Download as pdf or txt
Download as pdf or txt
You are on page 1of 33

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232082377

Visuospatial Reasoning

Chapter · January 2005

CITATIONS READS
63 2,409

1 author:

Barbara Tversky
Stanford University
292 PUBLICATIONS   17,268 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

SnapAR View project

INSTITUTE 2020 / Artificial and Human Intelligence (Nordic Chapter - Sweden) View project

All content following this page was uploaded by Barbara Tversky on 04 January 2017.

The user has requested enhancement of the downloaded file.


P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

CHAPTER 1 0

Visuospatial Reasoning

Barbara Tversky

Visuospatial reasoning is not simply a mat- uity of spatial reasoning, perhaps because
ter of running to retrieve a fly ball or wend- of the naturalness of mapping abstract el-
ing a way through a crowd or plotting a ements and relations to spatial ones, spatial
path to a destination or stacking suitcases reasoning serves as a basis for abstract knowl-
in a car trunk. It is a matter of deter- edge and inference. The prevalence of spa-
mining whether gears will mesh (Schwartz tial figures of speech in everyday talk attests
& Black, 1 996a), understanding how a car to that: We feel close to some people and
brake works (Heiser & Tversky, 2002), dis- remote from others; we try to keep our spir-
covering how to destroy a tumor without de- its up, to perform at the peak of our pow-
stroying healthy tissue (Duncker, 1 945 ; Gick ers, to avoid falling into depressions, pits,
& Holyoak, 1 980, 1 983 ), and designing a or quagmires; we enter fields that are wide
museum (Suwa & Tversky, 1 997). Perhaps open, struggling to stay on top of things and
more surprising, it is also a matter of decid- not get out of depth. Right now, in this sec-
ing whether a giraffe is more intelligent than tion, we establish fuzzy boundaries for the
a tiger (Banks & Flora, 1 977; Paivio, 1 978), current field of inquiry.
whether one event is later than another
(Boroditsky, 2000), and whether a conclu-
sion follows logically from its’ premises (Bar- Reasoning
wise & Etchemendy, 1 995 ; Johnson-Laird,
1 983 ). All these abstract inferences, and Before the research, a few words about the
more, appear to be based on spatial reason- words. The core of reasoning seems to be, as
ing. Why is that? People begin to acquire Bruner put it years ago, going beyond the in-
knowledge about space and the things in it formation given (Bruner, 1 973 ). Of course,
probably before they enter the world. In- nearly every human activity requires going
deed, spatial knowledge is critical to survival beyond the information given. The simplest
and spatial inference critical to effective sur- recognition or generalization task, as well as
vival. Perhaps because of the (literal) ubiq- the simplest action, require going beyond
2 09
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

21 0 the cambridge handbook of thinking and reasoning

the information given, as according to a far the brain, is another distinction fraught
more ancient saying, you never step into the with complexity and controversy. Evidence
same river twice. Yet many of these tasks brought to bear for one can frequently be
and actions do not feel cognitive, do not feel reinterpreted as evidence for the other (e.g.,
like reasoning. However, the border between Anderson, 1 978). Both representations and
perceptual and cognitive processes may be transformations themselves can each be de-
harder to establish than the borders between composed into representations and transfor-
countries in conflict. Fortunately, psychol- mations. Despite these complications, the
ogy is typically free of territorial politics, so distinction has been a productive way to
establishing boundaries between perception think about psychological processes. In fact,
and cognition is not essential. There seems to it is a distinction that runs deep in hu-
be a tacit understanding as to what counts as man cognition, captured in language as
perceptual and what as cognitive, although subject and predicate and in behavior as
for these categories just as for simpler ones, agent/object and action. The distinction will
such as chairs and cups, the centers of the prove useful here, more than as a way of
category enjoy more consensus than the bor- organizing the literature (for related discus-
ders. Invoking principles or requirements for sion, see Doumas & Hummel, Chap. 4).
the boundaries between perception and cog- It has been argued that the very estab-
nition – consciousness, for example – seems lishment of representations entails inferen-
to entail more controversy than the separa- tial operations. A significant example are
tion into territories. the Gestalt principles of perceptual orga-
How do we go beyond the information nization – grouping by similarity, proxim-
given? Going beyond the information given ity, common fate, and good continuity –
does not necessarily mean adding informa- that contribute to scene segmentation and
tion. One way to go beyond the information representation. These are surely a form of
given is to transform the information given. visuospatial inference. Representations are
This is the concern of the earlier part of the internal translations of external stimuli (or
manuscript. Going beyond the information internal data); as such, they not only elim-
given can also mean transforming the given inate information from the external world,
information, sometimes according rules, as they also add to it and distort it in the ser-
in deductive reasoning. Another way to go vice of interpretation or behavior. Thus, if
beyond the information given is to make in- inference is to be understood in terms of
ferences or judgments from it. Inference and operating on or manipulating information
judgment are the concerns of the later part to draw new conclusions, then it begins in
of the manuscript. Now some more distinc- the periphery of the sensory systems with
tions regarding the visuospatial portion of leveling and sharpening and feature detec-
the title. tion and organization. Nevertheless, the field
has accepted a level of description of repre-
sentations and transformations, one higher
Representations and Transformations
than the levels of sensory and perceptual
Truths are hard to come by in science, but processing; that level is reflected here.
useful fictions, approximate truths, abound.
One of these is the distinction between rep-
resentations and transformations, between
information and processes, between data and Visuospatial
the operations performed on data. Repre-
sentations place limits on transformations What makes visuospatial representations
as they select and structure the informa- visuospatial? Visuospatial transformations
tion captured from the world or the mind. visuospatial? First and foremost, visuospatial
Distinguishing representations and transfor- representations capture visuospatial proper-
mations, even under direct observation of ties of the world. They do this in a way
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 21 1

that preserves, at least in part, the spatial top-down by studying complex reasoning
structural relations of that information (see that has a visuospatial basis. Both ap-
Johnson-Laird, 1 983 ; Pierce in Houser & proaches have been productive. We begin
Kloesel, 1 992). This means that visuospa- with elements.
tial properties that are close or above or be-
low in the world preserve those relations Imagery as Internalized Perception
in the representations. Visual includes static
properties of objects, such as shape, texture, The major research tradition studying visu-
and color, or between objects and reference ospatial reasoning from a bottom-up per-
frames, such as distance and direction. It also spective has been the imagery program, pi-
includes dynamic properties of objects such oneered by Shepard (see Finke & Shepard,
as direction, path, and manner of movement. 1 986; Shepard & Cooper, 1 982; Shepard &
By this account, visuospatial transformations Podgorny, 1 978, for overviews) and Kosslyn
are those that change or use visuospatial in- (1 980, 1 994b), which has aimed to demon-
formation. Many of these properties of static strate parallels between visual perception
and dynamic objects and of spatial relations and visual imagery. There are two basic
between objects are available from modal- tenets of the approach, one regarding rep-
ities other than vision. This may explain resentations and the other regarding opera-
why well-adapted visually impaired individ- tions on representations: that mental images
uals are not disadvantaged at many spatial resemble percepts, and that mental trans-
tasks (e.g., Klatzky, Golledge, Cicinelli, & formations on images resemble observable
Pellegrino, 1 995 ). Visuospatial representa- changes in things in the world, as in men-
tions are regarded as contrasting with other tal rotation, or perceptual processes per-
forms of representation, notably linguistic. formed on things in the world, as in men-
The similarities (e.g., Talmy, 1 983 , 2001 ) tal scanning. Kosslyn (1 994b) has persisted
and differences between visuospatial and in these aims, more recently demonstrat-
linguistic representations provide insights ing that many of the same neural structures
into both. are used for both. Not the demonstrations
Demonstrating properties of internal rep- per se, but the interpretations of them have
resentations and transformations is tricky for met with controversy (e.g., Pylyshyn, 1 978,
another reason; representations are many 1 981 ). In attempting to demonstrate the sim-
steps from either (controlled) input or ilarities between imagery and perception,
(observed) output. For these reasons, the the imagery program has focused both on
study of internal representations and pro- properties of objects and on characteristics
cesses was eschewed not only by behavior- of transformations on objects – the former,
ists, but also by experimentalists. It was one representations, and the latter, operations or
of the first areas to flourish after the so- transformations. The thrust of the research
called Cognitive Revolution of the 1 960s, programs has been to demonstrate that im-
with a flurry of innovative techniques to ages are like internalized perceptions and
demonstrate form and content of internal transformations of images like transforma-
representations and the transformations per- tions of things in the world.
formed on them. It is that research that we
now turn. representations
In the service of demonstrating that im-
ages preserve characteristics of perceptions,
Representations and Transformations
Shepard and his colleagues brought evi-
Visuospatial reasoning can be approached dence from similarity judgments as sup-
bottom-up by studying the elementary rep- port. They demonstrated “second-order
resentations and processes that presumably isomorphisms,” similarity spaces for per-
form the building blocks for more com- ceived and imagined stimuli that have the
plex reasoning. It can also be approached same structure, that is, are fit by the
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

21 2 the cambridge handbook of thinking and reasoning

same underlying multidimensional space Additional support for the claim that images
(Shepard & Chipman, 1 970). For example, preserve properties of percepts comes from
similarity judgments of shapes of cutouts tasks requiring construction of images. Con-
of states conform to the same multidimen- structing images takes longer when there are
sional space as similarity judgments of imag- more parts to the image, even when the
ined shapes of states. The same logic was same figure can be constructed from more
used to show that color is preserved in or fewer parts (Kosslyn, 1 980).
images, as well as configurations of faces The imagery-as-internalized-perception
(see Gordon & Hayward, 1 973 ; Shepard, has proved to be too narrow a view of the
1 975 ). Similar reasoning was used to demon- variety of visuospatial representations. In ac-
strate qualitative differences between pic- counting for syllogistic reasoning, Johnson-
torial and verbal representations in a task Laird (1 983 ) proposed that people form
requiring sequential same–different judg- mental models of the situations described
ments on pairs of schematic faces and names by the propositions (see Johnson-Laird,
(Tversky, 1 969). The pictorial and verbal Chap. 9). Mental models contrast with clas-
similarity of the set of faces was orthogonal sic images in that they are more schematic
so the “different” responses were a clue to than classical images. Entities are repre-
the underlying representation; times to re- sented as tokens, not as likenesses, and
spond “different” are faster when more fea- spatial relations are approximate, almost
tures between the pairs differ. These times qualitative. A similar view was developed
indicated that when participants expected to account for understanding text and dis-
the target (second) stimulus would be a pic- course, that listeners and readers construct
ture, they encoded the first stimulus pictori- schematic models of the situations described
ally, whether it had been a picture of a face or (e.g., Kintsch & van Dijk, 1 983 ; Zwaan &
its name. The converse also held: When the Radvansky, 1 998). As is seen, visuospatial
target stimulus was expected to be a name, mental representations of environments, de-
participants coded the first stimulus verbally vices, and processes are often schematic,
irrespective of its presented modality. even distorted, rather than detailed and ac-
To demonstrate that mental images pre- curate internalized perceptions.
serve properties of percepts, Kosslyn and his
colleagues presented evidence from studies
of reaction times to detect features of imag- transformations
ined objects. One aim is to show that prop- Here, the logic is the same for most research
erties that take longer to verify in percepts programs, and in the spirit of Shepard’s
take longer to identify in images. For ex- notion of second-order isomorphisms: to
ample, when participants were instructed to demonstrate that the times to make par-
construct images of named animals in order ticular visuospatial judgments in memory
to judge whether the animal had a partic- increase with the times to observe or per-
ular part, they verified large parts of ani- form the transformations in the world. The
mals, such as the back of a rabbit, faster dramatic first demonstration was mental ro-
than small but highly associated ones, such tation (Shepard & Metzler, 1 971 ): time to
as the whiskers of a rat. When participants judge whether two figures in different ori-
were not instructed to use imagery to make entations (Figure 1 0.1 ) are the same or
judgments, they verified small associated mirror images correlates linearly with the
parts faster than large ones. When not in- angular distance between the orientations
structed to use imagery, participants used of the figures. The linearity of the relation-
their general world knowledge to make judg- ship – 1 2 points on a straight line – suggests
ments (Kosslyn, 1 976). Importantly, when smooth continuous mental transformation.
the participants explicitly used imagery, they Although linear functions have been ob-
took longer to verify parts, large or small, tained for the original stimuli, strings of
than when they relied on world knowledge. 1 0 cubes with two bends, monotonic, but
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 21 3

creases (Bundesen, Larsen, & Farrell, 1 981 ;


Moyer, 1 973 ). New objects can be con-
structed in imagery, a skill presumably re-
lated to design and creativity (e.g., Finke,
1 990, 1 993 ). In a well-known example,
Finke, Pinker, and Farah (1 989) asked stu-
dents to imagine a capital letter J centered
under an upside-down grapefruit half. Stu-
dents reported “seeing” an umbrella. Even
without instructions to image, certain tasks
spontaneously encourage formation of vi-
sual images. For example, when participants
are asked whether a described spatial array,
such as star above plus, matches a depicted
one, response times indicate that they trans-
form the description into a depiction when
given sufficient time to mentally construct
the situation (Glushko & Cooper, 1 978;
Tversky, 1 975 ).
In the cases of mental rotation, mental
movement, and mental size transformations,
objects or object parts undergo imagined
transformations. There is also evidence that
objects can be mentally scanned in a contin-
Figure 1 0.1 . Mental rotation task of Shepard uous manner. In a popular task introduced
and Metzler (1 971 ). Participants determine
by Kosslyn and his colleagues, participants
whether members of each pair can be rotated
into congruence.
memorize a map of an island with several
landmarks, such as a well and a cave. Partic-
ipants are then asked to conjure an image
of the map and to imagine looking first at
not linear functions are obtained for other
the well, and then mentally scanning from
stimuli, such as letters (Shepard & Cooper,
the well to the cave. The general finding is
1 982). There are myriad possible mental
that mental scanning between two imagined
transformations, only a few of which have
landmarks increases linearly as the distance
been studied in detail. They may be classified
between them increases (Denis & Kosslyn,
into mental transformations on other objects
1 999; Kosslyn, Ball, & Rieser, 1 978; Fig-
and individuals, and mental transformations
ure 1 0.2). The phenomenon holds for spa-
on one’s self. In both cases, the transforma-
tial arrays established by description rather
tions may be global, wholistic, of the entire
than depiction, again, under instructions to
entity, or the transformations may be opera-
form and use images (Denis, 1 996). Mental
tions on parts of entities.
scanning occurs for arrays in depth and for
Mental Transformations on Objects. Ro- flat perspectives on 3 D arrays (Pinker, 1 980).
tation is not the only transformation that In the previous studies, participants were
objects in the world undergo. They can trained to mentally scan, and directed to do
undergo changes of size, shape, color, in- so, leaving open the question of whether
ternal features, position, combination, and it occurs spontaneously. It seems to be in
more. Mental performance of some of these a task requiring direction judgments on re-
transformations has been examined. The membered arrays. Participants first saw an
time to mentally compare the shapes of array of dots. After the dots disappeared,
two rectangles differing in size increases as an arrow appeared on the screen. The task
the actual size difference between them in- was to say whether the arrow pointed to
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

