of morespecific factors (Paivio amp Harshman 1983) Moreovermen tended to score high on items that related to the useof imagery in problem solving and in imagining movingobjects whereas women tended to score high on itemsthat related to the use of imagery in remembering andgenerating mental images of previously perceivedscenes In an attempt to improve Paiviorsquos questionnaireA Richardson selected 15 of its most discriminative items

This research was supported by the National Science Foundationunder Grant REC-0106760 We thank Jason Tajima Marie Burrage andJennifer Walsh for helping to implement the tests and conduct the stud-ies We also thank Marija Petkovic Olessia Blajenkova and MichaelMotes for their help in conducting and analyzing the studies with mem-bers of different professions Correspondence concerning this articleshould be addressed to M Kozhevnikov Psychology Department Rut-gers University 101 Warren Street 333 Smith Hall Newark NJ 07102(e-mail mariapsychologyrutgersedu)

Spatial versus object visualizersA new characterization of visual cognitive style

MARIA KOZHEVNIKOVRutgers University Newark New Jersey

and

STEPHEN KOSSLYN and JENNIFER SHEPHARDHarvard University Cambridge Massachusetts

The visual system processes object properties (such as shape and color) and spatial properties (suchas location and spatial relations) in distinct systems and neuropsychological evidence reveals thatmental imagery respects this distinction The findings reported in this article demonstrate that verbal-izers typically perform at an intermediate level on imagery tasks whereas visualizers can be divided intotwo groups Specifically scores on spatial and object imagery tasks along with a visualizerndashverbalizercognitive style questionnaire identified a group of visualizers who scored poorly on spatial imagerytasks but excelled on object imagery tasks In contrast a second group of visualizers scored high onspatial imagery tasks but poorly on object imagery tasks The results also indicate that object visualiz-ers encode and process images holistically as a single perceptual unit whereas spatial visualizers gen-erate and process images analytically part by part In addition we found that scientists and engineersexcel in spatial imagery and prefer spatial strategies whereas visual artists excel in object imagery andprefer object-based strategies

SPATIAL VERSUS OBJECT VISUALIZERS 711

and composed a VerbalizerndashVisualizer QuestionnaireHowever VerbalizerndashVisualizer Questionnaire scoresproved to be only weakly related to the vividness of ex-perienced imagery and to be essentially unrelated toscores on visual-spatial tasks (eg Alesandrini 1981Green amp Schroeder 1990)

At the end of the 1970s research on the visualizerndashverbalizer cognitive style dimension began to appear inthe educational literature On the basis of clinical meth-ods Krutetskii (1976) proposed that individuals can beclassified into groups according to how they processmathematical information The first group the analytictype consists of people who prefer verbal-logical modeswhen attempting to solve problems The second groupthe geometric type consists of those who prefer to useimagery Following Krutetskiirsquos approach Moses (1980)Suwarsono (described in Lean amp Clements 1981) andPresmeg (1986a 1986b) proposed that individuals canbe placed on a continuum that specifies their preferencesfor using imagery while solving mathematical problemsSuwarsono developed the Mathematical Processing In-strument (MPI described in Lean amp Clements 1981) tomeasure an individualrsquos degree of visuality which is thetendency to use visual imagery when solving math prob-lems The questionnaire included a number of simplemathematics problems that could be solved by either vi-sual or analytical methods However its validity wasquestioned because researchers failed to find a clear re-lationship between the degree of visuality and studentsrsquoperformance on either mathematical or spatial abilitytests (Lean amp Clements 1981)

Indeed a host of studies have challenged the validityof the visualizerndashverbalizer dimension Although theverbal subscale of the VerbalizerndashVisualizer Question-naire is related to verbal ability measures (eg Green ampSchroeder 1990 Kirby et al 1988) the visual subscaleis only weakly related to ratings of the vividness of men-tal imagery and is not even moderately related to othervisual-spatial aptitude measures (eg Alesandrini 1981Edwards amp Wilkins 1981 Green amp Schroeder 1990) Amoderate correlation between the visual subscale of theVerbalizerndashVisualizer Questionnaire and spatial visual-ization ability was reported by Kirby et al (1988) butonly when all the items related to dream vividness wereexcluded from the questionnaire Such findings castdoubt on the usefulness of the visualizerndashverbalizer dis-tinction and interest in this dimension has declined overthe past decade In addition the visualizerndashverbalizer di-mension like other cognitive style dimensions suffersfrom the fact that it is not rooted in a more general the-ory of human information processing Instead the dis-tinction is based on intuition and anecdotal observations

In this article we describe a new approach to charac-terizing individual differences in cognitive functioningwhich grows out of a key distinction about informationprocessingmdashnamely the distinction between processingobject properties and processing spatial relations Thisprocessing distinction is rooted in the brain Neuro-

psychological data provide evidence that higher level vi-sual areas of the brain are divided into two functionallyand anatomically distinct pathways the object and thespatial relations pathways (Haxby et al 1991 Kosslyn ampKoenig 1992 Ungerleider amp Mishkin 1982) The objectpathway runs from the occipital lobe down to the inferiortemporal lobe and has been called the ventral systemthis system processes properties of objects such asshape and color The spatial relations pathway runs fromthe occipital lobe up to the posterior parietal lobe and hasbeen called the dorsal system this system processes ob-ject localization and spatial attributes and is also essen-tial for guiding movements

The distinction between perceptual processing of ob-ject properties versus spatial relations extends to visualmental imagery (eg Farah Hammond Levine amp Cal-vanio 1988 Levine Warach amp Farah 1985 Milner ampGoodale 1995) For example Levine et al demonstratedthat lesions in the temporal cortex disrupted performanceof tasks that relied on mental images of objects and theirproperties whereas such lesions did not disrupt spatialimagery in contrast lesions in the posterior parietal cor-tex had the reverse effects (see also Farah et al 1988)Similarly in neuroimaging studies spatial and objectimagery tasks led to very different patterns of brain ac-tivity (Kosslyn Ganis amp Thompson 2001) For exam-ple when participants visualized a route on a map thatthey had memorized prior to the experiment the parietallobes were activated but when participants visualizedfaces or colors the temporal lobes were activated (UhlGoldenberg Lang amp Lindinger 1990) Moreover Koss-lyn and Anderson (summarized in Kosslyn 1994 chap 9)found that the more complex the pattern the more timerequired to form an image of it using spatial imagery incontrast if one encoded the pattern as a single perceptualunit and later visualized it using object imagery the timeto create the image did not depend on image complexity

The distinction between spatial and object imagery isalso supported by findings of gender differences Maleshave performed better than females on a variety of spa-tial orientation and mental rotation tasks (Collins ampKimura 1997 Geary Gilger amp Elliot-Miller 1992 KailCarter amp Pellegrino 1979) whereas females have scoredhigher than males on imagery vividness questionnaires(eg Campos amp Sueiro 1993) Consistent with thesefindings Paivio and Clark (1991) showed that men couldgenerate dynamic images more quickly than women couldbut that women could generate static images more quicklythan men could Moreover the distinction between ob-ject and spatial imagery appears to characterize individ-ual differences Kirby et al (1988) found two visual fac-tors in several of their factor analyses one related tospatial visualization (ie the ability to manipulate andtransform complex spatial images) and the other relatedto vividness of imagerymdashin particular to dream im-agery Hegarty and Kozhevnikov (1999) found that thevisual-spatial representations used by elementary schoolchildren while solving mathematical problems can be re-

712 KOZHEVNIKOV KOSSLYN AND SHEPHARD

liably classified as primarily spatial or primarily pictor-ial some of the children consistently used schematicspatial representations and some preferred to use object-based pictorial representations Hegarty and Kozhevnikovalso showed that preference for spatial representations isassociated with both high spatial visualization abilityand better performance on mathematics problem solv-ing Furthermore they found that the use of spatial rep-resentations in problem solving is negatively correlatedwith the use of pictorial representations Aginsky Har-ris Rensink and Beusmans (1997) reported a similar dis-sociation between visual-object versus spatial strategiesin wayfinding tasks They found that some participantsbased their wayfinding decisions on visually recognizedlandmarks along a route whereas others represented theenvironment as a survey map from the start

Critically Kozhevnikov Hegarty and Mayer (2002)found that although the majority of verbalizers they testedhad average spatial ability visualizers fell into two con-trasting groups those who had high spatial ability and thosewho had low spatial ability with only a small number ofvisualizers having average spatial ability Kozhevnikovet al showed that visualizers who had high spatial abil-ity tended to generate schematic abstract images wheninterpreting science graphs whereas visualizers who hadlow spatial ability tended to generate detailed pictorialimages of individual objects

The focus of this article is to develop a new approachto the visualizerndashverbalizer cognitive style dimensionUnlike all previous efforts to characterize the visualizerndashverbalizer cognitive style we reject the idea that imagerycan be characterized as variation along a single dimen-sion We argue instead that there are in fact two distincttypes of visual cognitive style reflecting different ways ofgenerating mental images and processing visual-spatialinformation Specifically we predict that some people areespecially good in constructing vivid pictorial and de-tailed images of individual objects whereas other peopleexcel in creating images that represent spatial relationsamong objects and in imagining spatial transformations

