The history of science reveals many notable examples of the power of embodied thought in creativity and imagination. Scientists frequently ac-knowledge that their great discoveries are brought about not through for-mal, purely analytic reasoning, but by “gut feelings” that take shape in the form of rich sensory images and bodily sensations. Albert Einstein, who always recognized his weakness in mathematics, described his creative process in the following way:
The words of the language, as they are written or spoken, do not seem to play any role in my mechanisms of thought. The psychical entities which seem to serve as the elements in thought are certain signs and more or less clear images which can be voluntarily reproduced and combined. . . . The above mental entities are, in my case, of visual and some of muscular type. (Hadamard,1945: 142–3)
Einstein’s embodied thought processes took particular shape in one of his famous thought experiments where he pretended to be a photon moving at the speed of light. He first imagined what he saw and how he felt, and then became a second photon and imagined what he now experienced of the first photon.
Many scientists, like Einstein, have conceded that formal mathematics was useful for communicating their scientific discoveries, but that the lo-cus of their original ideas was rooted in embodied possibilities. Another scientist, Cyril Stanley Smith, purposefully studied graphic arts to better develop his sense of the structure of metals. When he was developing al-loys, Smith wrote “I certainly came to have a very strong feeling of natural understanding, a feeling of how I would behave if I were a certain alloy, a sense of hardness and softness and conductivity and fusability and de-formability and brittleness – all in a curious internal and quite literally sensuous way” (Smith,1981: 359). These embodied images were not inci-dental to Smith’s creative work, for his research depended upon “aesthetic 123
124 Embodiment and Cognitive Science feeling for a balanced structure and a muscular feeling of the interfaces pulling against one another” (Smith,1981: 359).
Similarly to scientific thinking, artistic creativity also involves imagined bodily sensations. One commentator on creativity in the arts suggested, To the pianist and sculptor, the instrumentalist, dancer, surgeon, and manual ar-tisans, they [ideas] burst upon awareness in a kinesthetic form, feeling their way into varying types of muscular experience. Fingers “itch” to play, music “flows”
from the hands, ideas “flow” from the pen. Movement expresses the “idea” of the dancer or orchestra conductor; the almost sensuous desire to model plastic form becomes compulsive in sculpture. (Hutchinson, 1959: 142)
Cognitive scientists rarely acknowledge the embodied nature of higher-order cognition. Gardner (1983) persuasively argued for the concept of kinesthetic thinking as one of the seven forms of multiple intelligence. But the tendency in cognitive science has been to view kinesthetic intelligence as a separate module of mind that does not necessarily interact with other aspects of mind and language. Yet there is an emerging literature in several areas of cognitive science that explicitly demonstrates direct links between higher-order cognition and embodied action, such that embodiment is es-sential to various cognitive functions. This chapter describes this work and its implications for theories of higher-order cognition.
Mental Imagery
The vast majority of scholarship on mental imagery within psychology, as well as many other disciplines, ignores the role that embodiment (e.g., people’s subjective felt experiences of their bodies in action) may play in mental imagery activities. For instance, the classic empirical work on mental imagery investigates possible correspondences between mental im-agery and visual perception (e.g., Finke,1989). Following this trend, most contemporary cognitive psychology textbooks talk about mental imagery only in terms of visual perception (and, to a much lesser extent, audition).
Although there are numerous studies examining people’s kinesthetic and motor imagery, few scholars, until recently, searched for explicit links be-tween kinesthetic activity and mental imagery. The recent work suggests that many aspects of visual and motor imagery share a common represen-tational, and possibly neuropsychological, substrate. As Paivio (1986: 72) once noted, “all mental transformations engage motor processes that derive originally from active manipulation of the referent objects.”
Imagining Human Movement
The embodied approach to mental imagery suggests that the long-noted equivalence between mental imagery and visual perception is not inac-curate, as long as one recognizes that visual perception is shaped by
Imagery, Memory, and Reasoning 125 kinesthetic activity (see Chapter3). My view of mental imagery is quite broad. Following Newton (1996), I use the term “image” to refer to any imaginary instance where one considers what it may be like to move one’s body in a certain way or to think what it may be like to manipulate an object in different ways or what it must feel like to act upon an object in a certain manner, all without actually physically doing what we are currently thinking (Gibbs & Berg,2002). For example, to entertain in our conscious minds the possible feeling that arises when we bend over to grab our left foot shows how people form proprioceptive mental images of an action.