21 4 the cambridge handbook of thinking and reasoning

the previous location of a dot. Reaction visuospatial reasoning has not only percep-
times increased with distance of the arrow tual, but also motor, foundations.
to the likely dot, suggesting that participants
Mental Transformations of Self. That men-
mentally scan from the arrow to answer
tal imagery is both perceptual and motor
the question (Finke & Pinker, 1 982, 1 983 ).
follows from broadening the basic tenets of
Mental scanning may be part of catching
the classical account for imagery. According
or hitting the ball in baseball, tennis, and
to that account, mental processes are inter-
other sports.
nalizations of external or externally driven
Applying Several Mental Transformations. processes, perceptual ones according to the
Other mental transformations on objects are classic view (e.g., in the chapter title of
possible, for example, altering the internal Shepard & Podgorny, 1 978, “Cognitive pro-
configuration of an object. To solve some cesses that resemble perceptual processes”).
problems, such as geometric analogies, peo- The acts of drawing a figure or construct-
ple need to apply more than one mental ing an object entail both perceptual and
transformation to a figure to obtain the an- motor processes working in concert, as do
swer. In most cases, the order of applying the many other activities performed in both real
transformations is optional; that is, first ro- and virtual worlds, from shaking hands to
tating and then moving a figure yields the wayfinding.
same answer as first moving and then rotat- Evidence for mental transformations of
ing. Nevertheless, people have a preferred self, or motor imagery, rather than or in addi-
order for performing a sequence of mental tion to visual imagery has come from a vari-
transformations, and when this order is vi- ety of tasks. The time taken to judge whether
olated, both errors and performance time a depicted hand is right or left correlates
increase (Novick & Tversky, 1 987). What with the time taken to move the hand into
accounts for the preferred order? Although the depicted orientation, as if participants
the mental transformations are performed were mentally moving their hands in or-
in working memory, the determinants of or- der to make the right/left decision (Parsons,
der do not seem to be related to working 1 987b; Sekiyama, 1 982). Mental reorienta-
memory demands. Move is one of the least tion of one’s body has been used to ac-
demanding transformations, and it is typi- count for reaction times to judge whether
cally performed first, whereas rotate is one a left or right arm is extended in pictures
of the most difficult transformations and is of bodies in varying orientations from up-
performed second. Then transformations of right (Parsons, 1 987a). In those studies, re-
intermediate difficulty are performed. What action times depend on the angle of rotation
correlated with the order of applying succes- and the degree of rotation. For some orienta-
sive mental transformations is the order of tions, notably the picture plane, the degree
drawing. Move determines where the pencil of rotation from upright has no effect. This
is to be put on the paper, the first act of draw- allows dissociating mental transformations
ing. Rotate determines the direction in which of other, in this case, mental rotation, from
the first stroke should be taken, and it is the mental transformations of self, in this case,
next transformation. The next transforma- perspective transformations, as the latter do
tions to be applied are those that determine yield increases in reaction times with de-
the size of the figure and its internal details gree of rotation from upright (Zacks, Mires,
(remove, add part, change size, change shading, Tversky, & Hazeltine, 2000; Zacks & Tver-
add part). Although the mental transforma- sky, in press). Imagining one’s self interacting
tions have been tied to perceptual processes, with a familiar object such as a ball or a ra-
the ordering of performing them appears zor, selectively activates left inferior parietal
to be tied to a motor process, the act of and sensorymotor cortex, whereas imagin-
drawing or constructing a figure. This finding ing another interacting with the same objects
presaged later work showing that complex selectively activates right inferior parietal,
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 21 5

same level, and are asked which would spill


first when tilted, they are typically incorrect
from visual imagery. However, if they close
their eyes and imagine tilting each glass un-
til it spills, they correctly tilt a wide glass
less than a narrow one (Schwartz & Black,
1 999). Think of turning a car versus turn-
ing a boat. To imagine making a car turn
right, you must imagine rotating the steer-
ing wheel to the right; however, to imagine
making a boat turn right, you must imagine
moving the rudder lever left. In mental ro-
tation of left and right hands, the shortest
motor path accounts for the reaction times
better than the shortest visual path (Parsons,
1 987b). Mental enactment also facilitates
memory, even for actions described ver-
bally (Englekamp, 1 998). Imagined motor
transformations presumably underlie men-
Figure 1 0.2 . Mental scanning. Participants tal practice of athletic and musical routines,
memorize map and report time to mentally scan techniques known to benefit performance
from one feature to another (after Kosslyn, Ball, (e.g., Richardson, 1 967).
& Rieser, 1 978).
The reasonable conclusion, then, is that
both internalized perceptual transforma-
tions and internalized motor transformations
precuneus, posterior cingulated, and fron-
can serve as bases for transformations in
topolar cortex (Ruby & Decety, 2001 ).
mental imagery. Perceptual and motor im-
There have been claims that visual and
agery can work in concert in imagery, just
motor imagery, or as we have put it, mental
as perceptual and motor processes work in
transformations of object and of self, share
concert in conducting the activities of life.
the same underlying mechanisms (Wexler,
Kosslyn, & Berthoz, 1 998; Wolschlager &
Wolschlager, 1 998). For example, perform- elementary transformations
ing clockwise physical rotations facilitates The imagery-as-internalized-perception ap-
performing clockwise mental rotations, but proach has provided evidence for myriad
interferes with performing counterclock- mental transformations. We have reviewed
wise mental rotations. However, this may evidence for a number of mental per-
be because planning, performing, and mon- ceptual transformations: scanning, change
itoring the physical rotation requires both orientation, location, size, shape, color, con-
perceptual and motor imagery. The work struct from parts, and rearrange parts. Then
of Zacks and collaborators (Zacks et al., we have motor transformations: motions of
2000; Zacks & Tversky, in press) and Ruby bodies, wholes, or parts. This approach has
and Decety (2001 ) suggests that these two the potential to provide a catalog of elemen-
classes of mental transformations are dis- tary mental transformations that are simple
sociable. Other studies directly comparing inferences and that can combine to enable
the two systems supports their dissociability: complex inferences.
The consequences of using one can be differ- The work on inference, judgment, and
ent from the consequences of using the other problem solving will suggest transformations
(Schwartz, 1 999; Schwartz & Black, 1 999; that have yet to be explored in detail. Here,
Schwartz & Holton, 2000). When people we propose a partial catalog of candidates
imagine wide and narrow glasses filled to the for elementary properties of representations
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

21 6 the cambridge handbook of thinking and reasoning

and transformations, expanding from the re- figures, finding simple figures in more com-
search reviewed: plex ones, loads high on spatial visualization,
and performance on mental rotation tasks
r Determining static properties of entities:
naturally loads high on the mental rotation
figure/ground, symmetry, shape, internal factor. As frequently as they are found, these
configuration, size, color, texture, and three abilities do not span the range of spa-
more tial competencies. Yet another partially in-
r Determining relations between static dependent visuospatial ability is visuospatial
entities: memory, remembering the layout of display
◦ With respect to a frame of reference: (e.g., Betrancourt & Tversky, in press). The
location, direction, distance, and more number of distinct spatial abilities as well as
◦ With respect to other entities, com- their distinctness remain controversial (e.g.,
paring size, color, shape, texture, loca- Carroll, 1 993 ; Hegarty & Waller, in press).
tion, orientation, similarity, and other More recent work explores the relations
attributes of spatial abilities to the kinds of men-
r Determining relations of dynamic and static tal transformations that have been distin-
entities: guished, for example, imagining an object
◦ With respect to other entities or rotate versus imagining changing one’s own
to a reference frame: direction, orientation. The mental transformations, in
speed, acceleration, manner, intersec- turn, are often associated with different
tion/collision brain regions (e.g., Zacks, Mires, Tversky, &
r Performing transformations on entities: Hazeltine, 2000; Zacks, Ollinger, Sheridan,
& Tversky, 2002; Zacks & Tversky, in
change location (scanning), change per-
press). Kozhevniikov, Kosslyn, and Shepard
spective, orientation, size, shape; mov-
(in press) proposed that spatial visualiza-
ing wholes, reconfiguring parts, zooming,
tion and mental rotation correspond respec-
enacting
r Performing transformations on self: change tively to the two major visual pathways in
the brain – the ventral “what” pathway un-
of perspective, change of location, derlying object recognition and the dorsal
change of size, shape, reconfiguring parts, “where” pathway underlying spatial loca-
enacting tion. Interestingly, scientists and engineers
score relatively high on mental rotation and
individual differences artists score relatively high on spatial visu-
Yes, people vary in spatial ability. However, alization. Similarly, architects and design-
spatial ability does not contrast with ver- ers score higher than average on embed-
bal ability; in other words, someone can be ded figure tasks but not on mental rota-
good or poor at both, as well as good in one tion (Suwa & Tversky, 2003 ). Associating
and poor in the other. In addition, spatial spatial ability measures to mental transfor-
ability (like verbal ability) is not a single, mations and brain regions are promising
unitary ability. Some of the separate spa- directions toward a systematic account of
tial abilities differ qualitatively; that is, they spatial abilities.
map well onto the kinds of mental transfor-
mations they require. A meta-analysis of a
number of factor analyses of spatial abili-
Inferences
ties yielded three recurring factors (Linn &
Peterson, 1 986): spatial perception, spatial
Inferences from Observing Motion
visualization, and mental rotation. Rod-and-
in Space
frame and water-level tasks load high on spa-
tial perception; this factor seems to reflect To ensure effective survival, in addition to
choice of frame of reference, within an ob- perceiving the world as it is we need to
ject or extrinsic. Performance on embedded also anticipate the world that will be. This
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 21 7

entails inference, inferences from visuospa- From abstract motion paths, adults can
tial information. Some common inferences, make further inferences about what gen-
such as determining where to intersect a erated the motion. In point-light films,
flying object, in particular, a fly ball (e.g., the only thing visible is the movement
McBeath, Shaffer, & Kaiser, 1 995 ), or what of lights placed at motion junctures of,
moving parts belong to the same object (e.g., for example, at the joints of people walk-
Spelke, Vishton, & von Hofsten, 1 995 ) are ing or along branches of bushes swaying.
beyond the scope of the chapter. From sim- From point-light films, people can determine
ple, abstract motions of geometric figures, whether the motion is walking, running, or
people, even babies, infer causal impact and dancing, of men or of women, of friends
basic ontological categories, notably, inani- (Cutting & Kozlowski, 1 977; Johannson,
mate and animate. A striking demonstration 1 973 ; Kozlowski & Cutting, 1 977), of bushes
of perception of causality comes from the or trees (Cutting, 1 986). Surprisingly, from
work of Michotte (1 946/1 963 ; see Buehner point-light displays of action, people are bet-
& Cheng, Chap. 7). Participants watch films ter at recognizing their own movements than
of a moving object, A, coming into contact those of friends, suggesting that motor ex-
with a stationary object, B. When object B perience contributes to perception of mo-
moves immediately, continuing the direc- tion (Prasad, Loula, & Shiffrar, 2003 ). Even
tion of motion suggested by object A, people abstract films of movements of geometric
perceive A as launching B, A as causing B to figures in sparse environments can be inter-
move. When A stops so both A and B are preted as complex social interactions, such
stationary before B begins to move, the per- as chasing and bullying, when they are espe-
ception of a causal connection between A’s cially designed for that (Heider & Simmel,
motion and B’s is lost; their movements are 1 944; Martin & Tversky, 2003 ; Oatley &
seen as independent events. This is a forceful Yuill, 1 985 ) or playing hide-and-seek, but in-
demonstration of immediate perception of terpreting these as intentional actions is not
causality from highly abstract actions, as well immediate; rather, it requires repeated ex-
as of the conditions for perception of causal- posure and possibly instructions to interpret
ity. What seems to underlie the perception the actions (Martin & Tversky, 2003 ).
of causality is the perception that object A Altogether, simply from abstract mo-
acts on object B. Actions on objects turn out tion paths or animated point-light displays,
to be the basis for segmenting events into people can infer several basic ontological
parts (Zacks, Tversky, & Iyer, 2001 ). categories: causal action, animate versus
In Michotte’s (1 946/1 963 ) demonstra- inanimate motion, human motion, motion
tions, the timing of the contact between the of males or females and familiar individuals,
initially moving object and the stationary ob- and social interactions.
ject that begins to move later is critical. If
A stops moving considerably before B be- Mental Spatial Inferences
gins to move, then B’s motion is perceived
to be independent of A’s. B’s movement inferences in real environments
in this case is seen as self-propelled. Self- Every kid who has figured out a short-cut,
propelled movement is possible only for ani- and who has not, has performed a spatial
mate agents, or, more recently in the history inference (for a more recent overview of
of humanity, for machines. Possible paths kids, see Newcombe & Huttenlocher, 2000).
and trajectories of animate motion differ Some of these inferences turn out to be
from those for inanimate motion. Preschool easier than others, often surprisingly. For
children can infer which motion paths are example, in real environments, inferences
appropriate for animate and inanimate mo- about where objects will be in relationship
tion, and even for abstract stimuli; they also to one’s self after imagined movement in the
offer sensible explanations for their infer- environment turn out to be relatively ac-
ences (Gelman, Durgin, & Kaufman, 1 995 ). curate when the imagined movement is a
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