In Studies 1 and 2 we examined the performance ofvisualizers and verbalizers on different imagery tasks Ifthere are two qualitatively different types of visualizerswe would expect to find a bimodal distribution withsome excelling at spatial imagery and some excelling atobject imagery We would expect no such distributionamong the verbalizers but instead we would expect a nor-mal distribution of verbalizersrsquo scores across all imagerytasks In Study 3 we compared the performance of visu-alizers and verbalizers on general and verbal intelligencetests Study 4 focused on systematic differences in im-agery abilities for people engaged in different types ofprofessional activitiesmdashin particular visual arts versusscience and engineering We hypothesized that scientistsand engineers should have excellent spatial imagery andprefer spatial strategies but visual artists should have ex-cellent object imagery and prefer object imagery strategies

STUDY 1

In this study we administered a test that was intendedto assess spatial imagery (the mental Paper Folding Test orPFT) a questionnaire that was intended to survey object-based imagery (the Vividness of Visual Imagery Ques-tionnaire or VVIQ) and a VisualizerndashVerbalizer Cogni-tive Style instrument modeled after an instrument usedin the educational literature to distinguish verbalizersfrom visualizers If in fact there are two types of visu-alizers and these styles are qualitatively distinct peoplewho score high on the spatial-based imagery task shouldnot score high on the object imagery questionnaire andvice versa

MethodParticipants One hundred sixty-two participants took part in

Study 1 This group consisted of 116 paid volunteers (56 males and60 females) 18ndash40 years of age recruited through sign-up sheetsposted around the Harvard University campus and 46 volunteers(21 males and 25 females) recruited from the psychology partici-pant pool at Rutgers University at Newark as well as through sign-up sheets posted around the Rutgers campus

Materials The materials consisted of a pretest questionnairewhich included questions about the studentrsquos age and gender theVisualizerndashVerbalizer Cognitive Style Questionnaire the PFT andthe VVIQ

VisualizerndashVerbalizer Cognitive Style Questionnaire The pur-pose of this questionnaire was to measure the participantsrsquo prefer-ence for using imagery as opposed to verbal-logical strategies whensolving problems The VisualizerndashVerbalizer Cognitive Style Ques-tionnaire includes 11 mathematical problems selected from theMPI1 and has been used in previous studies to distinguish betweenvisualizers and verbalizers (Kozhevnikov et al 2002 Lean ampClements 1981 Presmeg 1986a 1986b) specifically the prob-lems were taken from or modeled after those of Suwarsono (de-scribed in Lean amp Clements 1981) As in previous measures of thevisualizerndashverbalizer cognitive style dimension (Lean amp Clements1981) this test included two parts Part I contained 11 written math-ematics problems that can be solved by either visual or nonvisualmethods (see examples of the problems in Appendix A) A pilotstudy determined that these problems were challenging but not toodifficult for the students and that the participants used a variety ofstrategies to solve them Part II contained descriptions of the mostcommon ways of solving the problems in Part I (identified fromprior interviews with students about how they solved these problemseg Kozhevnikov et al 2002 Lean amp Clements 1981) In Part IIthe participants were asked to indicate the strategy they had used foreach problem Space was provided for individuals to describe al-ternative methods that were not listed We assigned a score of 2 forany visual solution or combination of visual solutions checked bythe students a score of 0 for any nonvisual solution (irrespective ofwhether the answer to the problem was correct or incorrect) and theintermediate score of 1 for a combination of verbal and visual meth-ods The internal reliability (alpha-Cronbach) of the questionnairewas 80

Paper Folding Test According to Ekstrom French and Harman(1976) the PFT measures spatial visualization ability which re-flects the ability to apprehend encode and mentally manipulate ab-stract spatial forms (Lohman 1988) Researchers (eg Poltrock ampAgnoli 1986 Poltrock amp Brown 1984) have shown that scores ontests of spatial visualization ability are related to how well peoplecan rotate and integrate imaged forms and essentially are not re-


lated to participantsrsquo ability to generate vivid or detailed visual im-ages The participants were administered Part I of the PFT whichconsists of 10 items Each item showed successive drawings of twoor three folds made in a square sheet of paper The final drawingshowed a hole being punched in the folded paper The participantselected one of five drawings to indicate what the punched sheetwould look like when fully opened The internal reliability of thePFT was 84 (Ekstrom et al 1976) Examples of items from the PFTare provided in Appendix B

The total score for the PFT was calculated using a correction forguessing according to the formula R W(n 1) where R thenumber of correct items W the number of incorrect items and n the number of response options for each item (n 5 for the PFT)

Vividness of Visual Imagery Questionnaire The VVIQ (Marks1972) is the most frequently used measure of how vividly individ-uals can form visual mental images The VVIQ consisted of 16items for which the participants rated on a 5-point scale the vivid-ness of mental images they were asked to create Typical itemswere ldquoThe sun is rising above the horizon into a hazy skyrdquo and ldquoAstrong wind blows on the trees and on the lake causing wavesrdquo Theinternal reliability of the questionnaire was 88

Procedure All the participants were tested individually They wereadministered a pretest questionnaire the VisualizerndashVerbalizer Cog-nitive Style Questionnaire the PFT and the VVIQ The order of thesemeasures was counterbalanced The participants were given only3 min for the PFT there were no time limits for the other measures

Results and DiscussionWe classified all the participants as either visualizers

(n 83) or verbalizers (n 79) on the basis of a mediansplit of their score on the VisualizerndashVerbalizer Cogni-tive Style Questionnaire2 Descriptive statistics for eachof the variables we assessed are provided in Table 1 as afunction of cognitive style The mean score on the cogni-tive style questionnaire was 750 (SD 212) for verbal-izers and 1347 (SD 232) for visualizers There was onlya marginal difference between visualizers and verbaliz-ers on the VVIQ [F(1161) 310 MSe 87783 p 08] and no hint of a difference on the PFT [F(1161) 1] The correlation between the VVIQ and the PFT wasnegative and marginally significant (r 014 p 07) Scores on the VisualizerndashVerbalizer CognitiveStyle Questionnaire did not correlate significantly withthose on the PFT (r 03 p 97) or the VVIQ (r 09 p 24) Thus when all visualizers are treated asthe same the results are consistent with previous re-search that did not show correlations between measuresof visualizer versus verbalizer style and performance onvisual-spatial tasks (eg Alesandrini 1981 Edwards ampWilkins 1981 Green amp Schroeder 1990 Lean ampClements 1981)

In order to differentiate between two types of visual-izers we grouped the participants with respect to theirspatial imagery ability as follows low (scores in the bot-tom 333 of the distribution on the PFT) intermediate(scores in the middle 333) or high (scores in the top333) A median split would not have allowed us to iden-tify the participants who scored unusually high or low ascompared with intermediate levels as is essential in orderto test the present hypothesis Figure 1 presents the num-ber of visualizers versus verbalizers who scored low inter-

Table 1Mean Scores and Standard Deviations

for the Paper Folding Test (PFT Out of a Total Possible Scoreof 10) and the Vividness of Visual Imagery Questionnaire

(VVIQ out of 80) as a Function of Cognitive Style

Visualizers Verbalizers

Test M SD M SD

PFT 649 234 627 221VVIQ 6310 986 6065 881

visualizersverbalizers

35

30

25

20

15

10

5

0low PFT

Nu

mb

er o

f p

arti

cip

ants

high PFTintermediate PFT

Spatial imagery

Figure 1 Number of visualizers versus verbalizers who scored lowintermediate and high on the Paper Folding Test (PFT)


mediate and high in spatial imagery as assessed by thePFT A chi-square test revealed that the distribution oflow intermediate and high spatial visualization abilitywas significantly different for visualizers and verbalizers[ χ 2(2) 681 p 05] As can be seen in Figure 1 thevisualizers were not a homogeneous group there weretwo distinct subgroups high spatial and low spatial vi-sualizers with a significantly smaller number of visual-izers of intermediate spatial imagery ability In contrastto the visualizers the distribution of the verbalizersrsquoscores did not show such bimodality These results areconsistent with previous findings in Kozhevnikov et al(2002) which documented two contrasting groups of vi-sualizers those who had good spatial imagery and thosewho had poor spatial imagery3