These mental images are not merely perceptual, but kinesthetic in the sense of entertaining what it is like to move our bodies in particular ways.
There is considerable research showing that our ability to imagine our-selves moving in certain ways subsequently influences our actual per-formance of those movements (a few recent studies include Corriss &
Kose, 1998; Hanrahan, Tetreau, & Sarrazin,1995; Hardy & Callow,1999;
Murphy,1990; Smyth & Waller,1998). A recent study, for example, showed that when participants had to copy figures, engaging in visual imagery benefited the drawing of overall form, whereas engaging in kinesthetic imagery facilitated fine-tuned movements of the two hands (Fery,2003).
Although there is much debate over which imagining techniques have the most measurable influence on learning and performance (see Ahsen,1995), these empirical studies show, at the very least, that imagining our bodies moving has some relationship to subsequent real-life human action.
Ideomotor action refers to the fact that just thinking about an action can make people perform the action without any special influence of the will.
Arnold (1946) found that the more vividly a person imaged a movement, the more it occurred. For instance, standing still and imagining falling over, by thinking about both what it would look like and what it would feel like, produces more teetering than does thinking about either the look or feel alone. Asking people to imagine bending their arms, without actu-ally doing so, provokes movement-relevant electrical activity in the arms’
biceps-brachial muscles (Jacobson,1932). Once more, thinking about acting can produce movement without the feeling of doing.
Merely thinking about a kind of person can induce ideomotor mimicry of that person’s behavior (Bargh, Chen, & Burrows,1996). In one study, college students completed a scrambled-sentence task in which some words repeatedly mentioned the idea of aging (e.g., sentences containing the words “wrinkled,” “gray,” “retired,” “wise,” and “old”). Afterward, each participant’s gait was secretly measured as he or she left the exper-iment room. People who earlier read words referring to the elderly in the scrambled-sentence task actually walked out of the room more slowly than did participants who were not presented with words referring to se-nior citizens. Postexperiment interviews suggested that participants were not consciously aware ofhaving been exposed to ideas about the elderly,
126 Embodiment and Cognitive Science or having walked slowly out of the experiment room. But reading words referring to characteristics of the elderly unconsciously prompted people to walk slowly. Interestingly, in a different experiment, when participants were told beforehand that the words mentioned in the sentences were ex-pected by the experimenter to influence their behavior, they did not exhibit the same slowness in walking afterward. In this way, the cognitive influ-ence on action seems to occur outside of participants’ conscious will.
Many other studies have replicated and extended these original findings. One study asked college students to think about professors (Dijksterhuis & van Knippenberg, 1998). Afterward, the students gave more correct answers to questions from the game “Trivial Pursuit” than did participants who did not first think about professors. On the other hand, when participants were first asked to think about soccer hooligans (in a study conducted in Holland), they subsequently were inferior at answering
“Trivial Pursuit” questions compared to control subjects. Related studies have also demonstrated that getting college students to think about old age can facilitate some loss of memory (Dijksterhuis, Bargh, & Miedema, 2001).
In each of the above studies, there is no mention of imagery in describing the locus of the experimental findings. Researchers simply assume that some kind of abstract knowledge is activated (e.g., when a person reads words such as “wrinkled,” “gray,” “wise,” and “old”) and integrated via symbolic mental processes. Yet participants may actually be creating rich mental images while engaged in tasks such as unscrambling words to form grammatical sentences. These mental images are not simply pictorial, but reflect complex interplay between image, somatic response, and meaning.
Much of the work on imagery focuses on people’s deliberate imaginings of physical events. Imagining the location of objects around us also depends on kinesthetic action. For example, one study asked people to memorize the location of objects in a room (Presson & Montello,1994). Afterward, the participants were blindfolded and asked to point to specific objects. People were quick and accurate in doing this. But when participants were then asked to imagine rotating 90 degrees and to point to specific objects again, they were slow and inaccurate. When participants were asked to actually rotate 90 degrees, when blindfolded, and to point to specific objects, they were just as fast and accurate as they were before rotating.