21 8 the cambridge handbook of thinking and reasoning

translation, that is, movement forward or also see that actual motor actions affect and
backward aligned with the body. How- reflect the character of mental ones.
ever, if the imagined movement is rota-
tional, a change in orientation, updating is inferences in mental environments
far less accurate (e.g., Presson & Montello, The section on inference opened with spatial
1 994; Reiser, 1 989). When asked to imagine inferences made in real environments. Of-
walking forward a certain distance, turning, ten, people make inferences about environ-
walking forward another distance, and then ments they are not currently in, for exam-
pointing back to the starting point, partic- ple, when they tell a friend where to how
ipants invariably err by not taking into ac- to get to their house and where to find the
count the turn in their pointing (Klatzky, key when they arrive. For familiar environ-
Loomis, Beall, Chance, & Golledge, 1 998). If ments, people are quite competent at these
they actually move forward, turn, and con- sorts of spatial inferences. The mental repre-
tinue forward, but blindfolded, they point sentations and processes underlying these in-
correctly. Spatial updating in real environ- ferences have been studied for several kinds
ments is more accurate after translation than of environments, notably the immediately
after rotation, and updating after rotation surrounding visible or tangible environment
is selectively facilitated by physical rotation. and the environment too large to be seen
This suggests a deep point about spatial in- at a glance. These two situations, the space
ferences and possibly other inferences, that around the body, and the space the body
in inference, mental acts interact with phys- navigates, seem to function differently in our
ical acts. lives, and consequently, to be conceptualized
differently (Tversky, 1 998).
Spatial updating for the space around the
gesture body was first studied using language alone
Interaction of mind and body in inference is to establish the environments (Franklin &
also revealed in gesture. When people de- Tversky, 1 990). It is significant that lan-
scribe space but are asked to sit on their guage alone, with no specific instructions
hands to prevent gesturing, their speech fal- to form images, was sufficient to establish
ters (Rauscher, Krauss, & Chen, 1 996), sug- mental environments that people could up-
gesting that the acts of gesturing promote date easily and without error. In the proto-
spatial reasoning. Even blind children ges- typical spatial framework task, participants
ture as they describe spatial layouts (Iverson read a narrative that describes themselves
& Goldin-Meadow, 1 997). in a 3 D spatial scene, such as a museum
The nature of spontaneous gestures sug- or hotel lobby (Franklin & Tversky, 1 990;
gests how this happens. When describing Figure 1 0.3 ). The narrative locates and de-
continuous processes, people make smooth, scribes objects appropriate to the scene be-
continuous gestures; when describing dis- yond the observer’s head, feet, front, back,
crete ones, people make jagged, discontin- left, and right (locations chosen randomly).
uous ones (Alibali, Bassok, Solomon, Syc, After participants have learned the scenes
& Goldin-Meadow, 1 999). For space, peo- described by the narratives, they turn to a
ple tend to describe environments as if they computer that describes them as turning in
were traveling through them or as if they the environment so they are now facing a dif-
were viewing them from above. The plane ferent object. The computer then cues them
of their gestures differs in each case, in cor- with direction terms, front, back, head, and so
respondence with the linguistic perspective on, to which the participants respond with
they adopt (Emmorey, Tversky, & Taylor, the name of the object now in that direc-
2000). Earlier, mental transformations that tion. Of interest are the times to respond,
appear to be internalized physical transfor- depending on the direction from the body.
mations, such as those underlying handed- The classical imagery account would predict
ness judgments, were described. Here, we that participants will imagine themselves in
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 21 9

and slowest to the left/right axis, the pat-


tern obtained for the prototypical situation.
When narratives describe observers as reclin-
ing in the scenes, turning from back to side
to front, then no axis of the body is corre-
lated with gravity so times depend on the
asymmetries of the body, and the pattern
changes. Times to retrieve objects in front
and back are then fastest because the per-
ceptual and behavioral asymmetries of the
front/back axis are most important. This is
the axis that separates the world that can be
seen and manipulated from the world that
cannot be seen or manipulated.
By now, dozens of experiments have ex-
amined patterns of response times to system-
atic changes in the described spatial envi-
Figure 1 0.3. Spatial framework situation. ronment (e.g., Bryant, Tversky, & Franklin,
Participants read a narrative describing objects 1 992; Franklin, Tversky, & Coon, 1 992). In
around an observer (after Bryant, Tversky, & one variant, narratives described participants
Franklin, 1 992).
at an oblique angle outside the environ-
ment looking onto a character (or two!) in-
the environment facing the selected object, side the environment; in that case, none
and then imagine themselves turning to face of the axes of the observer’s body is cor-
each cued object in order to retrieve the ob- related with axes of the characters in the
ject in the cued direction. The imagery ac- narrative, and the reaction times to all di-
count predicts that reaction times should be rections are equal (Franklin et al., 1 992).
fastest to the object in front, then to the ob- In another variant, narratives described the
jects 90 degrees away from front, that is, left, scene, a special space house constructed by
right, head, and feet, and slowest to objects NASA, as rotating around the observer in-
1 80 degrees from front, that is, objects to the stead of the observer turning in the scene
back. Data from dozens of experiments fail (Tversky, Kim, & Cohen, 1 999). That con-
to support that account. dition proved difficult for participants. They
Instead, the data conform to the spatial took twice as long to update the environ-
framework theory according to which partic- ment when the environment moved than
ipants construct a mental spatial framework when the observer moved, a case prob-
from extensions of three axes of the body, lematic for pure propositional accounts of
head/feet, front/back, and left/right. Times mental spatial transformations. Once they
to access objects depend on the asymmetries had updated the environment, retrieval
of the body axes, as well as the asymmetries times corresponded to the spatial frame-
of the axes of the world. The front/back and work pattern.
head/feet axes have important perceptual Yet other experiments have varied the
and behavioral asymmetries that are lack- way the environment was conveyed, com-
ing in the left/right axis. The world also has paring description, diagram, 3 D model, and
three axes, only one of which is asymmetric, life (Bryant & Tversky, 1 999; Bryant, Tver-
the axis conferred by gravity. For the upright sky, & Lanca, 2001 ). When the scene is con-
observer, the head/feet axis coincides with veyed by narrative, life, or a 3 D model, the
the axis of gravity, so responses to head and standard spatial framework pattern obtains.
feet should be fastest, and they are. Accord- However, when the scene is conveyed by
ing to the spatial framework account, times a diagram, participants spontaneously adopt
should be next fastest to the front/back axis an external perspective on the environment.
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

220 the cambridge handbook of thinking and reasoning

Their response times are consonant with route perspectives, the mental transforma-
performing a mental rotation of the entire tion needed to understand the location in-
environment rather than performing a men- formation is a transformation of self, an ego-
tal change of their own perspective with re- centric transformation of one’s viewpoint
spect to a surrounding environment (Bryant in an environment. For survey perspectives,
& Tversky, 1 999). Which viewpoint partici- the mental transformation needed to under-
pants adopt, and consequently which mental stand the location information is a transfor-
transformation they perform, can be altered mation of other, a kind of mental scanning
by instructions. When instructed to do so, of an object.
participants will adopt the internal perspec- The prevalence of these two perspectives
tive embedded in the environment in which in imagery, the external perspective viewing
the observer turns from a diagram or the ex- an object or something that can be repre-
ternal perspective from a model in which the sented as an object and the internal perspec-
entire environment is rotated, with the tive viewing an environment from within,
predicted changes in patterns of retrieval is undoubtedly associated with their preva-
times. Similar findings have been reported lence in the experience of living. In life, we
by Huttenlocher and Presson (1 979), Wraga, observe changes in the orientation, size, and
Creem, and Proffitt (2000), and Zacks et al. configuration of objects in the world, and
(in press). scan them for those changes. In life, we move
around in environments, updating our po-
sition relative to the locations of other ob-
route and survey perspectives jects in the environment. We are adept at
When people are asked to describe envi- performing the mental equivalents of these
ronments that are too large to be seen at a actual transformations. There is a natural
glance, they do so from one of two perspec- correspondence between the internal and
tives (Taylor & Tversky, 1 992a, 1 996). In a external perspectives and the mental trans-
route perspective, people address the listener formations of self and other, but the human
as “you,” and take “you” on a tour of the en- mind is flexible enough to apply either trans-
vironment, describing landmarks relative to formation to either perspective. Although
your current position in terms of your front, we are biased to take an external perspec-
back, left, and right. In a survey perspective, tive on objects and mentally transform them
people take a bird’s eye view of the envi- and biased to take an internal perspective on
ronment and describe locations of landmarks environments and mentally transform our
relative to one another in terms of north, bodies with respect to them, we can take
south, east, and west. Speakers (and writers) internal perspectives on objects and ex-
often mix perspectives, contrary to linguists ternal perspectives on events. The mental
who argue that a consistent perspective is world allows perspectives and transforma-
needed both for coherent construction of tions, whereas the physical world does not.
a message and for coherent comprehen- Indeed, conceptualizing a 3 D environment
sion (Taylor & Tversky, 1 992, 1 996; Tversky, that surrounds us and is too large to be seen
Lee, & Mainwaring, 1 999). In fact, con- at once as a small flat object before the eyes,
struction of a mental model is faster when something people, even children, have done
perspective is consistent, but the effect is for eons whenever they produce a map, is
small and disappears quickly during retrieval a remarkable feat of the human mind (cf.
from memory (Lee & Tversky, in press). Tversky, 2000a).
In memory for locations and directions of
landmarks, route and survey statements are
verified equally quickly and accurately irre- effects of language on spatial thinking
spective of the perspective of learning, pro- Speakers of Dutch and other Western lan-
vided the statements are not taken verbatim guages use both route and survey perspec-
from the text (Taylor & Tversky, 1 992b). For tives. Put differently, they can use either a
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 221

relative spatial reference system or an ab- shopping center. Indeed, visuospatial think-
solute (extrinsic) spatial reference system ing is fundamental to the reasoning processes
to describe locations of objects in space. described in other chapters in this handbook,
Relative systems use the spatial relations as discussed in the chapters on similarity (see
“left,” “right,” “front,” and “back” to locate Goldstone & Son, Chap. 2), categorization
objects; absolute or extrinsic systems use (see Medin & Rips, Chap. 3 ), induction (see
terms equivalent to “north,” “south,” “east,” Sloman & Lagnado, Chap. 5 ), analogical rea-
and “west.” A smattering of languages dis- soning (see Holyoak, Chap. 6), causality (see
persed around the world do not describe Buehner & Cheng, Chap. 7), deductive rea-
locations using “left” and “right” (Levinson, soning (see Evans, Chap. 8), mental models
2003 ). Instead, they rely on an absolute sys- (see Johnson-Laird, Chap. 9), and problem
tem, so a speaker of those languages would solving (see Novick & Bassok, Chap. 1 4). For-
refer to your coffee cup as the “north” cup tunately for both reader and author, there is
rather than the one on “your right.” Talk no need to repeat those discussions here.
apparently affects thought. Years of talking
about space using an absolute spatial refer- Distortions as Clues to Reasoning
ence system has fascinating consequences for
thinking about space. For example, speak- Another approach to revealing visuospa-
ers of absolute languages reconstruct a shuf- tial reasoning has been to demonstrate the
fled array of objects relative to extrinsic di- ways that visuospatial representations differ
rections in contrast to speakers of Dutch, systematically from situations in the world.
who reconstruct the array relative to their This approach, which can be called the dis-
own bodies. What’s more, when speakers of tortions program, contrasts with the classi-
languages with only extrinsic reference sys- cal imagery approach. The aim of the distor-
tems are asked to point home after being tions approach is to elucidate the processes
driven hither and thither, they point with involved in constructing and using men-
impressive accuracy, in contrast to Dutch tal representations by showing their conse-
speakers, who point at random. The view quences. The distortions approach has fo-
that the way people talk affects how they cused more on relations between objects
think has naturally aroused controversy (see and relations between objects and refer-
Gleitman & Papafragou, Chap. 26), but is re- ence frames, as these visuospatial properties
ceiving increasing support from a variety of seem to require more constructive processes
tasks and languages (e.g., Boroditsky, 2001 ; than those for establishing representations
Boroditsky, Ham, & Ramscar, 2002). Tak- of objects. Some systematic distortions have
ing a broader perspective, the finding that also been demonstrated in representations
language affects thought is not as startling. of objects.
Language is a tool, such as measuring in-
struments or arithmetic or writing; learn- representations
ing to use these tools also has consequences Early on, the Gestalt psychologists at-
for thinking. tempted to demonstrate that memory for
figures got distorted in the direction of good
figures (see Riley, 1 962). This claim was con-
tested and countered by increasingly sophis-
Judgments ticated empirical demonstrations. The dis-
pute faded in a resolution: visual stimuli
Complex visuospatial thinking is fundamen- are interpreted, sometimes as good figures;
tal to a broad range of human activity, from memory tends toward the interpretations.
providing directions to the post office and So if o – o is interpreted as “eyeglasses,” par-
understanding how to operate the latest ticipants later draw the connection curved,
electronic device to predicting the conse- whereas if it is interpreted as “barbells,”
quences of chemical bonding or designing a they do not (Carmichael, Hogan, & Walter,
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