At first glance the fact that there are low spatial visu-alizers may seem puzzling Why do these people preferto use imagery instead of analytical methods to processinformation when they have poor spatial imagery Weconsidered two alternatives (1) Low spatial visualizershave poor imagery abilities but nevertheless prefer touse imagery because they also lack verbal-analyticalability and (2) low spatial visualizers prefer to use men-

tal imagery because they are in fact good at some formsof imagerymdashjust not the forms required to perform wellon spatial visualization tests In this case they may excelat object imagery To investigate these alternatives weexamined how the visualizers and verbalizers scored onthe VVIQ In order to identify the participants whoscored especially high or low we again classified scoresas low (scores in the bottom 333 of the distribution)intermediate (scores in the middle 333) or high(scores in the top 333) Figures 2A and 2B present thenumber of low spatial intermediate spatial and highspatial visualizers and verbalizers respectively whoscored low intermediate and high on the VVIQ A chi-square test revealed significant differences among lowintermediate and high spatial visualizers with respect totheir VVIQ scores [ χ 2(4) 1481 p 01] As can beseen in Figure 2A for the most part the visualizersscored high either on the PFT or on the VVIQ Further-more most (625) low spatial visualizers scored highon the VVIQ At the same time most (633) visualiz-ers who scored high on the VVIQ had low spatial im-agery ability There were only a few visualizers whoscored high on both the PFT and the VVIQ (84) or

Figure 2 Number of low intermediate and high (A) spatial visualizers and(B) verbalizers who scored low intermediate and high on the Vividness of Vi-sual Imagery Questionnaire (VVIQ) PFT Paper Folding Test

low VVIQintermediate VVIQhigh VVIQ24

20

16

12

8

4

0

low PFT

Nu

mb

er o

f p

arti

cip

ants


Spatial imagery

low PFT high PFTintermediate PFT

Spatial imagery

Nu

mb

er o

f p

arti

cip

ants

20

16

12

8

4

0

low VVIQintermediate VVIQhigh VVIQ

Visualizers

Verbalizers

A

B


who scored low on both tests (72) As for the verbal-izers they generally did not score high on the PFT or onthe VVIQ A chi-square test revealed that the differenceamong low intermediate and high spatial verbalizerswith respect to their VVIQ scores was not significant[ χ 2(4) 414 p 38] Further analysis (a one-wayanalysis of variance [ANOVA]) confirmed that there wasa significant difference in the VVIQ scores for visualiz-ers of differing spatial imagery ability [F(283) 392MSe 92417 p 05] Pairwise comparisons4 revealedthat visualizers with low spatial imagery reported sig-nificantly more vivid mental imagery than did visualiz-ers with high spatial imagery ( p 05) No significantdifference was found in the VVIQ scores among verbal-izers of low intermediate and high spatial imagery[F(279) 144 MSe 81509 p 24] The meanVVIQ scores for visualizers and verbalizers of differingspatial imagery ability are shown in Table 2

Finally a chi-square analysis revealed marginal differ-ences between the numbers of females and males in thetwo groups of visualizers [ χ 2(1) 237 p 1] In thegroup that scored low on the PFT and high on the VVIQ36 were males and 64 were females in the group thatscored high on the PFT and low on the VVIQ 57 weremales and 43 were females

In summary as was predicted we found a bimodal dis-tribution of visualizersrsquo scores on the PFT and the VVIQwhereas the verbalizersrsquo scores were not distributed inthis way These findings suggest that visualizers in con-trast to verbalizers are not a homogeneous group butrather consist of two different types spatial visualizersand object visualizers Although the gender differenceswere not significant the trend might suggest that fe-males tend to be object visualizers and males tend to bespatial visualizers However the results show that the dif-ference between spatial and object visualizers cannot bereduced to a gender difference numerous men were objectvisualizers and numerous women were spatial visualizers

Furthermore the fact that most high spatial visualiz-ers scored low on the VVIQ whereas most low spatial vi-sualizers scored high provides evidence that the ability togenerate vivid high-resolution visual images and theability to perform spatial transformations rely on distinctprocesses These results help to explain the apparent lackof correlation between scores on the traditional visualizerndashverbalizer cognitive style dimension and those on visual-spatial tasks The relationship between the use of imagerystrategies and imagery ability cannot be characterized as

variation along a single dimension Failure to appreciatethis fact led previous researchers to the misleading con-clusion that ldquoverbalizers outperform more visual stu-dents on spatial testsrdquo (Lean amp Clements 1981 p 296)or that ldquoit is tenuous to equate a vivid imager with a vi-sualizer since in all probability the two are separateissuesrdquo (Strosahl amp Ascough 1981 p 429) Thus in-stead of a dichotomous cognitive style dimension (visu-alizers vs verbalizers) the results of Study 1 suggest theexistence of three dimensions verbalizers object visu-alizers and spatial visualizers

A limitation of Study 1 however is that the object im-agery survey (the VVIQ) was a self-report measure andtherefore subjective Study 2 was designed to examinewhether object visualizers are in fact better than spatialvisualizers on objective measures of object imagery

STUDY 2

To measure object imagery we designed two comput-erized tasks The first the grain resolution task was de-signed to assess participantsrsquo ability to generate detailedhigh-resolution images of individual objects In this taskthe participants read the names of pairs of objects andused imagery to evaluate which object in each pair hadthe surface with the finer texture or denser grain Thesecond task the degraded pictures task was designed toassess participantsrsquo ability to recognize and identify anobject presented in a degraded picture In this task we as-sumed that top-down processing would be used to completeobscured portions of the object and that this completionprocess would rely on the same mechanisms that under-lie object imagery (see Kosslyn 1994)

In addition we administered a version of the Shepardand Metzler (1971) mental rotation task to assess spatialimagery and we designed a computerized embedded pic-tures task to tap either spatial or object aspects of im-agery depending on the type of question asked about thestimuli In half of the items of the embedded picturestask the participants were to identify global propertiesthat characterized the entire picture such as symmetrywhereas in the other half of the items the participantswere to identify local properties that characterized onlya part of the picture such as a particular type of junctionconnecting two lines in the image Kosslyn and Anderson(summarized in Kosslyn 1994 chap 9) provided evidencethat object imagery is formed on the basis of individualperceptual units stored in visual memory whereas spa-

Table 2Mean Scores and Standard Deviations on the Vividness

of Visual Imagery Questionnaire (out of 80) as a Function of Cognitive Style and Spatial Ability

Low Spatial Intermediate Spatial High Spatial

M SD M SD M SD

Visualizers 6798 960 6373 905 6014 983Verbalizers 6240 987 6155 867 5812 817


tial images are necessarily formed by using spatial rep-resentations to arrange components sequentially Be-cause global properties characterize an entire picture wepredicted that the participants would rely on object im-agery to identify these properties in embedded picturesby the same token because local properties characterizeonly part of an image we predicted that in order to assesssuch properties the participants would generate theimage part by part and therefore would rely on spatialimagery

MethodParticipants All the participants who took part in Study 1 were

invited to participate in Study 2 Overall 51 participants (28 fe-males and 23 males) who previously had taken part in Study 1 vol-unteered to take part in this study On the basis of their scores fromStudy 1 the participants were divided into verbalizers (n 26) ob-ject visualizers (n 12) and spatial visualizers (n 13)

Materials and Procedure All the participants were tested in-dividually They sat 35ndash45 cm in front of a computer screen andsaw stimuli presented by a Macintosh PowerMac G3 on an NECE700 monitor with a resolution of 1024 768 (75 Hz) The par-ticipants received four computerized tasks Shepard and Metzler(1971) mental rotation grain resolution degraded pictures and em-bedded pictures All the tasks were presented with the PsyScopeprogram (Cohen MacWhinney Flatt amp Provost 1993) The order ofthe tasks was counterbalanced There was no set time limit on thetasks but the participants were instructed to respond as quickly andaccurately as possible speed and accuracy were measured in all thetasks

Shepard and Metzler mental rotation task In this computerizedversion of the Shepard and Metzler (1971) mental rotation task theparticipants saw pairs of two-dimensional pictures of angular three-dimensional forms The forms in a pair were rotated from 0ordm to 180ordmrelative to each other within the picture plane Half the pairs con-tained identical shapes and half contained mirror images The par-ticipants were asked to judge whether the forms in a pair had the sameshape or were mirror images There were 9 practice trials and 109test trials The internal reliability of the task for accuracy was 88

Degraded pictures task The degraded pictures task was adaptedfrom the Snow Pictures Test of Ekstrom et al (1976) In each trialof this task the participants saw a degraded line drawing of a com-mon object (eg an umbrella a pair of scissors or a table) In somecases in which it was judged that the line drawings were too easy todistinguish from the background noise further degradation was in-troduced using Adobe Photoshop Once the participants had iden-tified the object in the picture they were to press the Return key Afterdoing this the picture disappeared and the participants had to typethe name of the object after which the next degraded picture ap-peared on the screen There were 2 practice trials and 10 test trialsThe internal reliability of the task for accuracy was 73 An exam-ple from the degraded pictures task is shown in Figure 3A Figure 3Bshows the outline of the degraded object hidden in Figure 3A