A similar set of results has been reported with children (Rieser &
Rider,1991). Five- and 9-year-old children were tested on their ability to imagine (when at home) their classrooms and to point to objects from var-ious perspectives. When the perspective changes were accompanied by actually changing positions, 5-year-olds were correct in 100% of the trials, and the 9-year-olds 98% of the time. Yet when the children only imag-ined changing perspectives, the 5-year-olds were correct only 29% of the time and the 9-year-olds 27% of the time. A comparison group of adults
Imagery, Memory, and Reasoning 127 showed that when actually changing positions, 100% of the their responses required less than 2 seconds, whereas when they imagined the perspec-tive changes, only 29% of the responses required less than 2 seconds. Once more, important aspects of the imagination are shaped by bodily actions.
People also have the ability to imagine the environmental consequences of their actions. During tool use, people often change position, for exam-ple, when they use a wrench to turn a bolt. One set of studies specifically examined the idea that hand movements can facilitate imagery for object rotations but that the facilitation depends on people’s model of the tool in some situation (Schwartz & Holton,2000). Physically turning a block without vision reduced mental rotation times compared with imagining the same rotations without bodily movement. A second study showed that pulling a string from a spool facilitated participants’ mental rotation of an object sitting on the spool. Overall, people’s imagistic transformations are not dependent on the objective, geometric characteristics of an action.
But people’s imagistic abilities are dependent on their subjective model-ing of the tools that mediate motor action and the environmental conse-quences of that action, and how they can transfer that understanding to new situations.
These representative findings illustrate the importance of embodied ac-tion in how people learn over time to mentally imagine their locaac-tions, and the locations of objects, in the world around them. Thus, imagining the lo-cation of objects in space, a task that many cognitive psychologists view as purely cognitive and divorced from the body, is strongly influenced by body movements. In general, these data support the idea that mental im-agery for different real-world events incorporates embodied information.
Several studies have suggested that the mental representations of overt and covert actions are to a large degree “functionally equivalent” (Hall, Bernoties, & Schmidt,1995; Vogt,1995). For instance, there is a close re-lationship between the time needed to mentally rotate a hand or fist in line with the same orientation as a target picture, and the time needed to physically perform the same rotations (Parsons, 1987b, 1994). More-over, imagined representations of the human body in motion are lim-ited by the same biomechanical factors that constrain real movements (Kourtzi & Shiffrar,1999).
Planning movements, and not just moving covertly or overtly, is the common element underlying embodied action and mental imagery per-formance (Salway & Logie,1995). For example, Johnson (2000) reports the findings from a series of studies showing that motor imagery, or mentally simulated actions, is essential in people’s prospective judgments of the awkwardness in prehension (e.g., people’s judgments for executing differ-ent hand movemdiffer-ents with a dowel). People appear to think ahead about their embodied movements not by activating a completed motor plan, but by planning the simulated action.
128 Embodiment and Cognitive Science People often overestimate their physical abilities when they mentally simulate their possible future actions (Landau, Libkuman, & Wildman, 2002). In one study, a group of participants mentally simulated lifting a heavy object (a refrigerator) and then estimated how much weight they could lift. A different group of people estimated the weight they could lift without first engaging in the mental simulation exercise. People who mentally simulated lifting a heavy object reported being able to lift more weight than did participants in the no-simulation condition. Follow-up studies showed that people who mentally simulated an event, such as lifting a 100-lb. weight, many times beforehand reported that they could lift more weight than did participants who completed fewer simulations.
Furthermore, people who simulated lifting a larger amount of weight es-timated that they could lift more weight than did people who imagined lifting a smaller amount of weight. These findings nicely illustrate how even brief mental simulations of embodied activities shape people’s abili-ties to predict future physical performance.
Predicting the outcome of a future action seems to require some ability to internally represent a model of the situation and then draw a conclusion on the basis of these representational structures. One study investigated whether people’s verbal reports about imagined actions can be explained without appeal to representational structures (van Rooij, Bongers, &
Haselager,2002). While standing in one spot, participants were handed rods of different lengths that they then held at an upward 45-degree angle.