222 the cambridge handbook of thinking and reasoning

1 93 2). Little noticed is that the effect does Tversky, 1 992, 2000b, 2000c), it is diffi-
not appear in recognition memory (Prentice, cult to clearly attribute error to either rep-
1 95 4). Since then, and relying on the sophis- resentations or processes. Rather the errors
ticated methods developed, there has been seem to be consequences of both, of schema-
more evidence for shape distortion in repre- tized, hence distorted, representations con-
sentations. Shapes that are nearly symmet- structed ad hoc in order to enable specific
ric are remembered or judged as more sym- judgments, such as the direction or distance
metric than they actually are, as if people between pairs of cities. When answering
code nearly symmetric objects as symmetric such questions, it is unlikely that people con-
(Freyd & Tversky, 1 984; McBeath, Schiano, sult a library of “cognitive maps.” Rather, it
& Tversky, 1 997; Tversky & Schiano, 1 989). seems that they draw on whatever informa-
Given that many of the objects and be- tion they have that seems relevant, organiz-
ings that we encounter are symmetric, but ing it for the question at hand. The reliability
are typically viewed at an oblique angle, of the errors under varying judgments makes
symmetry may be a reasonable assump- it reasonable to assume erroneous represen-
tion, although one that is wrong on occa- tations are reliably constructed. Some of the
sion. Size is compressed in memory (Kerst organizing principles that yield systematic
& Howard, 1 978). When portions of ob- errors are reviewed in the next section.
jects are truncated by picture frames, the
objects are remembered as more complete Hierarchical Organization. Dots that are
than they actually were (Intraub, Bender, & grouped together by good continuation, for
Mangels, 1 992). example, parts of the same square out-
lined in dots, are judged to be closer than
representations and transformations: spatial dots that are actually closer but parts of
configurations and cognitive maps separate groups (Coren & Girgus, 1 980).
The Gestalt psychologists also produced An analogous phenomenon occurs in judg-
striking demonstrations that people organize ments of distance between buildings (Hirtle
the visual world in principled ways, even & Jonides, 1 985 ): Residents of Ann Arbor
when that world is a meaningless array (see think that pairs of university (or town) build-
Hochberg, 1 978). Entities in space, espe- ings are closer than actually closer pairs of
cially ones devoid of meaning, are difficult buildings that belong to different groups, one
to understand in isolation, easier to grasp in to the university and the other to the town.
context. People group elements in an array Hierarchical organization of essentially flat
by proximity or similarity or good continua- spatial information also affects accuracy and
tion. One inevitable consequence of percep- time to make judgments of direction. People
tual organizing principles is distorted repre- incorrectly report that San Diego is west of
sentations. Reno. Presumably this error occurs because
Many of the distortions reviewed here people know the states to which the cities
have been instantiated in memory for per- belong and use the overall directions of the
ceptual arrays that do not stand for anything. states to infer the directions between cities in
They have also been illustrated in memory the states (Stevens & Coupe, 1 978). People
for cognitive maps and for environments. As are faster to judge whether one city is east or
such, they have implications for how people north of another when the cities belong to
reason in navigating the world, a visuospa- separate geographic entities than when they
tial reasoning task that people of all ages and are actually farther, but part of the same ge-
parts of the world need to solve. Even more ographic entity (Maki, 1 981 ; Wilton, 1 979).
intriguing, many of these phenomena have A variant of hierarchical organization
analogs in abstract thought. occurs in locating entities belonging to a
For the myriad spatial distortions de- bounded region. When asked to remember
scribed here (and analyzed more fully in the location of a dot in a quadrant, people
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 223

place it closer to the center of the quadrant,


as if they were using general information
about the area to locate entity contained in
it (Huttenlocher, Hedges, & Duncan, 1 991 ;
Newcombe & Huttenlocher, 2000).

Amount of Information. That representa-


tions are constructed on the fly in the ser-
vice of particular judgments seems to be the
case for other distance estimates. Distances
between A and B, say two locations within a
town, are greater when there are more cross
streets or more buildings or more obstacles
or more turns on the route (Newcombe &
Liben, 1 982; Sadalla & Magel, 1 980; Sadalla
& Staplin, 1 980a, 1 980b; Thorndyke, 1 981 ),
as if people mentally construct a represen-
tation of a path from A to B from that in-
formation and use the amount of informa-
tion as a surrogate for the missing exact
distance information. There is an analogous Figure 1 0.4. Alignment. A significant majority
visual illusion: A line appears longer if bi- of participants think the incorrect lower map is
correct. The map has been altered so the United
sected, and longer still with more tick marks
States and Europe and South American and
(at some point of clutter, the illusion ceases Africa are more aligned (after Tversky, 1 981 ).
or reverses).

Perspective. Steinberg regaled generations


of readers of the New Yorker and denizens of Alignment. Hierarchical, perspective, and
dormitory rooms with his maps of views of landmark effects can all be regarded as con-
the world. In the each view, the immediate sequences of the Gestalt principle of group-
surroundings are stretched and the rest of ing. Even groups of two equivalent entities
the world shrunk. The psychological reality can yield distortion. When people are asked
of this genre of visual joke was demonstrated to judge which of two maps is correct, a map
by Holyoak and Mah (1 982). They asked stu- of North and South America in which South
dents in Ann Arbor to imagine themselves America has been moved westward to over-
on either coast and to estimate the distances lap more with North America, or the ac-
between pairs of cities distributed more or tual map, in which the two continents barely
less equally on an east-west axis across the overlap, the majority of respondents pre-
states. Regardless of imagined perspective, fer the former (Tversky, 1 981 ; Figure 1 0.4).
students overestimated the near distances A majority of observers also prefer an in-
relative to the far ones. correct map of the Americas and Europe/
Africa/Asia in which the Americas are
Landmarks. Distance judgments are also moved northward so the United States and
distorted by landmarks. People judge the dis- Europe and South America and Africa are
tance of an undistinguished place to be closer more directly east-west. This phenomenon
to a landmark than vice versa (McNamara has been called alignment; it occurs when
& Diwadkar, 1 997; Sadalla, Burroughs, & people group two spatial entities and then
Staplin, 1 980). Landmark asymmetries vi- remember them more in correspondence
olate elementary metric assumptions, as- than they actually are. It appears not only
sumptions that are more or less realized in in judgments of maps of the world, but also
real space. in judgments of directions between cities, in
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

224 the cambridge handbook of thinking and reasoning

memory for artificial maps, and in memory What might a representation that cap-
for visual blobs. tures all these distortions look like? Nothing
Spatial entities cannot be localized in that can be sketched on a sheet of paper, that
isolation; they can be localized with re- is, coherent in two dimensions. Landmark
spect to other entities or to frames of ref- asymmetries alone disallow that. It does not
erence. When they are coded with respect seem likely that people make these judg-
to another entity, alignment errors are likely. ments by retrieving a coherent prestored
When entities are coded with respect to a mental representation, a “cognitive map,”
frame of reference, rotation errors, described and reading the direction or distance from it.
in the next section, are likely. Rather, it seems that people construct repre-
sentations on the fly, incorporating only the
Rotation. When people are asked to place
information needed for that judgment, the
a cutout of South American in a north-south
relevant region, the specific entities within
east-west frame, they upright it. A large spa-
it. Some of the information may be visuospa-
tial object, such as South America, induces
tial, from experience or from maps, some
its own coordinates along an axis of elon-
may be linguistic. For these reasons, “cog-
gation and an axis parallel to that one. The
nitive collage” seems a more apt metaphor
actual axis of elongation of South America is
than “cognitive map” for whatever represen-
tilted with respect to north-south, and peo-
tations underlie spatial judgment and mem-
ple upright it in memory. Similarly, people
ory (Tversky, 1 993 ). Such representations
incorrectly report that Berkeley is east of
are schematic, they leave out much infor-
Stanford, when it is actually slightly west.
mation and simplify others. Schematization
Presumably this occurs because they up-
occurs for at least two reasons. More exact
right the Bay Area, which actually runs
information may not be known and there-
at an angle with respect to north-south.
fore cannot be represented. More exact in-
This error has been called rotation; it oc-
formation may not even be needed as the
curs when people code a spatial entity with
situation on the ground may fill it in. More
respect to a frame of reference (Tversky,
information may overload working mem-
1 981 ; Figure 1 0.5 ). As for rotation, it ap-
ory, which is notoriously limited. Not only
pears in memory for artificial maps and un-
must the representation be constructed in
interpreted blobs, as well as in memory for
working memory, but also a judgment made
real environments. Others have replicated
on the representation. Schematization may
this error in remembered directions and in
hide incoherence, or it may not be noticed.
navigation (e.g., Glicksohn, 1 994; Lloyd &
Schematization necessarily entails system-
Heivly, 1 987; Montello, 1 991 ; Presson &
atic error.
Montello, 1 994).
Are Spatial Representations Incoherent?. Why do Errors Persist?. It is reasonable to
This brief review has brought evidence for wonder why so many systematic errors per-
distortions in memory and judgment for sist. Some reasons for the persistence of er-
shapes of objects, configurations of objects, ror have already been discussed, that there
and distances and directions between objects may be correctives on the ground, that some
that are a consequence of the organization errors are a consequence of the schematiza-
of the visuospatial information. These are tion processes that are an inherent part of
not errors of lack of knowledge; even ex- memory and information processing. Yet an-
perienced taxi drivers make them (Chase other reason is that the correctives are spe-
& Chi, 1 981 ). Moreover, many of these bi- cific – now I know that Rome is north of
ases have parallels in abstract domains, such Philadelphia – and do not affect or even
as judgments about members of one’s own make contact with the general information
social or political groups relative to judg- organizing principle that generated the error,
ments about members of other groups (e.g., and that serves us well in many situations
Quattrone, 1 986). (e.g., Tversky, 2003 a).
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 225

Figure 1 0.5. Rotation. When asked to place a cutout of South American in a


NSEW framework, most participants upright it, as in the left example (after
Tversky, 1 981 ).

From Spatial to Abstract Reasoning Indeed, spatial reasoning is often studied


in the context of graphics, maps, diagrams,
Visuospatial reasoning does not only entail graphs, and charts. External representations
visuospatial transformations on visuospatial bear similarities to internal representations,
information. Visuospatial reasoning also in- if only because they are creations of the hu-
cludes making inferences from visuospatial man mind, cognitive tools to increase the
information, whether that information is in power of the human mind. They also bear
the mind or in the world. An early demon- formal similarities, in that both internal and
stration was the symbolic distance effect (e.g., external representations are mappings be-
Banks & Flora, 1 977; Moyer, 1 973 ; Paivio, tween elements and relations. External rep-
1 978). The time to judge which of two ani- resentations are constrained by a medium
mals is more intelligent or pleasant is faster and unconstrained by working memory; for
when the entities are farther on the dimen- this reason, inconsistencies, ambiguities, and
sion than when they are closer, as if people incompleteness may be reduced in external
were imagining the entities arrayed on a line representations.
corresponding to the abstract dimension. It
is easier, hence faster, to discriminate larger
Graphics: Elements
distances than smaller ones. Note that a sub-
jective experience of creating and using an The readiness with which people map ab-
image does not necessarily accompany mak- stract information onto spatial information
ing these and other spatial and abstract judg- is part of the reason for the widespread use
ments. Spatial thinking can occur regardless of diagrams to represent and convey ab-
of whether thinkers have the sensation of stract information, from the sublime, the
using an image. So many abstract concepts harmonies of the spheres, rampant in re-
have spatial analogs (for related discussion, ligions spanning the globe, to the mun-
see Holyoak, Chap. 6). dane corporate charts and statistical graphs.
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

226 the cambridge handbook of thinking and reasoning

Graphics, such as these, consist of elements poral, quantitative, and preference informa-
and spatial relations among the elements. In tion, for example, to place stickers for TV
contrast to written (alphabetic) languages, shows they loved, liked, or disliked. Almost
both elements and use of space in graph- all the preschoolers put the stickers on a line,
ics can convey meaning rather directly (e.g., preserving ordinal information. Children in
Bertin, 1 967/1 983 ; Pinker, 1 994; Tversky, the middle school years were able to repre-
1 995 , 2001 ; Winn, 1 989). Elements may sent interval information, but representing
consist of likenesses, such as road signs de- more than ordinal information was unusual
picting picnic tables, falling rocks, or deer. El- for younger children, despite strong manipu-
ements may also be figures of depiction, sim- lations to encourage them. Not only did chil-
ilar to figures of speech: synecdoche, where dren (and adults) spontaneously use spatial
a part represents a whole, common in ideo- relations to represent abstract relations, but
graphic writing, for example, using a ram’s children also showed preferences for the di-
horns to represent a ram; or metonomy, rection of increases in abstract dimensions.
where an association represents an entity or Increases were represented from right to left
action, common in computer menus, such or left to right (irrespective of direction of
as scissors to denote cut text or a trashcan to writing for quantity and preference) or down
allow deletion of files. to up. Representing increasing time or quan-
tity from up to down was avoided. Rep-
Graphics: Relations resenting increases as upward is especially
robust; it affects people’s ability to make
Relations among entities preserve different inferences about second-order phenomena
levels of information. The information pre- such as rate, which is spontaneously mapped
served is reflected in the mapping to space. In to slope, from graphs (Gattis, 2002; Gattis &
some cases, the information preserved is sim- Holyoak, 1 996). The correspondence of up-
ply categorical; space is used to separate en- ward to more, better, and stronger appears
tities belonging to different categories. The in language – on top of the world, rising to
spaces between words, for example, indi- higher levels of platitude – and in gesture –
cate that one set of letters belongs to one thumbs up, high five – as well as in graph-
meaning and another set to another mean- ics. These spontaneous and widespread cor-
ing. Space can also be used to represent ordi- respondences between spatial and abstract
nal information, for example, listing historic relations suggest they are cognitively natural
events in their order of occurrence, groceries (e.g., Tversky, 1 995 a, 2001 ).
by the order of encountering them in the The demonstrations of spontaneous use
supermarket, and companies by their prof- of spatial language and diagrammatic space
its. Space can be used to represent interval to represent abstract relations suggests that
or ratio information, as in many statistical spatial reasoning forms a foundation for
graphs, where the spatial distances among more abstract reasoning. In fact, children
entities reflect their distances on some used diagrammatic space to represent ab-
other dimension. stract relations earlier for temporal relations
than for quantitative ones, and earlier for
spontaneous use of space to represent quantitative relations than for preference re-
abstract relations lations (Tversky et al., 1 991 ). Corrobora-
Even preschool children spontaneously use tive evidence comes from simple spatial and
diagrammatic space to represent abstract in- temporal reasoning tasks, judging whether
formation (e.g., diSessa, Hammer, Sherin, & one object or person is before another. In
Kolpakowski, 1 991 ; Tversky, Kugelmass, & many languages, words for spatial and tem-
Winter, 1 991 ). In one set of studies (Tver- poral relations, such as before, after, and
sky et al., 1 991 ), children from three lan- in between, are shared. That spatial terms
guage communities were asked to place are the foundation for the temporal comes
stickers on paper to represent spatial, tem- from research showing priming of temporal
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 227