Grain resolution task On each trial of this task the participantssaw a pair of names of objects written on the computer screen andidentified which of the two named objects had a finer texture ordenser grain Grain referred to bits or particles per unit of area orvolume Some examples of grain include the density of spots perarea (leopard skin vs giraffe skin) number of particles per unit ofvolume (a heap of grains of salt vs a heap of poppy seeds) andnumber of air bubbles per volume (soda vs shampoo) The pairs ofwords were presented on the computer screen one pair at a timeand the two words were separated by a dash If the participant be-lieved that the object named on the left side of the dash had a densergrain he or she was to press the ldquobrdquo key on the keyboard whereas

if he or she believed that the object named on the right had the densergrain he or she was to press the ldquonrdquo key After the participant hadresponded the pair disappeared and a new pair appeared on thescreen The task consisted of 2 practice and 20 test trials The inter-nal reliability of the task for accuracy (alpha-Cronbach) was 62

Computerized embedded pictures task This task was adaptedfrom that of Rouw Kosslyn and Hamel (1997) The participantsviewed line drawings of common objects (eg a piano a sailboator a stool) They had to decide whether specific global or localproperties were present The global properties were symmetry(along any axis) and parallelism (ie two or more lines in the draw-ing were parallel) the local properties were an arrow junction (threeline segments that meet to form an arrow) and a T-junction (twoperpendicular lines that intersect at a point to form a T) Figure 4provides examples of the stimuli The drawings were from Snod-grass and Vanderwart (1980) We recorded four auditory cues (thewords symmetry parallelism arrow junction and T-junction) withMacromedia SoundEdit 16 each recording was approximately 1 secin duration During the trials a fixation asterisk was presented for075 sec and then the picture was displayed on the screen for 25 secThe picture then disappeared and the participants visualized it onthe blank white screen The screen remained blank for 2 sec andthen one of the four property cues was presented auditorily Theparticipants decided whether or not the picture contained the namedproperty The 16 test trials were divided so that there were 4 trialsfor each of the four properties The internal reliability of the task foraccuracy on global properties items (alpha-Cronbach) was 56 andon local properties items was 54

A

B

Figure 3 (A) Example item (umbrella) from the degraded pic-tures task (B) The outline of the umbrella hidden in the picture


Results and DiscussionWe assessed overall accuracy and response latency

from the mental rotation grain resolution and degradedpictures tasks The mean proportions correct and re-sponse times for each task are presented in Figures 5 and6 respectively as a function of cognitive style The datawere analyzed in a 3 3 repeated measures ANOVAwith task as a within-subjects variable and cognitivestyle (object visualizers spatial visualizers and verbal-izers) as a between-subjects variable The accuracy andresponse time data were considered in separate analysesFor accuracy there was a main effect of task [F(296) 161 MSe 1437Endash02 p 0001] indicating that theparticipants were not equally accurate on all tasks Themain effect of cognitive style was not reliable [F(250) 162 MSe 4154Endash03 p 2] However there was areliable interaction between task and cognitive style[F(496) 539 MSe 1437Endash02 p 001] whichindicates that object visualizers spatial visualizers and

verbalizers performed differently on different imagerytasks Pairwise comparisons revealed that object visual-izers were less accurate than either spatial visualizers orverbalizers ( p 01) on the mental rotation task whereasno difference was found between verbalizers and spatialvisualizers ( p 7) At the same time object visualiz-ers were more accurate than spatial visualizers ( p 05)on the grain resolution task whereas there was no dif-ference between verbalizers and object visualizers ( p 9) For the degraded pictures task object visualizersperformed better than verbalizers ( p 05) and margin-ally better than spatial visualizers ( p 06) but therewas no difference in how accurately the verbalizers andspatial visualizers performed this task ( p 9)

For response times there also was a main effect oftask [F(296) 4474 MSe 13E07 p 001] in-dicating that the participants required more time forsome tasks than for others The main effect of cognitivestyle was not reliable [F(296) 162 MSe 14E07

Figure 4 Example of the picture stimuli presented to participants in the em-bedded pictures task (A) Participants were asked to decide whether the picturewas symmetrical (global property the answer is ldquonordquo) (B) Participants wereto decide whether a T-junction was present in the picture (local property theanswer is ldquoyesrdquo)

object visualizers verbalizers spatial visualizers1

9

8

7

6

5

4

3

2

1

0Grain Degraded pictures Mental rotation

Pro

po

rtio

n c

orr

ect

Figure 5 Mean proportions correct for the grain resolution degraded pictures and Shep-ard and Metzler mental rotation tasks as a function of cognitive style


p 2] However there was a reliable interaction be-tween task and cognitive style [F(496) 307 MSe 13E07 p 05] The spatial visualizers were fasterthan both the object visualizers and the verbalizers ( p 05) on the mental rotation task whereas the object vi-sualizers and the verbalizers required comparable amountsof time ( p 8) The finding that verbalizers were as ac-curate as spatial visualizers on the mental rotation taskwhereas their response times were significantly longersuggests that verbalizers might use nonspatial strategiesto solve the mental rotation problems

There were no signif icant differences in responsetimes among the three groups of participants in the grainresolution task ( p 7) That there was no difference be-tween verbalizers and object visualizers in either speedor accuracy on the grain resolution task may imply thatthe task could be performed on the basis of verbally coded

knowledge about the density or grain of the two objectsIn fact Blajenkova and Kozhevnikov (2002) showed thatperformance on many of the grain resolution items cor-relates highly with scores on verbal analogy tests On thedegraded pictures task the performance of object visu-alizers was faster than that of verbalizers ( p 05) andmarginally faster than that of spatial visualizers ( p 08)We did not find a significant difference between the re-sponse times of spatial visualizers and verbalizers ( p 9)

In addition we assessed accuracy and latency for twoproperty types (global and local) in the embedded pic-tures task The mean scores for each property type areshown in Figure 7 as a function of cognitive style The la-tency data are shown in Table 3 We conducted a 2 3ANOVA for the embedded pictures task using propertytype (global vs local) as a within-subjects variable andcognitive style (object visualizers spatial visualizers


14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)

Figure 6 Mean response times of object visualizers verbalizers and spatial visualizers onthe grain resolution degraded pictures and Shepard and Metzler mental rotation tasks


6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties

Figure 7 Mean scores on the items of the embedded pictures task that required the iden-tification of local and global properties as a function of cognitive style


and verbalizers) as a between-subjects variable first foraccuracy and then for response times For accuracy therewas no difference between the two types of properties[F(149) 192 MSe 2022 p 17] and no main ef-fect of cognitive style [F(149) 1] However we didfind that the two variables interacted [F(249) 692MSe 2022 p 05] which shows that the groups dif-fered in how accurately they identified global versuslocal properties Pairwise comparisons revealed that ob-ject visualizers were more accurate than spatial visualiz-ers when they identified global properties ( p 05)whereas spatial visualizers were more accurate thanobject visualizers when they identified local properties( p 01) For both types of property judgments verbal-izers scored in between the object visualizers and thespatial visualizers (see Figure 7) For response time therewas no effect of property type or cognitive style nor wasthe interaction between property type and cognitive stylesignificant

In short the results from Study 2 nicely dovetail withthose from Study 1 Object visualizers are indeed clearlydifferent from spatial visualizers Object visualizerswere more accurate and faster on the degraded picturestask which requires recognizing individual shapes Inaddition they were more accurate than spatial visualiz-ers on the grain resolution task which requires generat-ing detailed high-resolution images In contrast spatialvisualizers were more accurate and faster than object vi-sualizers in the mental rotation task which requires ro-tating images of three-dimensional figures On the basisof these results one could argue that spatial visualizersare more accurate and faster in generating and manipu-lating dynamic images whereas object visualizers aremore accurate and faster in generating static imagesHowever the differences between object and spatial vi-sualizers cannot be reduced to differences in their perfor-mance of dynamic versus static tasks The results fromthe embedded pictures task showed that object visualiz-

ers were also more holistic and performed better on thetrials that required identifying global properties ofshapes whereas spatial visualizers were more successfulin identifying local properties

STUDY 3

One possible explanation for the results of Study 2 isthat object visualizers in contrast to spatial visualizershave difficulty processing abstract information and thusthey perform worse on spatial imagery tasks that requiremanipulation of abstract mental images The goal ofStudy 3 was to compare the general and verbal intelli-gence of object visualizers spatial visualizers and ver-balizers and to examine how the three groups process ab-stract information

MethodParticipants All the participants who had participated in Study 2

were invited to participate in Study 3 Overall 42 participants (23 males and 19 females) who had participated in Study 2 volun-teered to take part in this study On the basis of their scores fromStudy 1 the participants were divided into verbalizers (n 21) ob-ject visualizers (n 11) and spatial visualizers (n 10)

Procedure The participants were given two tests the AdvancedProgressive Matrices (APM Raven Raven amp Court 1998) whichis a widely used nonverbal measure of fluid reasoning and theWechsler Abbreviated Scale of Intelligence Similarities (WASIWechsler 1999) which is considered to be a measure of verbal in-telligence The APM consisted of a complete set of 36 items eachof which contained a pattern or design A portion of the pattern wasmissing and the task was to find which of eight possible answersbest completed the pattern or design In the WASI the participantswere presented with a complete set of 19 pairs of words and had todescribe how two named qualities characteristics or states werealike (eg red and blue love and hate freedom and law) The scorefor each item depended on the degree to which the response describeda general property primarily pertinent to both words in the pair