The participants’ task was simply to say whether or not they could use the rod to touch a distant object. Across the series of trials, the rods presented to participants either increased in length and then decreased, decreased in length and then increased, or were of random lengths.
Determining whether a rod can reach an object involves assessing in-formation on the rod’s length and one’s own bodily abilities (i.e., one’s posture, ability to lean forward with feet planted on one spot, arm length, and so on). A traditional representational account would argue that par-ticipants must calculate via some internal standard a comparison between the representations of the rod’s length, the postural possibilities, and the estimated distance to the object. Successful imagined action, therefore, is based on mental calculations that transform these different, separate rep-resentations.
A dynamical systems account, however, maintains that a person’s be-havior is best described at the level of the whole embodied system, as a self-organizing pattern, emerging from the interaction among subsystems.
As with all dynamical accounts of human performance, the emphasis here is on the temporal dynamics of the participants’ behaviors across the dif-ferent trials of the experiment. Difdif-ferential equations are used to show how different potential functions capture the long-term dynamics under-lying the participants’ performance. These potential functions describe an
Imagery, Memory, and Reasoning 129 attractor landscape, which, at different times for a participant, reflects rela-tively stable and unstable states of behavior. Van Rooij et al. tested a specific two-attractor space model that made specific predictions about the relative frequencies of different dynamical patterns, once more, across different se-quences of trials in the experiment (i.e., presentation of rods from shortest to longest length, longest to shortest length, and rods of varying lengths presented randomly).
In fact, the results of van Rooij et al. (2002) showed that several dynamic patterns explained the participants’ performance. Participants tended to give the same categorical responses in the random sequence condition.
This assimilative effect is consistent with the dynamical view that the system tends to cling to the state it resides in. Second, there was an in-verse relationship between rod length and probability of “yes” responses.
This contrastive effect was enhanced when the coupled sequence ran from shorter to longer rods rather than the opposite, exactly what is expected be-cause the multistable region is relatively large here. Finally, three dynamic patterns of hysteresis, critical boundary, and enhanced contrast were all ob-served, to different extents, in each participant’s behavior throughout the trials in the experiment. Most generally, these data are consistent with a dy-namical account in which the participants’ imagined actions arise from the interplay between a control parameter (a parameter that leads the system through various dynamical patterns) and a collective variable governing the entire system. These parameters are not represented internally, but pro-vide an “imagining landscape” that is an emergent property of the entire embodied system. Van Rooij et al. argue that it is difficult to imagine how a traditional representational theory could explain the dynamical patterns observed in participants’ task-behaviors, given the complexity of having to integrate different internal mechanisms that are usually postulated for each finding (i.e., the problem of integrating hysteresis and enhanced contrast within a single mechanism). Nonetheless, the complex patterns in partic-ipants’ performance can be explained within a more general dynamical model of self-organizing behavior. For the present purposes, a dynamical model also gives proper acknowledgment to the body’s role in cognitive behavior, such as that used to momentarily imagine different human ac-tions. This work, more generally, demonstrates how a dynamical model of human behavior can scale up to explain higher-order cognitive behavior.
Motoric Processes in Mental Imagery
It may not be terribly surprisingly that people are able to mentally imag-ine their bodies in action, and sometimes use their specific embodied ex-periences when imagining physical events in the real world. But some very interesting studies have more directly explored the links between motor processes and visual mental imagery abilities. For example, the transformation of mental images, as done in the classic Cooper and
130 Embodiment and Cognitive Science Shepard (1982) task, depends on motor processes (Wexler et al., 1998;
Wohlschlager & Wohlschlager,1998). “Visuomotor anticipation is the en-gine that drive mental rotation” (Wexler et al., 1998: 79). Similar mech-anisms may underlie both visual image transformation and the
Wohlschlager & Wohlschlager,1998). “Visuomotor anticipation is the en-gine that drive mental rotation” (Wexler et al., 1998: 79). Similar mech-anisms may underlie both visual image transformation and the