perspective from spatial perspective, but not 65 .1 1 .1 2, ca. 5 00–480 b.c.) depicts a spear
vice versa (Boroditsky, 2000). More support that appears in one piece from the desired
for the primacy of spatial thinking for ab- viewing angle, but in three pieces when
stract thought comes from studies of prob- viewed straight on (J. P. Small, personal com-
lem solving (Carroll, Thomas, & Mulhotra, munication, May 27, 2003 ).
1 980). One group of participants was asked The perceptual and cognitive processes
to solve a spatial problem under constraints, and biases that people bring to graphics in-
arranging offices to facilitate communica- clude the catalog of mental representations
tion among key people. Another group was and transformations that was begun earlier.
asked to solve a temporal analog, arranging In that spirit, several researchers have devel-
processes to facilitate production. The solu- oped models for graph understanding, no-
tions to the spatial analog were superior to tably Pinker (1 990), Kosslyn (1 989, 1 994a),
those to the temporal analog. When exper- and Carpenter and Shah (1 998) (see Shah
imenters suggested using a diagram to yet 2003 /2004, for an overview). These mod-
another group solving the temporal analog, els take account of the particular perceptual
their success equaled that of the spatial ana- or imaginal processes that need to be ap-
log group. plied to particular kinds of graphs to yield
the right inferences. Others have taken ac-
diagrams facilitate reasoning count of perceptual and cognitive processing
in the construction of guidelines for design of
Demonstrating that using a spatial diagram
(e.g., Carswell & Wickens, 1 990; Cleveland,
facilitates temporal problem solving also il-
1 985 ; Kosslyn, 1 994a; Tufte, 1 983 , 1 990,
lustrates the efficacy of diagrams in thinking,
1 997; Wainer, 1 984, 1 997). In some cases the
a finding amply supported, even for infer-
design principles are informed by research,
ences entailing complex logic, such as double
but in most they are informed by the au-
disjunctions, although to succeed, diagrams
thors’ educated sensibilities and/or rules of
have to be designed with attention to the
thumb from graphic design.
ways that space and spatial entities are used
to make inferences (Bauer & Johnson-Laird,
Inferences from Diagrams: Structural and
1 993 ). Middle school children studying sci-
Functional. The existence of spontaneous
ence were asked to put reminders on pa-
mapping of abstract information onto spatial
per. Those children who sketched diagrams
does not mean that the meanings of diagrams
learned the material better than those who
are transparent and can be automatically and
did not (Rode & Stern, in press).
easily extracted (e.g., Scaife & Rogers, 1 995 ).
Diagrams can support many different classes
diagrams for communicating of inferences, notably, structural and func-
Many maps, charts, diagrams, and graphs are tional (e.g., Mayer & Gallini, 1 990). Struc-
meant to communicate clearly for travel- tural inferences, or inferences about quali-
ers, students, and scholars, whether they are ties of parts and the relations among them,
professionals or amateurs. To that end, they can be readily made from inspection of a di-
are designed to be clear and easy to com- agram. Distance, direction, size, and other
prehend, and they meet with varying suc- spatial qualities and properties can be “read
cess. Good design takes account of human off” a diagram (Larkin & Simon, 1 987), at
perceptual and cognitive skills, biases, and least with some degree of accuracy. “Reading
propensities. Even ancient Greek vases take off” entails using the sort of mental trans-
account of how they will be seen. Because formations discussed earlier, mental scan-
they are curved round structures, creating ning, mental distance, size, shape, or direc-
a veridical appearance requires artistry. The tion judgments or comparisons. Functional
vase “Achilles and Ajax playing a game” by inferences, or inferences about the behav-
the Kleophrades Painter in the Museum of ior of entities, cannot be readily made from
Metropolitan Art in New York City (Art. inspection of a diagram in the absence of
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

228 the cambridge handbook of thinking and reasoning

additional knowledge or assumptions, often depend on the order of that pulley in the
a consequence of expertise. Spatial infor- mechanics of the system. To answer func-
mation may provide clues to functional in- tional questions, it is as if participants men-
formation, but it is not sufficient for con- tally animate the pulley system in order to
cepts such as force, mass, and friction. generate an answer. Mental animation, how-
Making functional inferences requires link- ever, does not seem to be a continuous pro-
ing perceptual information to conceptual cess in the same way as physical animation.
information; it entails both knowing how to Rather, mental animation seems to be a se-
“read” a diagram, that is, what visuospatial quence of discrete steps, for example, the
features and relations to inspect or trans- first pulley goes clockwise, and the rope goes
form, and knowing how to interpret that vi- under the next pulley to the left of it, so it
suospatial information. must go counterclockwise. That continuous
Structural and functional inferences re- events are comprehended as sequences of
spectively correspond to two senses of men- steps is corroborated by research on segmen-
tal model prevalent in the field. In both cases, tation and interpretation of everyday events,
mental model contrasts with image. In one such as making a bed (Zacks, Tversky, &
sense, a mental model contrasts with an im- Iyer, 2001 ).
age in being more skeletal or abstract. This is It has long been known that domain ex-
the sense used by Johnson-Laird in his book, perts are more adept at functional inferences
Mental Models (1 983 ), in his explication of from diagrams than novices. Experts can
how people solve syllogisms (see Johnson- “see” sequences of organized chess moves
Laird, Chap. 9, and Evans, Chap. 8). Here, in a midgame display (Chase & Simon,
a mental model captures the structural re- 1 973 ; De Groot, 1 965 ). Similarly, experts
lations among the parts of a system. In the in Go (Reitman, 1 976), electricity (Egan
other sense, a mental model contrast with & Schwartz, 1 979), weather (Lowe, 1 989),
an image in having moving parts, in being architecture (Suwa & Tversky, 1 997), and
“runnable” to derive functional or causal in- more make functional inferences with ease
ferences (for related discussion on causal- from diagrams in their domain. Novices
ity, see Buehner and Cheng, Chap. 7, and are no different from experts in structural
on problem solving, see Chi and Ohlsson, inferences.
Chap. 1 6). This is the sense used in another
book also titled Mental Models (Gentner & Inferences from Diagrams of Systems. The
Stevens, 1 983 ). One goal of diagrams is to distinction between structural and func-
instill mental models in the minds of their tional inferences is illustrated by work on
users. To that end, diagrams abstract the es- production and comprehension of diagrams
sential elements and relations of the system for mechanical systems, such as a car brake,
they are meant to convey. As is seen, convey- a bicycle pump, or a pulley system (Heiser
ing structure is more straightforward than & Tversky, 2002; Figure 1 0.6). Participants
conveying function. were asked to interpret a diagram of one of
What does it mean to say that a mental the systems. On the whole, their interpreta-
model is “runnable?” One example comes tions were structural, that is, they described
from research on pulley systems (Hegarty, the relations among the parts of the system.
1 992). Participants were timed to make two Another set of participants was given the
kinds of judgments from diagrams of three- same diagrams, enriched by arrows indicat-
pulley systems. For true-false judgments of ing the sequence of action in the systems.
structural questions, such as “The upper left Those participants gave functional descrip-
pulley is attached to the ceiling,” response tions; that is, they described the step-by-step
times did not depend on which pulley in operation of the system. Reversing the tasks,
the system was queried. For judgments of other groups of participants read structural
functional questions, such as “The upper left or functional descriptions of the systems
pulley goes clockwise,” response times did and produced diagrams of them. Those who
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 229

Figure 1 0.6. Diagrams of a car brake and a bicycle pump (both after Mayer & Gallini, 1 990), and a
pulley system (after Hegarty, 1 992). Diagrams without arrows encouraged structural descriptions and
diagrams with arrows yielded functional descriptions (Heiser and Tversky, in press).

read functional descriptions used arrows in were able to infer functional information
their diagrams far more than those who read from functional text. This finding suggests
structural descriptions. Arrows are an ex- that people with high expertise/ability can
trapictorial device that have many mean- form unitary diagrammatic mental models
ings and functions in diagrams, for exam- of mechanical systems that allow spatial and
ple, pointing, indicating temporal sequence, functional inferences with relative ease, but
causal sequence, and path and manner of people with low expertise/ability have and
motion (Tversky, 2001 ). use diagrammatic mental models for struc-
Expertise came into play in a study of tural information, but rely on propositional
learning rather than interpretation. Partic- representations for functional information.
ipants learned one of the mechanical sys-
tems from a diagram with or without ar- Enriching Diagrams to Facilitate Functional
rows or from structural or functional text. Inferences. As noted, conveying spatial or
They were later tested on both structural and structural information is relatively straight-
functional information. Participants high in forward in diagrams. Diagrams can use space
expertise/ability (self-assessed) were able to to represent space in direct ways that are
infer both structural and functional infor- readily interpreted, as in maps and archi-
mation from either diagram. In contrast, tectural sketches. Conveying information
participants low in expertise/ability could that is not strictly spatial, such as change
derive structural but not functional informa- over time, forces, and kinematics, is less
tion from the diagrams. Those participants straightforward. Some visual conventions for
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

2 30 the cambridge handbook of thinking and reasoning

conveying information about dynamics or in problem solving. For example, a diagram


forces have been developed in comics and containing many arrows moving toward the
in diagrams (e.g., Horn, 1 998; Kunzle, 1 990; center of a display was superior to a diagram
McCloud, 1 994), and many of these con- with static arrows in suggesting the solution
ventions are cognitively compelling. Arrows to the Duncker radiation problem, how to
are a good example. As lines, arrows in- destroy a tumor without destroying healthy
dicate a relationship, a link. As asymmet- tissue (Pedone, Hummel, & Holyoak, 2001 ;
ric lines, they indicate an asymmetric rela- see Holyoak, Chap. 6, Figure 6.4). The fail-
tionship. The arrowhead is compelling as an ure of animations to improve learning itself
indicator of the direction of the asymme- becomes intuitive on further reflection. For
try because of its correspondence to arrow- one thing, animations are often complex, so
heads common as weapons in the world or it is difficult for a viewer to know where to
its correspondence to Vs created by paths look and to make sense of the timing of many
of downward moving water. A survey of moving components. However, even simple
diagrams in science and engineering texts animations, such as the path of a single mov-
shows wide use of extrapictorial diagram- ing circle, are not superior to static graphics
matic devices, such as arrows, lines, brack- (Morrison & Tversky, in press). The second
ets, and insets, although not always consis- reason for the lack of success of anima-
tently (Tversky, Heiser, Lozano, MacKenzie, tions is one reviewed earlier. If people think
& Morrison, in press). As a consequence, of dynamic events as sequences of steps
these devices are not always correctly in- rather than continuous animations, then
terpreted. Some diagrams of paradigmatic presenting change over time as sequences
processes, such as the nitrogen cycle in bi- of steps may make the changes easier
ology or the rock cycle in geology, contain to comprehend.
the same device, typically an arrow, with
multiple senses, pointing or labeling, indi-
Diagrams for Insight
cating movement path or manner, suggest-
ing forces or sequence, in the same diagram. Maps for highways and subways, diagrams
Of course, there is ambiguity in many words for assembly and biology, graphs for eco-
that appear commonly in scientific and other nomics and statistics, and plans for electri-
prose, words that parallel these graphic de- cians and plumbers are designed to be con-
vices, such as line and relationship. Neverthe- cise and unambiguous, although they may
less, the confusion caused by multiple senses not always succeed. Their inventors want to
of diagrammatic devices in interpreting di- communicate clearly and without error. In
agrams suggests that greater care in design contrast are graphics created to be ambigu-
is worthwhile. ous, to allow reinterpretation and discovery.
An intuitive way to visualize change over Art falls into both those categories. Early de-
time is by animations. After all, an animation sign sketches are meant to be ambiguous, to
uses change over time to convey change over commit the designer to only those aspects
time, a cognitively compelling correspon- of the design that are likely not to change,
dence. Despite the intuitive appeal, a sur- and to leave open other aspects. One reason
vey of dozens of studies that have compared for this is fixation; it is hard to “think out
animated graphics to informationally com- of the box.” Visual displays express, suggest,
parable static graphics in teaching a wide more than what they display. That expres-
variety of concepts, physical, mechanical, sion in fact, came from solution attempts to
and abstract, did not find a single example the famous nine-dot problem (see Novick
of superior learning by animations (Tversky, & Bassok, Chap. 1 4, Fig. 1 4.4). Connect all
Morrison, & Betrancourt, 2002). Animations nine dots in a 3 × 3 array using four straight
may be superior for purposes other than lines without lifting the pen from the pa-
learning, for example, in maintaining per- per. The solution that is hard to see is to
spective or in calling attention to a solution extend the lines beyond the “box” suggested
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 2 31

Figure 1 0.7. A sketch by an architect designing a museum. Upon


reinspection, he made an unintentional discovery (Suwa, Tversky, Gero, &
Purcell, 2001 ).