The participants first completed the APM and then the WASIFor the APM the participants were allotted 30 min there was notime limit for the WASI

Table 3Latency Data (Means and Standard Deviationsin Milliseconds) for the Embedded Pictures Task

(Global and Local Properties) as a Function of Cognitive Style

Verbalizers Object Visualizers Spatial Visualizers

Task Property M SD M SD M SD

Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933

Table 4Mean Scores and Standard Deviations for the Wechsler Abbreviated Scale of Intelligence

(WASI) and the Advanced Progressive Matrices (APM) Tests as a Function of Cognitive Style


Test M SD M SD M SD

WASI 3852 360 3727 307 3680 673APM figural 1438 530 1540 392 1608 437APM analytical 819 374 320 103 1027 261


Results and DiscussionThe mean scores for verbalizers object visualizers and

spatial visualizers on the WASI and the APM are presentedin Table 4 There was no difference among the threegroups on the WASI However a one-way ANOVA re-vealed a difference among the groups on the APM[F(241) 453 MSe 27174 p 05] Pairwise com-parisons indicated that the spatial visualizers performedsignificantly better on this test than did the object visu-alizers ( p 05)

To examine further the performance of the object ver-sus the spatial visualizers on the APM we distinguishedbetween two different classes of matrices problems fig-ural and analytical This distinction has previously beenoffered by several researchers (Carpenter Just amp Shell

1990 Hunt 1975 Prabhakaran Smith Desmond Gloveramp Gabrieli 1997) Figural problems require primarilyvisual-spatial analysis and are based on perceptual oper-ations or characteristics such as continuation or super-imposition Note that these characteristics can be processedusing either spatial or object imagery Analytical problemsin contrast are thought to require abstract-analytical rea-soning Examples of figural and analytical problems areprovided in Figure 8 We identified 19 figural and 17 an-alytical problems in the APM Although the groups per-formed comparably on the figural problems [F(241) 245 MSe 34145 p 10] they differed markedly onthe analytical problems [F(241) 1517 MSe 18877p 001] Object visualizers were worse ( p 001) thanboth spatial visualizers and verbalizers on this type ofproblem

In summary object visualizers were comparable to theother groups in abstract verbal reasoning and in using vi-sual analysis to solve the figural items in the APM In con-trast object visualizers performed significantly worse thanspatial visualizers and verbalizers on the analytical prob-lems in the APM A closer look at these items suggests anaccount for this finding The analytical problems in theAPM in fact appear to require spatial reasoning In-deed when Prabhakaran et al (1997) examined brain ac-tivation while participants solved figural and analyticalproblems they found that analytical problems activatedposterior parieto-occipital areas associated with spatialprocessing Because object visualizers tend to processvisual images globally their poor performance on the an-alytical problems might reflect difficulty in forming cat-egorical spatial relations rather than a lack of generalabstract thinking These results are consistent with thoseof Kozhevnikov et al (2002) who reported that objectvisualizers had difficulty interpreting abstract graphproblems they consistently considered only the globalshape of the graph and did not attempt to restructure thegraph visually and consider each interval successively

STUDY 4

Although relying on object imagery may impair per-formance on tasks that require one to consider spatial re-lations or mentally transform the orientation or locationof patterns such imagery should be useful for many othertypes of tasks For example we have provided evidencethat object imagery can help one to recognize degradedpictures (where one can use imagery to complete a shape)and to generate high-resolution detailed images of shapeIn addition object visualizers who are especially goodin generating vivid pictorial images may excel in suchtasks as drawing or painting whereas spatial visualizersmay succeed in such tasks as science and engineeringStudy 4 was designed as a preliminary investigation tocompare how members of different professionsmdashnamelyvisual artists versus scientistsmdashperform on object andspatial imagery tests

A

B

Figure 8 Examples of Advanced Progressive Matrices prob-lem types (A) a figural problem and (B) an analytical problemThe participants were instructed to use the eight patterns in theupper array to determine which of the eight alternative responsesshould appear in the bottom right position


MethodParticipants The participants were 10 visual artists and 14 sci-

entists The group of scientists included physicists (n 10) and engineers (n 4) with at least a graduate degree in the field andseveral years of research or professional experience The group ofvisual artists included professional painters (n 6) photographers(n 2) and interior designers (n 2) with at least 2 years of pro-fessional experience in the field All the participants were selectedon the basis of pilot interviews in which they reported their prefer-ences for using imagery versus verbal-analytical strategies in theirwork and in everyday life

Procedure The participants were asked to complete the pretestquestionnaire They were then given the same version of the PFT asthat used in Study 1 and a paper-and-pencil version of the grain res-olution task The paper-and-pencil version of the grain resolutiontask included the same 20 pairs of items that we used in the com-puterized version in Study 2 The participants were given only 3 minto take the PFT and 5 min to complete the grain resolution task

In addition the participants were shown a graph that depicted anobjectrsquos position as a function of time (see Figure 9) and were askedto visualize describe and draw the situation represented by thegraph The correct description of the graph is that the object ini-tially is stationary (its position remains the same in the first inter-val of the graph) then moves at a constant velocity in the second in-terval (its position changes linearly with time) and finally comesto a stop (its position remains constant again)

Results and DiscussionThe mean scores on the paper folding and grain reso-

lution tasks are shown in Table 5 as a function of profes-sion We performed a two-way ANOVA with task (paperfolding and grain resolution) as a within-subjects vari-able and profession (scientists vs artists) as a between-subjects variable The results revealed neither an effectof task nor an effect of profession However two vari-ables did interact [F(122) 629 MSe 5670 p 05]Pairwise comparisons showed that the scientists per-formed better than the visual artists on the PFT ( p 02)whereas the visual artists performed better than the sci-entists on the grain resolution task ( p 03)

In addition we coded all the participantsrsquo responseson the graph problem into three categories (1) a pictor-ial interpretation in which the participant interpreted

the graph literally as a pictorial illustration of a situation(2) an irrelevant interpretation in which the participantinterpreted the graph (correctly or incorrectly) by visu-alizing irrelevant but not pictorial features of the graphor (3) a schematic interpretation in which the participantreferred to the graph as an abstract schematic represen-tation (independently of whether the actual interpreta-tion was correct or incorrect) Two independent ratersanalyzed the participantsrsquo responses to the graph ques-tion and the interrater reliability was 93 Consistent withprevious findings that object visualizers tend to interpretgraphs as literal pictures of the situation (Kozhevnikov et al 2002) 8 (out of 10) visual artists interpreted thegraph problem as a literal pictorial illustration of a situ-ation or as the path of the actual motion and did not at-tempt to interpret the graph as an abstract schematic rep-resentation The artists consistently referred to the globalshape of the graph and expected the shape of the graphto resemble the path of the actual motion An example ofan artistrsquos report is illustrated in Figure 10 In contrastall 14 scientists produced schematic descriptions of thesituation depicted in the graph an example of a typicalscientistrsquos response is also given in Figure 10 The dif-ference between visual artistsrsquo and scientistsrsquo interpreta-tions of the graph was highly significant [ χ2(2) 2016p 0001]

The scientists and the artists clearly had different sortsof imagery Different professions might promote objectversus spatial imagery or perhaps people with one sortof imagery select a field on the basis of their imageryabilities or preferences The fact that the scientists weremuch more successful when interpreting graphs is notsurprising given their experience with graphs duringtheir professional work What was surprising howeveris the clear qualitative difference in the scientistsrsquo andthe visual artistsrsquo drawings and descriptions One mightargue however that the graph presented to the partici-pant is abstract and its interpretation requires prior ex-perience with graphs However it is interesting to notethat previous findings reported by Kozhevnikov et al(2002) showed the same qualitative difference in graphinterpretation between object and spatial visualizers whohad the same mathematical background but who did nottake any science or physics classes at high school or col-lege levels (and thus neither group was exposed at leastin courses to scientific graphs) These diverse responsessuggest that the two types of people do in fact visualizethe world in different ways

Figure 9 Example graph problem presented to the scientistsand visual artists


for the Paper Folding and Grain Resolution Tasks as a Function of Profession

Visual Artists Scientists

Task M SD M SD

Paper folding 515 276 714 101Grain resolution 1251 158 1127 255


GENERAL DISCUSSION

The findings in these four studies support an interpre-tation that there are two qualitatively different types ofvisualizers Object visualizers use imagery to constructhigh-quality images of the shapes of individual objectswhereas spatial visualizers use imagery to represent andtransform spatial relations Furthermore the results ofStudies 1 and 2 show that people who score high on spa-tial imagery tasks tend to score below average on objectimagery tasks and vice versa for people who score highon object imagery tasks No such dissociation was foundamong verbalizers Verbalizers did not show a clearlymarked preference for any particular type of imageryTherefore the visualizerndashverbalizer cognitive style di-mension should be revised to include two types of visualizersmdashthose whose imagery is based primarily oncharacteristics of objects and those whose imagery isprimarily spatialmdashin addition to verbalizers