by the 3 × 3 array. The Gestalt psychologists configuration in his own design, he was more
made us aware of the visual inferences the likely to invent a new design idea; similarly,
mind makes without reflection, grouping by when he invented a new design idea, he was
proximity, similarity, good continuation, and more likely to see a new configuration in his
common fate. sketch (Suwa et al., 2001 ; Figure 1 0.7).
Underlying these unintended discoveries
in sketches is a cognitive skill termed con-
inferences from sketches structive perception, which consists of two
Initial design sketches are meant to be am- independent processes: a perceptual one,
biguous for several reasons. In early stages of mentally reorganizing the sketch, and a con-
design, designers often do not want to com- ceptual one, relating the new organization
mit to the details of a solution, only the gen- to some design purpose (Suwa & Tversky,
eral outline, leaving open many possibilities; 2003 ). Participants adept at generating mul-
gradually, they will fill in the details. Per- tiple interpretations of ambiguous sketches
haps more important, skilled designers are excelled at the perceptual ability of finding
able to get new ideas by reexamining their hidden figures and at the cognitive ability of
own sketches, by having a conversation with finding remote meaningful associations, yet
their sketches, bouncing ideas off them (e.g., these two abilities were uncorrelated.
Goldschmidt, 1 994; Schon, 1 983 ; Suwa Expertise affects the kinds of inferences
& Tversky, 1 997; Suwa, Tversky, Gero, & designers are able to make from their
Purcell, 2001 ). They may construct sketches sketches. Novice designers are adept at per-
with one set of ideas in mind, but on later ceptual inferences, such as seeing proxim-
reexamination they see new configurations ity and similarity relations. Expert design-
and relations that generate new design ideas. ers are also adept at functional inferences,
The productive cycle between reexamining such as “seeing” the flow of traffic or the
and reinterpreting is revealed in the protocol changes in light from sketches (Suwa &
of one expert architect. When he saw a new Tversky, 1 997).
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

2 32 the cambridge handbook of thinking and reasoning

Conclusions and Future Directions Acknowledgments

Starting with the elements of visuospatial I am grateful to Phil Johnson-Laird and


representations in the mind, we end with Jeff Zacks for insightful suggestions on a
visuospatial representations created by the previous draft. Preparation of this chapter
mind. Like language, graphics serve to ex- and some of the research reported were
press and clarify individual spatial and ab- supported by Office of Naval Research,
stract concepts. Graphics have an advantage Grant Numbers NOOO1 4-PP-1 -O649,
over language in expressiveness (Stenning N0001 401 1 071 7, and N0001 4021 05 3 4 to
& Oberlander, 1 995 ); graphics use elements Stanford University.
and relations in graphic space to convey el-
ements and relations in real or metaphoric
space. As such, they allow inference based References
on the visuospatial processing that people
have become expert in as a part of their
Alibali, M. W., Bassok, M., Solomon, K. O., Syc,
everyday interactions with space (Larkin & S. E., & Goldin-Meadow, S. (1 999). Illuminat-
Simon, 1 997). As cognitive tools, graphics ing mental representations through speech and
facilitate reasoning, both by externalizing, gesture. Psychological Science, 1 0, 3 27–3 3 3 .
thus offloading memory and processing, and Anderson, J. R. (1 978). Arguments concerning
by mapping abstract reasoning onto spatial representations for mental imagery. Psychologi-
comparisons and transformations. Graphics cal Review, 85 , 249–277.
organize and schematize spatial and abstract Banks, W. P., & Flora, J. (1 977). Semantic and
information to highlight and focus the es- perceptual processes in symbolic comparisons.
sential information. Like language, graphics Journal of Experimental Psychology: Human Per-
serve to convey spatial and abstract concepts ception and Performance, 3 , 278–290.
to others. They make private thoughts pub- Barwise, J., & Etchemendy. (1 995 ). In J. Glasgow,
lic to a community that can then use and N. H. Naryanan, & G. Chandrasekeran (Eds.),
revise those concepts collaboratively. Diagrammatic reasoning: Cognitive and compu-
Of course, graphics and physical and men- tational perspectives (pp. 21 1 –23 4). Cambridge,
tal transformations on them are not identi- MA: MIT Press.
cal to visuospatial representations and rea- Bauer, M. I., & Johnson-Laird, P. N. (1 993 ). How
soning, they are an expression of it. Talk diagrams can improve reasoning. Psychological
Science, 6, 3 72–3 78.
about space and actions in it were probably
among the first uses of language, telling oth- Bertin, J. (1 967/1 983 ). Semiology of graphics: Di-
agrams, networks, maps. (Translated by W. J.
ers how to find their way and what to look for
Berg.) Madison: University of Wisconsin Press.
when they get there. Cognitive tools to pro-
Betrancourt, M., & Tversky, B. (in press). Sim-
mote visuospatial reasoning were among the
ple animations for organizing diagrams. Inter-
first to be invented from tokens for property
national Journal of Human-Computer Studies.
counts, believed to be the precursor of writ-
Beveridge, M., & Parkins, E. (1 987). Visual repre-
ten language (Schmandt-Besserat, 1 992), to
sentation in analogical problem solving. Mem-
trail markers to maps in the sand. Spatial ory and Cognition, 1 5 , 23 0–23 7.
thought, spatial language, and spatial graph-
Boroditsky, L. (2000). Metaphoric structuring:
ics reflect the importance and prevalence Understanding time through spatial
of visuospatial reasoning in our lives, from metaphors. Cognition, 75 , 1 –28.
knowing how to get home to knowing how Boroditsky, L. (2001 ). Does language shape
to design a house, from explaining how to thought?: Mandarin and English speakers’ con-
find the freeway to explaining how the judi- ceptions of time. Cognitive Psychology, 43 , 1 –
cial system works, from understanding basic 23 .
science to inventing new conceptions of the Boroditsky, L., Ham, W., & Ramscar, M. (2002).
origins of the universe. Where do we go from What is universal in event perception? Com-
here? Onward and upward! paring English and Indonesian speakers. In
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 2 33

W. D. Gray & C. D. Schunn (Eds.), Proceed- Configurality, stimulus homogeneity, and ob-
ings of the 2 4th annual meeting of the Cognitive ject integration. Perception and Psychophysics,
Science Society (pp. 1 3 6–1 441 ). Mahwah, NJ: 47, 1 5 7–1 68.
Erlbaum. Chase, W. G., & Chi, M. T. H. (1 981 ). Cognitive
Bruner, J. S. (1 973 ). Beyond the information given: skill: Implications for spatial skill in large-scale
Studies in the psychology of knowing. Oxford, environments. In J. H. Harvey (Ed.), Cognition,
UK: W. W. Norton. social behavior, and the environment (pp. 1 1 1 –
Bryant, D. J., & Tversky, B. (1 999). Mental rep- 1 3 6). Hillsdale, NJ: Erlbaum.
resentations of spatial relations from diagrams Chase, W. G., & Simon, H. A. (1 973 ). The mind’s
and models. Journal of Experimental Psychology: eye in chess. In W. G. Chase (Ed.), Visual infor-
Learning, Memory, and Cognition, 2 5 , 1 3 7–1 5 6. mation processing. New York: Academic Press.
Bryant, D. J., Tversky, B., & Franklin, N. (1 992). Cleveland, W. S. (1 985 ). The elements of graphing
Internal and external spatial frameworks for data. Monterey, CA: Wadsworth.
representing described scenes. Journal of Mem- Coren, S., & Girgus, J. S. (1 980). Principles of
ory and Language, 3 1 , 74–98. perceptual organization and spatial distortion:
Byrant, D. J., Tversky, B., & Lanca, M. (2001 ). The gestalt illusions. Journal of Experimental
Retrieving spatial relations from observation Psychology: Human Performance and Perception,
and memory. In E. van der Zee & U. Nikanne 6, 404–41 2.
(Eds.), Conceptual structure and its interfaces Cutting, J. E. (1 986). Perception with an eye for
with other modules of representation (pp. 1 1 6– motion. Cambridge, MA: Bradford Books/MIT
1 3 9). Oxford: Oxford University Press. Press.
Bundesen, C., & Larsen, A. (1 975 ). Visual trans- Cutting J. E., & Kozlowski L. T. (1 977). Recog-
formation of size. Journal of Experimental nizing friends by their walk: Gait perception
Psychology: Human Perception and Performance, without familiarity cues. Bulletin of the Psycho-
1 , 21 4–220. nomic Society, 9, 3 5 3 –3 5 6.
Bundesen, C., Larsen, A., & Farrell, J. E. (1 981 ). De Groot, A. D. (1 965 ). Thought and choice in
Mental transformations of size and orientation. chess. The Hague: Mouton.
In A. Baddeley & J. Long (Eds.), Attention and Denis, M. (1 996). Imagery and the description
performance IX (pp. 279–294). Hillsdale, NJ: of spatial configurations. In M. de Vega & M.
Erlbaum. Marschark (Eds.), Models of visuospatial cogni-
Carmichael, R., Hogan, H. P., & Walter, A. A. tion (pp. 1 28–1 1 97). New York: Oxford Univer-
(1 93 2). An experimental study of the effect of sity Press.
language on the reproduction of visually per- Denis, M., & Kosslyn, S. M. (1 999). Scanning vi-
ceived forms. Journal of Experimental Psychol- sual mental images: A window on the mind.
ogy, 1 5 , 73 –86. Cahiers de Psychologie Cognitive, 1 8, 409–
Carpenter, P. A., & Shah, P. (1 998). A model 465 .
of the perceptual and conceptual processes in diSessa, A. A., Hammer, D., Sherin, B., &
graph comprehension. Journal of Experimental Kolpakowski, T. (1 991 ). Inventing graphing:
Psychology: Applied, 4, 75 –1 00. Meta-representational expertise in children.
Carroll, J. (1 993 ). Human cognitive abilities: A sur- Journal of Mathematical Behavior, 1 0, 1 1 7–1 60.
vey of factor-analytical studies. New York: Cam- Duncker, K. (1 945 ). On problem solving. Psycho-
bridge University Press. logical Monographs, 5 8, (Whole No. 270).
Carroll, J. M., Thomas, J. C., & Malhotra, A. Egan, D. E., & Schwartz, B. J. (1 979). Chunk-
(1 980). Presentation and representation in de- ing in recall of symbolic drawings. Memory and
sign problem solving. British Journal of Psychol- Cognition, 7, 1 49–1 5 8.
ogy, 71 , 1 43 –1 5 3 . Emmorey, K., Tversky, B., & Taylor, H. A. (2000).
Carswell, C. M. (1 992). Reading graphs: Interac- Using space to describe space: Perspective in
tion of processing requirements and stimulus speech, sign, and gesture. Journal of Spatial Cog-
structure. In B. Burns (Ed.), Percepts, concepts, nition and Computation, 2 , 1 5 7–1 80.
and categories (pp. 605 –645 ). Amsterdam: Englekamp, J. (1 998). Memory for action. Hove,
Elsevier. UK: Psychology Press.
Carswell, C. M., & Wickens, C. D. (1 990). The Finke, R. A. (1 990). Creative imagery. Hillsdale,
perceptual interaction of graphic attributes: NJ: Erlbaum.
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

2 34 the cambridge handbook of thinking and reasoning

Finke, R. A. (1 993 ). Mental imagery and creative Glicksohn, J. (1 994). Rotation, orientation, and
discovery. In B. Roskos-Evoldsen, M. J. Intons- cognitive mapping. American Journal of Psy-
Peterson, & R. E. Anderson (Eds.), Imagery, chology, 1 07, 3 9–5 1 .
creativity, and discovery. Amsterdam: North- Glushko, R. J., & Cooper, L. A. (1 978). Spa-
Holland. tial comprehension and comparison processes
Finke, R. A., & Pinker, S. (1 982). Spontaneous im- in verification tasks. Cognitive Psychology, 1 0,
agery scanning in mental extrapolation. Journal 3 91 –421 .
of Experimental Psychology: Learning, Memory, Goldschmidt, G. (1 994). On visual design
and Cognition, 8, 1 42–1 47. thinking: The vis kids of architecture. Design
Finke, R. A., & Pinker, S. (1 983 ). Directional Studies, 1 5 , 1 5 8–1 74.
scanning of remembered visual patterns. Jour- Gordon, I. E., & Hayward, S. (1 973 ). Second-
nal of Experimental Psychology: Learning, Mem- order isomorphism of internal representations
ory, and Cognition, 9, 3 98–41 0. of familiar faces. Perception and Psychophysics,
Finke, R. A., Pinker, S., & Farah, M. J. (1 989). 1 4, 3 3 4–3 3 6.
Reinterpreting visual patterns in mental im- Hegarty, M. (1 992). Mental animation: Inferring
agery. Cognitive Science, 1 2 , 5 1 –78. motion from static displays of mechanical sys-
Finke, R., & Shepard, R. N. (1 986). Visual func- tems. Journal of Experimental Psychology: Learn-
tions of mental imagery. In K. R. Boff, L. Kauf- ing, Memory, and Cognition, 1 8, 1 084–1 1 02.
man, & J. P. Thomas (Eds.), Handbook of percep- Hegarty, M., & Waller, D. (in press). Individ-
tion and human performance (vol. II, pp. 1 9–5 5 ). ual differences in spatial abilities. In P. Shah
New York: John Wiley & Sons. & A. Miyake (Eds.), Handbook of higher-level
Franklin, N., & Tversky, B. (1 990). Searching visuospatial thinking and cognition. Cambridge:
imagined environments. Journal of Experimen- Cambridge University Press.
tal Psychology: General, 1 1 9, 63 –76. Heider, F., & Simmel, M. (1 944). An experimen-
Franklin, N., Tversky, B., & Coon, V. (1 992). tal study of apparent behavior. American Jour-
Switching points of view in spatial mental mod- nal of Psychology, 5 7, 243 –25 9.
els acquired from text. Memory and Cognition, Heiser, J., & Tversky, B. (2002). Diagrams and
2 0, 5 07–5 1 8. descriptions in acquiring complex systems. Pro-
Freyd, J., & Tversky, B. (1 984). The force of sym- ceedings of the meetings of the Cognitive Science
metry in form perception. American Journal of Society.
Psychology, 97, 1 09–1 26. Heiser, J., Tversky, B., Agrawala, M., & Hanra-
Gattis, M. (2002). Structure mapping in spatial han, P. (2003 ). Cognitive design principles for
reasoning. Cognitive Development, 1 7, 1 1 5 7– visualizations: Revealing and instantiating. In
1 1 83 . Proceedings of the Cognitive Science Society meet-
Gattis, M., & Holyoak, K. J. (1 996). Mapping ings.
conceptual to spatial relations in visual reason- Hirtle, S. C., & Jonides, J. (1 985 ). Evidence of hi-
ing. Journal of Experimental Psychology: Learn- erarchies in cognitive maps. Memory and Cog-
ing, Memory, and Cognition, 2 2 , 1 –9. nition, 1 3 , 208–21 7.
Gelman, R., Durgin, F., & Kaufman, L. Hochberg, J. (1 978). Perception. Englewood
(1 995 ). Distinguishing between animates and Cliffs, NJ: Prentice-Hall.
inanimates: Not by motion alone. In D. Sper- Holyoak, K. J., & Mah, W. A. (1 982). Cognitive
ber, D. Premack, & A. J. Premack (Eds.), Causal reference points in judgments of symbolic mag-
cognition: A multidisciplinary debate (pp. 1 5 0– nitude. Cognitive Psychology, 1 4, 3 28–3 5 2.
1 84). Oxford: Clarendon Press. Horn, R. E. (1 998). Visual language. Bainbridge
Gentner, D., & Stevens, A. (1 983 ). Mental models. Island, WA: MacroVu, Inc.
Hillsdale, NJ: Erlbaum. Houser, N., & Kloesel, C. (1 992). The essential
Gick, M. L., & Holyoak, K. J. (1 980). Analogical Pierce, Vol. 1 and Vol. 2 . Bloomington: Indiana
problem solving. Cognitive Psychology, 1 2 , 3 06– University Press.
355. Huttenlocher, J., Hedges, L. V., & Duncan, S.
Gick, M. L., & Holyoak, K. J. (1 983 ). Schema in- (1 991 ). Categories and particulars: Prototype
duction and analogical transfer. Cognitive Psy- effects in estimating spatial location. Psycholog-
chology, 1 5 , 1 –28. ical Review, 98, 3 5 2–3 76.
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 2 35