The finding that the two types of visualization are typi-cally mutually exclusive is a puzzle Logically they couldhave been simply two separate dimensions and a givenpersonrsquos score on one need not be correlated with his or herscore on the other But the results presented in Figure 2show that there are in fact two groups of visualizers thosewho excel in spatial imagery and not object imagery andthose with the opposite proclivities It is possible that acompensatory mechanism is at work If one is effective atspatial imagery one may tend to use this skill more fre-quently than (or sometimes in lieu of) object imagery indaily life and vice versa and thus one sort of imagery willbe practiced and used whereas another sort will not Manytasks can be solved using either sort of imagery for ex-ample planning a route can be accomplished using deadreckoning (relying on spatial imagery) or by recalling a se-quence of landmarks (relying on object imagery) How-ever our results have shown that not all tasks can be solvedequally effectively using either sort of imagery

Visual artists response

Scientists response

Figure 10 Example of a visual artistrsquos response (top) and a scientistrsquos re-sponse (bottom) to the graph problem


Previous research (Paivio amp Clark 1991 Paivio ampHarshman 1983 J T E Richardson 1999) on the visualizerndashverbalizer cognitive style dimension sug-gested that male visualizers use abstract imagery to vi-sualize moving objects whereas female visualizers usevivid imagery to generate static images Taking this dis-tinction further J T E Richardson proposed that fe-males are more efficient in the generation and mainte-nance of static visual information whereas males aremore efficient in the dynamic manipulation and trans-formation of such information However the results ofStudies 1 and 2 show that the distinction between spatialand object visualizers cannot be reduced to gender dif-ferences or to differences in processing static versus dy-namic visual information

Rather our results suggest that object and spatial vi-sualizers encode and process mental images in differentways Object visualizers tend to encode images globallyas a single perceptual unit which they process holisti-cally In contrast spatial visualizers tend to encode andprocess images analytically part by part using spatialrelations to arrange and analyze the components Be-cause they allow the part structure of objects to be madeexplicit spatial images may be more flexible and trans-formable which might explain why spatial visualizersexcel in dynamic image transformations In contrast ob-ject images encoded as a single unit are not as easily trans-formed However the time needed to generate and activatesuch an image should not depend on image complexityand this could explain why object visualizers tend to befaster and more accurate in recognition and memory tasks

One might be tempted to characterize the differencebetween object and spatial types of visualizers in termsof the operation of the central executive used in workingmemory (Baddeley 1992) The central executive is thecomponent of working memory that controls attentionand coordinates the activities of verbal and visual-spatialshort-term stores the central executive also allocates re-sources for solving a given problem or suppresses infor-mation that is irrelevant to the current task (Engle Kaneamp Tuholski 1999 OrsquoReilly Braver amp Cohen 1999)From this perspective the central executive of object vi-sualizers might have difficulty suppressing details thatare not relevant for a given task which would hamper theability to abstract the key characteristics of a stimulusHowever the results of Study 3 show that object visual-izers perform well on abstract nonvisual tasks such as averbal similarities task Other studies (Kozhevnikov etal 2002) have also reported that object visualizers scoreas well as spatial visualizers and verbalizers on the quan-titative part of the Scholastic Aptitude Test moreoverobject visualizers are able to successfully use abstractmathematical strategies to solve science problems Suchfindings suggest that the difficulties object visualizershave in dealing with abstract spatial relations do notarise because the central executive is not able to suppressirrelevant information but rather because of their ten-dency to process visual information globally

Could the distinction between object and spatial im-agery simply reflect a difference between concrete andabstract representations Although we cannot rule outthis possibility with confidence our interviews with vi-sual artists indicated that these object visualizers can useabstract mental images that are not based on previouslyseen real objects but rather represent nonexisting ob-jects or a combination of features Moreover Miller(1996) cited several abstract artists (eg Jackson Pol-lock and Mark Rothko) who claimed that their paintingsdid not take themes from the real world but representedan abstract mixture of feelings and emotions with ldquovi-brating strips of colorrdquo Conversely spatial visualizerscan use concrete images that represent objects at specificlocations Indeed to transform most images properlyone needs to attend to the shape and appearance of thespecific object (such as would occur if one is visualizinghow best to pack luggage into a carrsquos trunk or reconfig-ure the furniture in a room)

Neuroimaging may provide solid evidence that thetwo groups do in fact differ qualitatively in the type ofimagery they prefer to use If this hypothesis is correctdifferent patterns of brain activation should arise whilethe two groups perform visual-spatial problem-solvingtasks For instance spatial visualizers should show moreactivation in areas associated with spatial representation(such as the posterior parietal lobes) whereas object visualizers should show more activation in areas associ-ated with the representation of shapes (such as the infe-rior temporal lobes)

In closing we note that if the present findings hold upover the course of further investigation they will havedirect implications for education personnel selectionand professional training If we are correct it will not beuseful to debate whether ldquohigh-imagery individualsrdquo aremore successful in learning than ldquolow-imagery individ-ualsrdquo or whether imagery in general enhances or impairsperformance on cognitive tasks In order to make opti-mal use of the strengths of visual-spatial processing weneed to explore the relationships between different typesof imagery and performance in various domains More-over the question remains as to why people who aregood at object imagery tend not to develop their spatialimagery ability and vice versa One way to grapple withthis issue is to discover whether people can be trained touse their less preferred type of imagery effectively

REFERENCES

Aginsky V Harris C Rensink R amp Beusmans J (1997) Twostrategies for learning a route in a driving simulator Journal of En-vironmental Psychology 17 317-331

Alesandrini K L (1981) Pictorialndashverbal and analyticalndashholisticlearning strategies in science learning Journal of Educational Psy-chology 73 358-368

Ausburn L J amp Ausburn F B (1978) Cognitive styles Some in-formation and implications for instructional design EducationalCommunications amp Technology Journal 26 337-354

Baddeley A [D] (1992) Is working memory working The FifteenthBartlett Lecture Quarterly Journal of Experimental Psychology44A 1-31


Bartlett F C (1932) Remembering A study in experimental and so-cial psychology Cambridge Cambridge University Press

Blajenkova O amp Kozhevnikov M (2002 November) Two types ofimagers The new self-report questionnaire Paper presented at the43rd Annual Meeting of the Psychonomic Society Kansas City

Campos A amp Suerio E (1993) Sex and age differences in visual im-agery vividness Journal of Mental Imagery 17 91-94

Carpenter P Just M A amp Shell P (1990) What one intelligencetest measures A theoretical account for the processing in the RavenProgressive Matrices test Psychological Review 97 404-431

Cohen J D MacWhinney B Flatt M amp Provost J (1993)PsyScope An interactive graphic system for designing and controllingexperiments in the psychology laboratory using Macintosh computersBehavior Research Methods Instruments amp Computers 25 257-271

Collins D W amp Kimura D (1997) A large sex difference on a two-dimensional mental rotation task Behavioral Neuroscience 111845-849

Edwards J E amp Wilkins W (1981) Verbalizerndashvisualizer ques-tionnaire Relationship with imagery and verbalndashvisual ability Jour-nal of Mental Imagery 5 137-142

Ekstrom R B French J W amp Harman H H (1976) Manual forkit of factor referenced cognitive tests Princeton NJ EducationalTesting Service

Engle R W Kane M J amp Tuholski S W (1999) Individual dif-ferences in working memory capacity and what they tell us aboutcontrolled attention general fluid intelligence and functions of theprefrontal cortex In A Miyake amp P Shah (Eds) Models of workingmemory Mechanisms of active maintenance and executive control(pp 102-135) Cambridge Cambridge University Press

Farah M J Hammond K M Levine D N amp Calvanio R(1988) Visual and spatial mental imagery Dissociable systems ofrepresentations Cognitive Psychology 20 439-462

Geary D C Gilger J W amp Elliot-Miller B (1992) Gender dif-ferences in three-dimensional mental rotation A replication Jour-nal of Genetic Psychology 153 115-117

Green K E amp Schroeder D H (1990) Psychometric quality of theverbalizerndashvisualizer questionnaire as a measure of cognitive stylePsychological Reports 66 939-945

Haxby J V Grady C L Horwitz B Ungerleider L GMishkin M Carson R E Herscovitch P Schapiro M B ampRapoport S I (1991) Dissociation of object and spatial visual pro-cessing pathways in human extrastriate cortex Proceedings of theNational Academy of Sciences 88 1621-1625

Hegarty M amp Kozhevnikov M (1999) Types of visualndashspatial rep-resentations and mathematical problem solving Journal of Educa-tional Psychology 91 684-689

Hunt E (1975) Quote the raven Nevermore In L W Gregg (Ed)Knowledge and cognition (pp 129-158) Hillsdale NJ Erlbaum

Kail R Carter P amp Pellegrino J (1979) The locus of sex dif-ferences in spatial ability Perception amp Psychophysics 26 182-186

Keefe J W (1979) Learning style An overview In J W Keefe (Ed)Student learning styles Diagnosing and prescribing programs (pp 1-17) Reston VA National Association of Secondary SchoolPrincipals

Kirby J Moore P amp Shofield N (1988) Verbal and visual learn-ing styles Contemporary Educational Psychology 13 169-184

Kosslyn S M (1994) Image and brain The resolution of the imagerydebate Cambridge MA MIT Press

Kosslyn S M Ganis G amp Thompson W L (2001) Neural foun-dations of imagery Nature Reviews Neuroscience 2 635-642

Kosslyn S M amp Koenig O (1992) Wet mind The new cognitiveneuroscience New York Free Press

Kozhevnikov M Hegarty M amp Mayer R E (2002) Revising thevisualizerverbalizer dimension Evidence for two types of visualiz-ers Cognition amp Instruction 20 47-77

Krutetskii V A (1976) The psychology of mathematical abilities inschoolchildren Chicago University of Chicago Press

Lean C amp Clements M A (1981) Spatial ability visual imageryand mathematical performance Educational Studies in Mathemat-ics 12 267-299

Levine D N Warach J amp Farah M J (1985) Two visual systemsin mental imagery Dissociation of ldquowhatrdquo and ldquowhererdquo in imagery

disorders due to bilateral posterior cerebral lesions Neurology 351010-1018

Lohman D F (1988) Spatial abilities as traits processes and knowl-edge In R J Sternberg (Ed) Advances in the psychology of humanintelligence (pp 181-232) Hillsdale NJ Erlbaum

Marks D F (1972) Visual imagery differences in the recall of pic-tures British Journal of Psychology 64 17-24

Messer S B (1976) Reflectionndashimpulsivity A review PsychologicalBulletin 83 1026-1053

Messick S (1976) Individuality in learning Implications of cognitivestyle and creativity for human development San Francisco Jossey-Bass

Messick S (1984) The nature of cognitive styles Problems and promisein educational practice Educational Psychologist 19 59-74

Miller A (1996) Insights of genius Imagery and creativity in sci-ence and art New York Springer-Verlag

Milner A D amp Goodale M A (1995) The visual brain in actionOxford Oxford University Press

Moses B E (1980 April) The relationship between visual thinkingtasks and problem-solving performance Paper presented at the annualmeeting of the American Education Research Association Boston

OrsquoReilly R C Braver T S amp Cohen J D (1999) A biologicallybased computational model of working memory In A Miyake amp P Shah (Eds) Models of working memory Mechanisms of activemaintenance and executive control (pp 102-135) Cambridge Cam-bridge University Press

Paivio A (1971) Imagery and verbal processes New York HoltRinehart amp Winston

Paivio A amp Clark J M (1991) Static versus dynamic imagery InC Cornoldi amp M A McDaniels (Eds) Imagery and cognition(pp 221-245) New York Springer-Verlag

Paivio A amp Harshman R A (1983) Factor analysis of a question-naire on imagery and verbal habits and skills Canadian Journal ofPsychology 37 461-483

Poltrock S E amp Agnoli F (1986) Are spatial visualization and vi-sual imagery ability equivalent In R J Sternberg (Ed) Advances inthe psychology of human intelligence (pp 255-296) Hillsdale NJErlbaum

Poltrock S E amp Brown P (1984) Individual differences in spatialability Intelligence 8 93-138

Prabhakaran V Smith J A L Desmond J E Glover G H ampGabrieli J D E (1997) Neural substrates of fluid reasoning Anf MRI study of neocortical activation during performance of theRavenrsquos progressive matrices test Cognitive Psychology 33 43-63

Presmeg N C (1986a) Visualization and mathematical giftednessEducational Studies in Mathematics 17 297-311

Presmeg N C (1986b) Visualization in high school mathematics Forthe Learning of Mathematics 63 42-46

Raven J Raven J C amp Court J H (1998) Manual for RavenrsquosProgressive Matrices and Vocabulary Scales Section 4 AdvancedProgressive Matrices Sets I and II 1998 Edition Oxford OxfordPsychologists Press

Richardson A (1977) Verbalizerndashvisualizer A cognitive style di-mension Journal of Mental Imagery 1 109-126

Richardson J T E (1999) Imagery Hove UK Psychology PressRiding R amp Cheema I (1991) Cognitive styles An overview and in-

tegration Educational Psychology 11 193-215Rouw R Kosslyn S amp Hamel R (1997) Detecting high-level and

low-level properties in visual images and visual percepts Cognition63 209-226

Shepard R N amp Metzler J (1971) Mental rotation of three-dimensional objects Science 191 952-954

Snodgrass J G amp Vanderwart M A (1980) Standardized set of260 pictures Norms for name agreement image agreement famil-iarity and visual complexity Journal of Experimental Psychology6 174-215

Strosahl K D amp Ascough J C (1981) Clinical uses of mental im-agery Experimental foundations theoretical misconceptions and re-search issues Psychological Bulletin 89 422-438

Uhl F Goldenberg G Lang W amp Lindinger G (1990) Cere-bral correlates of imagining colours faces and a map II Negativecortical DC potentials Neuropsychologia 28 81-93

Ungerleider L G amp Mishkin M (1982) Two cortical visual systems


In D J Ingle M A Goodale amp R J W Mansfield (Eds) Analysis ofvisual behavior (pp 549-586) Cambridge MA MIT Press

Wechsler D (1999) WASI Wechsler Abbreviated Scale of Intelli-gence San Antonio Psychological Corporation Harcourt Brace

Witkin H A Moore C A Goodenough D R amp Cox P W(1977) Field-dependent and field-independent cognitive styles andtheir implications Review of Educational Research 47 1-64

NOTES

1 The MPI has been extensively used in the mathematics educationliterature (Lean amp Clements 1981 Moses 1980 Presmeg 1986a1986b) and the results consistently show that people can be reliablyclassified as visualizers or verbalizers according to this questionnaire

This research found that the MPI has a high internal reliability and islargely independent of the sample being used

2 Historically a median split criterion has been used to distinguishbetween visualizers and verbalizers on different visualizerndashverbalizerquestionnaires To be consistent with previous research on the visualizerndashverbalizer dimension we defined visualizers and verbalizers in ourstudies in terms of a median split

3 It is interesting to note that according to the results of Kozhevnikovet al (2002) neither visualizers nor verbalizers exhibit a bimodal dis-tribution of verbal ability These results suggest that verbal abilities incontrast to mental imagery abilities do not dissociate into two distinctgroups

4 All pairwise comparisons reported in the present article relied onTukeyrsquos HSD test

APPENDIX AExamples of Problems From the VisualizerndashVerbalizer Cognitive Style Questionnaire

Part IInstructionsBelow are some problems to solve Different people solve these problems in different ways Youmay solve these problems any way you like There is no right or wrong way of doing them Youmay use the page to write down calculations or draw diagrams or if you prefer you may do theproblems in your head After you do the problems we will ask you some questions about howyou solved them

Problem 1In an athletics race Jim is four feet ahead of Tom and Peter is three feet behind Jim How far isPeter ahead of Tom

Problem 2At each of the two ends of a straight path a man planted a tree and then every 5 feet along thepath he planted another tree The length of the path is 15 feet How many trees were planted

Problem 3Jack Paul and Brian all have birthdays on January 1st but Jack is one year older than Paul andJack is three years younger than Brian If Brian is 10 years old how old is Paul

Part IIInstructionsPlease indicate which (if any) of the methods described below you used for solving the problemsof Part I If you used several methods indicate all methods you used If you think that yourmethod(s) for solving the problems was unlike any of those described in this part describe yourmethod giving as many details as possible

Problem 1mdash I wrote mathematical expressions to represent the problemmdash I solved the problem by imagining Jim Peter and Tom running in an athletics racemdash I solved the problem by drawing a diagram representing the relative position of Jim Peter

and Tom in an athletics racemdash I did not use any of the above methodsmdash I attempted the problem in this way

Problem 2mdash I wrote mathematical expressions to represent the problemmdash I solved the problem by imagining the path and the trees and then counting the trees in my mindmdash I solved the problem by drawing a diagram representing the problemmdash I did not use any of the above methodsmdash I attempted the problem in this way

Problem 3mdash I wrote mathematical expressions to represent the problemmdash I solved the problem by drawing a diagram representing the age of Jack Paul and Brianmdash I solved the problem by drawing a chart or table representing the problemmdash I did not use any of the above methodsmdash I attempted the problem in this way

Figure B1 Example item from the Paper Folding Test

(Manuscript received October 14 2003 revision accepted for publication July 22 2004)

APPENDIX BExamples of Items From the Paper Folding Test

In this test you are to imagine the folding and unfolding of pieces of paper [see Figure B1] The fig-ures on the left represent a square piece of paper being folded and the last of these figures has one ortwo small circles drawn on it to show where the paper has been punched One of the five figures on theright shows where the holes will be when the paper is unfolded You are to decide which one of thesefigures is correct



and composed a VerbalizerndashVisualizer QuestionnaireHowever VerbalizerndashVisualizer Questionnaire scoresproved to be only weakly related to the vividness of ex-perienced imagery and to be essentially unrelated toscores on visual-spatial tasks (eg Alesandrini 1981Green amp Schroeder 1990)