Huttenlocher, J., Newcombe, N., & Sandberg, scanning. Journal of Experimental Psychology:
E. H. (1 994). The coding of spatial location in Human Perception and Performance, 4,
young children. Cognitive Psychology, 2 7, 1 1 5 – 47–60.
1 47. Kozhevnikov, M., Kosslyn, S., & Shepard, J. (in
Huttenlocher, J., & Presson, C. C. (1 979). The press). Spatial versus object visualizers: A new
coding and transformation of spatial informa- characterization of visual cognitive style. Mem-
tion. Cognitive Psychology, 1 1 , 3 75 –3 94. ory and Cognition.
Intraub, H., Bender, R. S., & Mangels, J. A. (1 992). Kozlowski, L. T., & Cutting, J. E. (1 977). Recog-
Looking at pictures but remembering scenes. nizing the sex of a walker from a dynamic point
Journal of Experimental Psychology: Learning, light display. Perception and Psychophysics, 2 1 ,
Memory, and Cognition, 1 8, 1 80–1 91 . 5 75 –5 80.
Iverson, J., & Goldin-Meadow, S. (1 997). What’s Kunzle, D. (1 990). The history of the comic strip.
communication got to do with it? Gesture in Berkeley: University of California Press.
children blind from birth. Developmental Psy- Lakoff, G., & Johnson, M. (1 980). Metaphors we
chology, 3 3 , 45 3 –467. live by. Chicago: University of Chicago Press.
Johansson, G. (1 973 ). Visual perception of bio- Larkin, J. H., & Simon, H. A. (1 987). Why a
logical motion and a model for its analysis. Per- diagram is (sometimes) worth ten thousand
ception and Psychophyics, 1 4, 201 –21 1 . words. Cognitive Science, 1 1 , 65 –99.
Johnson-Laird, P. N. (1 983 ). Mental models. Cam- Lee, P. U., & Tversky, B. (in press). Costs to switch
bridge, MA: Harvard. perspective in acquiring but not in accessing
Kerst, S. M., & Howard, J. H. (1 978). Memory spatial information.
psychophysics for visual area and length. Mem- Levinson, S. C. (2003 ). Space in language and
ory and Cognition, 6, 3 27–3 3 5 . cognition: Explorations in cognitive diversity.
Kieras, D. E., & Bovair, S. (1 984). The role of a Cambridge: Cambridge University Press.
mental model in learning to operate a device. Linn, M. C., & Petersen, A. C. (1 986). A
Cognitive Science, 1 1 , 25 5 –273 . meta-analysis of gender differences in spatial
Klatzky, R. L., Golledge, R. G., Cicinelli, J. G., & ability: Implications for mathematics and sci-
Pellegrino, J. W. (1 995 ). Performance of blind ence achievement. In J. S. Hyde & M. C.
and sighted persons on spatial tasks. Journal of Linn (Eds.), The psychology of gender: Advances
Visual Impairment and Blindness, 89, 70–82. through metaanalysis (pp. 67–1 01 ). Baltimore:
Klatzky, R. L., Loomis, J. M., Beall, A. C., Johns Hopkins University Press.
Chance, S. S., & Golledge, R. G. (1 998). Spatial Lowe, R. K. (1 989). Search strategies and infer-
updating of self-position and orientation during ence in the exploration of scientific diagrams.
real, imagined, and virtual locomotion. Psycho- Educational Psychology, 9, 27–44.
logical Science, 9, 293 –298.
Maki, R. H. (1 981 ). Categorization and distance
Kosslyn, S. M. (1 976). Can imagery be dis- effects with spatial linear orders. Journal of
tinguished from other forms of internal Experimental Psychology: Human Learning and
representation? Memory and Cognition, 4, 291 – Memory, 7, 1 5 –3 2.
297.
Martin, B., & Tversky, B. (2003 ). Segmenting am-
Kosslyn, S. M. (1 980). Image and mind. Cam- biguous events. Proceedings of the Cognitive Sci-
bridge, MA: Harvard University Press. ence Society meetings, Boston.
Kosslyn, S. M. (1 989). Understanding charts and Mayer, R. E. (1 998). Instructional technology. In
graphs. Applied Cognitive Psychology, 3 , 1 85 – F. Durso (Ed.), Handbook of applied cognition.
223 . Chichester, England: Wiley.
Kosslyn, S. M. (1 994a). Elements of graph design. Mayer, R. E., & Gallini, J. K. (1 990). When is an
Freeman illustration worth ten thousand words? Journal
Kosslyn, S. M. (1 994b). Image and brain: The res- of Educational Psychology, 82 , 71 5 –726.
olution of the imagery debate. Cambridge, MA: McBeath, M. K., Schiano, D. J., & Tversky,
MIT Press. B. (1 997). Three-dimensional bilateral sym-
Kosslyn, S. M., Ball, T. M., & Rieser, B. J. metry bias in judgments of figural identity
(1 978). Visual images preserve metric spatial and orientation. Psychological Science, 8, 21 7–
information: Evidence from studies of image 223 .
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

2 36 the cambridge handbook of thinking and reasoning

McBeath, M. K., Shaffer, D. M., & Kaiser, M. K. Parsons, L. M. (1 987a). Imagined spatial transfor-
(1 995 ). How baseball outfielders determine mation of one’s body. Journal of Experimental
where to run to catch fly balls. Science, Psychology: General, 1 1 6, 1 72–1 91 .
2 68(5 21 0), 5 69–5 73 . Parsons, L. M. (1 987b). Imagined spatial trans-
McCloud, S. (1 994). Understanding comics. New formations of one’s hands and feet. Cognitive
York: HarperCollins. Psychology, 1 9, 1 92–1 91 .
McNamara, T. P., & Diwadkar, V. A. (1 997). Sym- Pedone, R., Hummel, J. E., & Holyoak, K. J.
metry and asymmetry of human spatial mem- (2001 ). The use of diagrams in analogical prob-
ory. Cognitive Psychology, 3 4, 1 60–1 90. lem solving. Memory and Cognition, 2 9, 21 4–
Michotte, A. E. (1 946/1 963 ). The perception of 221 .
causality. New York: Basic Books. Pinker, S. (1 980). Mental imamgery and the third
Milgram, S., & Jodelet, D. (1 976). Psychological dimension. Journal of Experimental Psychology:
maps of Paris. In H. Proshansky, W. Ittelson, & General, 1 09, 3 5 4–3 71 .
L. Rivlin (Eds.), Environmental psychology (2nd Pinker, S. (1 990). A theory of graph comprehen-
edition, pp. 1 04–1 24). New York: Holt, Rine- sion. In R. Freedle (Ed.), Artificial intelligence
hart and Winston. and the future of testing (pp. 73 –1 26). Hillsdale,
Montello, D. R. (1 991 ). Spatial orientation and NJ: Erlbaum.
the angularity of urban routes: A field study. Pinker, S., Choate, P., & Finke, R. A. (1 984). Men-
Environment and Behavior, 2 3 , 47–69. tal extrapolation in patterns constructed from
Montello, D. R., & Pick, H. L., Jr. (1 993 ). Integrat- memory. Memory and Cognition, 1 2 , 207–21 8.
ing knowledge of vertically-aligned large-scale Pinker, S., & Finke, R. A. (1 980). Emergent
spaces. Environment and Behavior, 2 5 , 45 7– two-dimensional patterns in images rotated in
484. depth. Journal of Experimental Psychology: Hu-
Morrison, J. B., & Tversky, B. (in press). Failures man Perception and Performance, 6, 224–264.
of simple animations to facilitate learning. Prasad, S., Loula, F., & Shiffrar, M. (2003 ). Who’s
Moyer, R. S. (1 973 ). Comparing objects in there? Comparing recognition of self, friend,
memory: Evidence suggesting an internal psy- and stranger movement. Proceedings of the Ob-
chophysics. Perception and Psychophysics, 1 3 , ject Perception and Memory meeting.
1 80–1 84. Prentice, W. C. H. (1 95 4). Visual recognition of
Newcombe, N., & Huttenlocher, J. (2000). Mak- verbally labeled figures. American Journal of
ing space. Cambridge, MA: MIT Press. Psychology, 67, 3 1 5 –3 20.
Newcombe, N., Huttenlocher, J., Sandberg, E., Presson, C. C., & Montello, D. (1 994). Up-
Lee, E., & Johnson, S. (1 999). What do mises- dating after rotational and translational body
timations and asymmetries in spatial judgment movements: Coordinate structure of perspec-
indicate about spatial representation? Journal tive space. Perception, 2 3 , 1 447–1 45 5 .
of Experimental Psychology: Learning, Memory, Pylyshyn, Z. W. (1 973 ). What the mind’s eye tells
and Cognition, 2 5 , 986–996. the mind’s brain: A critique of mental imagery.
Newcombe, N., & Liben, L. S. (1 982). Barrier Psychological Bulletin, 80, 1 –24.
effects in the cognitive maps of children and Pylyshyn, Z. W. (1 979). The rate of “mental ro-
adults. Journal of Experimental Child Psychol- tation” of images: A test of a holistic analogue
ogy, 3 4, 46–5 8. hypothesis.. Memory and Cognition, 7, 1 9–28.
Novick, L. R., & Tversky, B. (1 987). Cognitive Pylyshyn, Z. W. (1 981 ). The imagery debate: Ana-
constraints on ordering operations: The case logue media versus tacit knowledge. Psycholog-
of geometric analogies. Journal of Experimental ical Review, 88, 1 6–45 .
Psychology: General, 1 1 6, 5 0–67. Quattrone, G. A. (1 986). On the perception of a
Oatley, K., & Yuill, N. (1 985 ). Perception of per- group’s variability. In S. Worchel & W. Austin
sonal and inter-personal action in a cartoon (Eds.), The psychology of intergroup relations
film. British Journal of Social Psychology, 2 4, (pp. 25 –48). New York: Nelson-Hall.
1 1 5 –1 24. Rauscher, F. H., Krauss, R. M., & Chen, Y. (1 996).
Paivio, A. (1 978). Mental comparisons involving Gesture, speech, and lexical access: The role of
abstract attributes. Memory and Cognition, 6, lexical movements in speech production. Psy-
1 99–208. chological Science, 7, 226–23 1 .
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 2 37

Reitman, W. (1 976). Skilled perception in GO: Schwartz, D. L., & Black, J. B. (1 996a). Analog
Deducing memory structures from interre- imagery in mental model reasoning: Depictive
sponse times. Cognitive Psychology, 8, 3 3 6– models. Cognitive Psychology, 3 0, 1 5 4–21 9.
3 5 6. Schwartz, D., & Black, J. B. (1 996b). Shuttling
Richardson, A. (1 967). Mental practice: A review between depictive models and abstract rules:
and discussion. Research Quarterly, 3 8, 95 – Induction and feedback. Cognitive Science, 2 0,
1 07. 45 7–497.
Rieser, J. J. (1 989). Access to knowledge of spatial Schwartz, D. L., & Black, T. (1 999). Inferences
structure at novel points of observation. Journal through imagined actions: Knowing by simu-
of Experimental Psychology: Learning, Memory, lated doing. Journal of Experimental Psychology:
and Cognition, 1 5 , 1 1 5 7–1 1 65 . Learning, Memory, and Cognition, 2 5 , 1 1 6–1 3 6.
Riley, D. A. (1 967). Memory for form. In L. Post- Schwartz, D. L., & Holton, D. L. (2000). Tool
man (Ed.), Psychology in the making (pp. 402– use and the effect of action on the imagina-
465 ). New York: Knopf. tion. Journal of Experimental Psychology: Learn-
Rode, C., & Stern, E. (in press). Diagrammatic ing, Memory, and Cognition, 2 6, 1 65 5 –1 665 .
tool use in male and female secondary school Sekiyama, K. (1 982). Kinesthetic aspects of men-
students. Learning and Instruction. tal representations in the identification of left
Ruby, P., & Decety, J. (2001 ). Effect of subjective and right hands. Perception and Psychophysics,
perspective taking during simulation of action: 3 2 , 89–95 .
A PET investigation of agency. Nature Neuro- Shah, P., & Carpenter, P. A. (1 995 ). Conceptual
science, 4, 5 46–5 5 0. limitations in comprehending line graphs. Jour-
Sadalla, E. K., Burroughs, W. J., & Staplin, nal of Experimental Psychology: General, 1 2 4,
L. J. (1 980). Reference points in spatial cog- 43 –61 .
nition. Journal of Experimental Psychology: Hu- Shah, P., Freedman, E. O., & Vekiri, I.
man Learning and Memory, 5 , 5 1 6–5 28. (2003 /2004). Graphical displays. In P. Shah &
Sadalla, E. K., & Magel, S. G. (1 980). The per- A. Miyake (Eds.), Handbook of higher-level vi-
ception of traversed distance. Environment and suospatial thinking and cognition. Cambridge:
Behavior, 1 2 , 65 –79. Cambridge University Press.
Sadalla, E. K., & Montello, D. R. (1 989). Remem- Shah, P., & Miyake, A. (Eds.). (2003 /2004).
bering changes in direction. Environment and Handbook of higher-level visuospatial thinking
Behavior, 2 1 , 3 46–3 63 . and cognition. Cambridge: Cambridge Univer-
Sadalla, E. K., & Staplin, L. J. (1 980a). An in- sity Press.
formation storage model or distance cognition. Shepard, R. N. (1 975 ). Form, formation, and
Environment and Behavior, 1 2 , 1 83 –1 93 . transformation of internal representations. In
Sadalla, E. K., & Staplin, L. J. (1 980b). The per- R. Solso (Ed.), Information processing and
ception of traversed distance: Intersections. En- cognition: The Loyola symposium. Hillsdale, NJ:
vironment and Behavior, 1 2 , 1 67–1 82. Erlbaum.
Scaife, M., & Rogers, Y. (1 996). External Shepard, R. N., & Chipman, S. F. (1 970). Second-
cognition: How do graphical representa- order isomorphism of internal representations:
tions work? International Journal of Human- Shapes of states. Cognitive Psychology, 1 , 1 –1 7.
Computer Studies, 45 , 1 85 –21 3 . Shepard, R. N., & Cooper, L. (1 982). Mental im-
Schiano, D., & Tversky, B. (1 992). Structure strat- ages and their transformation. Cambridge, MA:
egy in viewing simple graphs. Memory and Cog- MIT Press.
nition, 2 0, 1 2–20. Shepard, R. N., & Feng, C. (1 972). A chronomet-
Schmandt-Besserat, D. (1 992). Before writing, vol- ric study of mental paper folding. Cognitive Psy-
ume 1 : From counting to cuneiform. Austin: Uni- chology, 3 , 228–243 .
versity of Texas Press. Shepard, R. N., & Metzler, J. (1 971 ). Mental rota-
Schon, D. A. (1 983 ). The reflective practitioner. tion of three-dimensional objects. Science, 1 71 ,
Harper Collins. 701 –703 .
Schwartz, D. L. (1 999). Physical imagery: Kine- Shepard, R. N., & Podgorny, P. (1 978). Cognitive
matic vs. dynamic models. Cognitive Psychol- processes that resemble perceptual processes.
ogy, 3 8, 43 3 –464. In W. K. Estes (Ed.), Handbook of learning and
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

2 38 the cambridge handbook of thinking and reasoning

cognitive processes (Vol. 5 , pp. 1 89–23 7). Hills- Tufte, E. R. (1 983 ). The visual display of quantita-
dale, NJ: Erlbaum. tive information. Chesire, CT: Graphics Press.
Shiffrar, M., & Freyd, J. J. (1 990). Apparent mo- Tufte, E. R. (1 990). Envisioning information.
tion of the human body. Psychological Science, Cheshire, CT: Graphics Press.
1 , 25 7–264. Tufte, E. R. (1 997). Visual explanations. Cheshire,
Spelke, E. P., Vishton, P. M., & von Hofsten, CT: Graphics Press.
C. (1 995 ). Object perception, object-directed Tversky, B. (1 969). Pictorial and verbal encod-
action, and physical knowledge in infancy. In ing in short-term memory. Perception and Psy-
M. S. Gazzaniga (Ed.), The cognitive neuro- chophysics, 5 , 275 –287.
sciences (pp. 275 –3 40). Cambridge, MA: MIT Tversky, B. (1 975 ). Pictorial encoding in
Press. sentence-picture comparison. Quarterly Jour-
Stenning, K., & Oberlander, J. (1 995 ). A cognitive nal of Experimental Psychology, 2 7, 405 –
theory of graphical and linguistic reasoning: 41 0.
Logic and implementation. Cognitive Science, Tversky, B. (1 981 ). Distortions in memory for
1 9, 97–1 40. maps. Cognitive Psychology, 1 3 , 407–43 3 .
Stevens, A., & Coupe, P. (1 978). Distortions in Tversky, B. (1 985 ). Categories and parts. In C.
judged spatial relations. Cognitive Psychology, Craig & T. Givon (Eds.), Noun classes and
1 0, 422–43 7. categorization (pp. 63 –75 ). Philadelphia: John
Suwa, M., & Tversky, B. (1 997). What architects Benjamins.
and students perceive in their sketches: A pro- Tversky, B. (1 991 ). Spatial mental models. In
tocol analysis. Design Studies, 1 8, 3 85 –403 . G. H. Bower (Ed.), The psychology of learning
Suwa, M., & Tversky, B. (2003 ). Constructive and motivation: Advances in research and theory
perception: A skill for coordinating perception (Vol. 27, pp. 1 09–1 45 ). New York: Academic
and conception. In Proceedings of the Cognitive Press.
Science Society meetings. Tversky, B. (1 992). Distortions in cognitive maps.
Suwa, M., Tversky, B., Gero, J., & Purcell, T. Geoforum, 2 3 , 1 3 1 –1 3 8.
(2001 ). Seeing into sketches: Regrouping parts Tversky, B. (1 993 ). Cognitive maps, cognitive col-
encourages new interpretations. In J. S. Gero, lages, and spatial mental models. In A. U. Frank
B. Tversky, & T. Purcell (Eds.), Visual and spa- & I. Campari (Eds.), Spatial information theory:
tial reasoning in design (pp. 207–21 9). Sydney, A theoretical basis for GIS (pp. 1 4–24). Berlin:
Australia: Key Centre of Design Computing Springer-Verlag.
and Cognition.
Tversky, B. (1 995 a). Cognitive origins of graphic
Talmy, L. (1 983 ). How language structures space. conventions. In F. T. Marchese (Ed.), Un-
In H. L. Pick, Jr., & L. P. Acredolo (Eds.), Spa- derstanding images (pp. 29–5 3 ). New York:
tial orientation: Theory, research and application Springer-Verlag.
(pp. 225 –282). New York: Plenum.
Tversky, B. (1 995 b). Perception and cognition of
Talmy, L. (2001 ). Toward a cognitive semantics. 2D and 3 D graphics. Human factors in comput-
Vol. 1 : Concept-structuring systems. Vol. 2 : Ty- ing systems. New York: ACM.
pology and process in concept structuring. Cam-
Tversky, B. (1 998). Three dimensions of spatial
bridge, MA: MIT Press.
cognition. In M. A. Conway, S. E. Gathercole,
Taylor, H. A., & Tversky, B. (1 992a). Descriptions & C. Cornoldi (Eds.), Theories of memory II
and depictions of environments. Memory and (pp. 25 9–275 ). Hove, East Sussex: Psychologi-
Cognition, 2 0, 483 –496. cal Press.
Taylor, H. A., & Tversky, B. (1 992b). Spatial men- Tversky, B. (2000a). Some ways that maps and di-
tal models derived from survey and route de- agrams communicate. In C. Freksa, W. Brauer,
scriptions. Journal of Memory and Language, 3 1 , C. Habel, & K. F. Wender (Eds.), Spatial cog-
261 –282. nition II: Integration abstract theories, empirical
Taylor, H. A., & Tversky, B. (1 996). Perspective studies, formal models, and powerful applications
in spatial descriptions. Journal of Memory and (pp. 72–79). Berlin, Springer.
Language, 3 5 , 3 71 –3 91 . Tversky, B. (2000b). Levels and structure of cog-
Thorndyke, P. (1 981 ). Distance estimation from nitive mapping. In R. Kitchin & S. M. Freund-
cognitive maps. Cognitive Psychology, 1 3 , 5 26– schuh (Eds.), Cognitive mapping: Past, present
5 5 0. and future (pp. 24–43 ). London: Routledge.
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

visuospatial reasoning 2 39

Tversky, B. (2000c). Remembering spaces. In E. Tversky, B., Morrison, J. B., & Betrancourt, M.
Tulving & F. I. M. Craik (Eds.), Handbook (2002). Animation: Does it facilitate? Interna-
of memory (pp. 3 63 –3 78). New York: Oxford tional Journal of Human-Computer Studies, 5 7,
University Press. 247–262.
Tversky, B. (2001 ). Spatial schemas in depictions. Tversky, B., Morrison, J. B., & Zacks, J. (2002). On
In M. Gattis (Ed.), Spatial schemas and ab- bodies and events. In A. Meltzoff & W. Prinz
stract thought (pp. 79–1 1 1 ). Cambridge, MA: (Eds.), The imitative mind: Development evolu-
MIT Press. tion, and brain bases (pp. 221 –23 2). Cambridge:
Tversky, B. (2003 a). Navigating by mind and by Cambridge University Press.
body. In C. Freksa (Ed.), Spatial cognition III Tversky, B., & Schiano, D. (1 989). Perceptual and
(pp. 1 –1 0). Berlin: Springer-Verlag. conceptual factors in distortions in memory
Tversky, B. (2003 b). Structures of mental spaces: for maps and graphs. Journal of Experimental
How people think about space. Environment Psychology: General, 1 1 8, 3 87–3 98.
and Behavior, 3 5 , 66–80. Ullman, S. (1 996). High-level vision: Object recog-
Tversky, B. (in press). Functional significance of nition and visual cognition. Cambridge, MA:
visuospatial representations. In P. Shah & A. MIT Press.
Miyake (Eds.), Handbook of higher-level visu- van Dijk, T. A., & Kintsch, W. (1 983 ). Strategies of
ospatial thinking. Cambridge: Cambridge Uni- discourse comprehension. New York: Academic
versity Press. Press.
Tversky, B., Heiser, J., Lozano, S., MacKenzie, R., Wainer, H. (1 984). How to display data badly.
& Morrison, J. B. (in press). Enriching anima- The American Statistician, 3 8, 1 3 7–1 47.
tions. In R. Lowe & W. Schwartz (Eds.), Leaving Wainer, H. (1 997). Visual revelations. Graphi-
with animals: Research and Innovation Design. cal tales of fate and deception from Napoleon
Cambridge University Press. Bonaparte to Ross Perot. New York: Springer
Tversky, B., & Hemenway, K. (1 984). Objects, Verlag.
parts, and categories. Journal of Experimental Wexler, M., Kosslyn, S. M., & Berthoz, A. (1 998).
Psychology: General, 1 1 3 , 1 69–1 93 . Motor processes in mental rotation. Cognition,
Tversky, B., Kim, J., & Cohen, A. (1 999). Men- 68, 77–94.
tal models of spatial relations and transforma- Wilton, R. N. (1 979). Knowledge of spatial
tions from language. In C. Habel & G. Rickheit relations: The specification of information used
(Eds.), Mental models in discourse processing and in making inferences. Quarterly Journal of Ex-
reasoning (pp. 23 9–25 8). Amsterdam: North- perimental Psychology, 3 1 , 1 3 3 –1 46.
Holland. Winn, W. (1 989). The design and use of instruc-
Tversky, B., Kugelmass, S., & Winter, A. (1 991 ). tional graphics. In H. Mandl & J. R. Levin
Cross-cultural and developmental trends in (Eds.), Knowledge acquisition from text and pic-
graphic productions. Cognitive Psychology, 2 3 , tures (pp. 1 25 –1 43 ). North Holland: Elsevier.
5 1 5 –5 5 7. Wohlschlager, A., & Wolschlager, A. (1 998).
Tversky, B., & Lee, P. U. (1 998). How space struc- Mental and manual rotation. Journal of Exper-
tures language. In C. Freksa, C. Habel, & K. F. imental Psychology: Human Perception and Per-
Wender (Eds.), Spatial cognition: An interdisci- formance, 2 4, 3 7–41 2.
plinary approach to representation and process- Wraga, M., Creem, S. H., & Proffitt, D. R.
ing of spatial knowledge (pp. 1 5 7–1 75 ). Berlin: (2000). Updating displays after imagined ob-
Springer-Verlag. ject and viewer rotations. Journal of Experimen-
Tversky, B., & Lee, P. U. (1 999). Pictorial and ver- tal Psychology: Learning, Memory, and Cogni-
bal tools for conveying routes. In C. Freksa & tion, 1 5 1 –1 68.
D. M. Mark (Eds.), Spatial information theory: Zacks, J. M., Mires, J., Tversky, B., & Hazeltine, E.
Cognitive and computational foundations of geo- (2000). Mental spatial transformations of ob-
graphic information science (pp. 5 1 –64). Berlin: jects and perspective. Journal of Spatial Cogni-
Springer. tion and Computation, 2 , 3 1 5 –3 3 2.
Tversky, B., Lee, P. U., & Mainwaring, S. (1 999). Zacks, J. M., Ollinger, J. M., Sheridan, M., & Tver-
Why speakers mix perspectives. Journal of sky, B. (2002). A parametric study of mental
Spatial Cognition and Computation, 1 , 3 99– spatial transformations of bodies. Neuroimage,
41 2. 1 6, 85 7–872.
P1 : KOD/FQV-NHX P2: IKB-GFZ-KOD
05 21 8241 76c1 0.xml CB798B/Holyoak 0 5 21 8241 7 6 October 3 1 , 2004 1 4:9

2 40 the cambridge handbook of thinking and reasoning

Zacks, J. M., & Tversky, B. (in press). Multiple structure in events. Journal of Experimental
systems for spatial imagery: Transformations of Psychology: General, 1 3 6, 29–5 8.
objects and perspective. Zwaan, R. A., & Radvansky, G. A. (1 998). Situ-
Zacks, J., Tversky, B., & Iyer, G. (2001 ). ation models in language comprehension and
Perceiving, remembering and communicating memory. Psychological Bulletin, 1 2 3 , 1 62–1 85 .

View publication stats

You might also like