At the end of the 1970s research on the visualizerndashverbalizer cognitive style dimension began to appear inthe educational literature On the basis of clinical meth-ods Krutetskii (1976) proposed that individuals can beclassified into groups according to how they processmathematical information The first group the analytictype consists of people who prefer verbal-logical modeswhen attempting to solve problems The second groupthe geometric type consists of those who prefer to useimagery Following Krutetskiirsquos approach Moses (1980)Suwarsono (described in Lean amp Clements 1981) andPresmeg (1986a 1986b) proposed that individuals canbe placed on a continuum that specifies their preferencesfor using imagery while solving mathematical problemsSuwarsono developed the Mathematical Processing In-strument (MPI described in Lean amp Clements 1981) tomeasure an individualrsquos degree of visuality which is thetendency to use visual imagery when solving math prob-lems The questionnaire included a number of simplemathematics problems that could be solved by either vi-sual or analytical methods However its validity wasquestioned because researchers failed to find a clear re-lationship between the degree of visuality and studentsrsquoperformance on either mathematical or spatial abilitytests (Lean amp Clements 1981)

Indeed a host of studies have challenged the validityof the visualizerndashverbalizer dimension Although theverbal subscale of the VerbalizerndashVisualizer Question-naire is related to verbal ability measures (eg Green ampSchroeder 1990 Kirby et al 1988) the visual subscaleis only weakly related to ratings of the vividness of men-tal imagery and is not even moderately related to othervisual-spatial aptitude measures (eg Alesandrini 1981Edwards amp Wilkins 1981 Green amp Schroeder 1990) Amoderate correlation between the visual subscale of theVerbalizerndashVisualizer Questionnaire and spatial visual-ization ability was reported by Kirby et al (1988) butonly when all the items related to dream vividness wereexcluded from the questionnaire Such findings castdoubt on the usefulness of the visualizerndashverbalizer dis-tinction and interest in this dimension has declined overthe past decade In addition the visualizerndashverbalizer di-mension like other cognitive style dimensions suffersfrom the fact that it is not rooted in a more general the-ory of human information processing Instead the dis-tinction is based on intuition and anecdotal observations

In this article we describe a new approach to charac-terizing individual differences in cognitive functioningwhich grows out of a key distinction about informationprocessingmdashnamely the distinction between processingobject properties and processing spatial relations Thisprocessing distinction is rooted in the brain Neuro-

psychological data provide evidence that higher level vi-sual areas of the brain are divided into two functionallyand anatomically distinct pathways the object and thespatial relations pathways (Haxby et al 1991 Kosslyn ampKoenig 1992 Ungerleider amp Mishkin 1982) The objectpathway runs from the occipital lobe down to the inferiortemporal lobe and has been called the ventral systemthis system processes properties of objects such asshape and color The spatial relations pathway runs fromthe occipital lobe up to the posterior parietal lobe and hasbeen called the dorsal system this system processes ob-ject localization and spatial attributes and is also essen-tial for guiding movements

The distinction between perceptual processing of ob-ject properties versus spatial relations extends to visualmental imagery (eg Farah Hammond Levine amp Cal-vanio 1988 Levine Warach amp Farah 1985 Milner ampGoodale 1995) For example Levine et al demonstratedthat lesions in the temporal cortex disrupted performanceof tasks that relied on mental images of objects and theirproperties whereas such lesions did not disrupt spatialimagery in contrast lesions in the posterior parietal cor-tex had the reverse effects (see also Farah et al 1988)Similarly in neuroimaging studies spatial and objectimagery tasks led to very different patterns of brain ac-tivity (Kosslyn Ganis amp Thompson 2001) For exam-ple when participants visualized a route on a map thatthey had memorized prior to the experiment the parietallobes were activated but when participants visualizedfaces or colors the temporal lobes were activated (UhlGoldenberg Lang amp Lindinger 1990) Moreover Koss-lyn and Anderson (summarized in Kosslyn 1994 chap 9)found that the more complex the pattern the more timerequired to form an image of it using spatial imagery incontrast if one encoded the pattern as a single perceptualunit and later visualized it using object imagery the timeto create the image did not depend on image complexity

The distinction between spatial and object imagery isalso supported by findings of gender differences Maleshave performed better than females on a variety of spa-tial orientation and mental rotation tasks (Collins ampKimura 1997 Geary Gilger amp Elliot-Miller 1992 KailCarter amp Pellegrino 1979) whereas females have scoredhigher than males on imagery vividness questionnaires(eg Campos amp Sueiro 1993) Consistent with thesefindings Paivio and Clark (1991) showed that men couldgenerate dynamic images more quickly than women couldbut that women could generate static images more quicklythan men could Moreover the distinction between ob-ject and spatial imagery appears to characterize individ-ual differences Kirby et al (1988) found two visual fac-tors in several of their factor analyses one related tospatial visualization (ie the ability to manipulate andtransform complex spatial images) and the other relatedto vividness of imagerymdashin particular to dream im-agery Hegarty and Kozhevnikov (1999) found that thevisual-spatial representations used by elementary schoolchildren while solving mathematical problems can be re-






STUDY 1



















Test M SD M SD

PFT 649 234 627 221VVIQ 6310 986 6065 881


35

30

25

20

15

10

5

0low PFT

Nu

mb

er o

f p

arti

cip

ants


Spatial imagery








20

16

12

8

4

0

low PFT

Nu

mb

er o

f p

arti

cip

ants


Spatial imagery


Spatial imagery

Nu

mb

er o

f p

arti

cip

ants

20

16

12

8

4

0


Visualizers

Verbalizers

A

B








STUDY 2






M SD M SD M SD












A

B









9

8

7

6

5

4

3

2

1


Pro

po

rtio

n c

orr

ect








14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)



6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties






STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES


























STUDY 1



















Test M SD M SD

PFT 649 234 627 221VVIQ 6310 986 6065 881


35

30

25

20

15

10

5

0low PFT

Nu

mb

er o

f p

arti

cip

ants


Spatial imagery








20

16

12

8

4

0

low PFT

Nu

mb

er o

f p

arti

cip

ants


Spatial imagery


Spatial imagery

Nu

mb

er o

f p

arti

cip

ants

20

16

12

8

4

0


Visualizers

Verbalizers

A

B








STUDY 2






M SD M SD M SD












A

B









9

8

7

6

5

4

3

2

1


Pro

po

rtio

n c

orr

ect








14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)



6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties






STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES

































Test M SD M SD

PFT 649 234 627 221VVIQ 6310 986 6065 881


35

30

25

20

15

10

5

0low PFT

Nu

mb

er o

f p

arti

cip

ants


Spatial imagery








20

16

12

8

4

0

low PFT

Nu

mb

er o

f p

arti

cip

ants


Spatial imagery


Spatial imagery

Nu

mb

er o

f p

arti

cip

ants

20

16

12

8

4

0


Visualizers

Verbalizers

A

B








STUDY 2






M SD M SD M SD












A

B









9

8

7

6

5

4

3

2

1


Pro

po

rtio

n c

orr

ect








14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)



6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties






STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES



























20

16

12

8

4

0

low PFT

Nu

mb

er o

f p

arti

cip

ants


Spatial imagery


Spatial imagery

Nu

mb

er o

f p

arti

cip

ants

20

16

12

8

4

0


Visualizers

Verbalizers

A

B








STUDY 2






M SD M SD M SD












A

B









9

8

7

6

5

4

3

2

1


Pro

po

rtio

n c

orr

ect








14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)



6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties






STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES




























STUDY 2






M SD M SD M SD












A

B









9

8

7

6

5

4

3

2

1


Pro

po

rtio

n c

orr

ect








14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)



6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties






STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES































A

B









9

8

7

6

5

4

3

2

1


Pro

po

rtio

n c

orr

ect








14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)



6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties






STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES




























9

8

7

6

5

4

3

2

1


Pro

po

rtio

n c

orr

ect








14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)



6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties






STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES



























14

12

10

8

6

4

2


Res

po

nse

tim

e (s

ec)



6

5

4

3

2

1

0Local properties

Mea

n s

core

Global properties






STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES

























STUDY 3










Global 2541 1109 2704 927 2919 1161Local 3251 1198 3496 1336 3187 933




Test M SD M SD M SD








STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES



























STUDY 4


A

B
















Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES



































Task M SD M SD



GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES






















GENERAL DISCUSSION




Scientists response









REFERENCES































































NOTES




























REFERENCES































































NOTES















































































NOTES

























NOTES


























Documents

Spatial versus object visualizers: A new characterization ...nmr.mgh.harvard.edu/mkozhevnlab/wp-content/uploads/pdfs/spatial_… · than men could. Moreover, the distinction between