Читать книгу Learning Bodies - Группа авторов - Страница 22

We will now turn from the philosophical analysis to a couple of studies in cognitive psychology that corroborate the theory of embodied cognition.

In the first study, published in 1971 by Shepard and Metzler, subjects had to decide whether two geometrical figures were identical to each other, apart from potential differences in their angular orientation (Fig 1a), or whether they were mirror images of each other (Fig 1b). By measuring reaction time, Shepard and Metzler were able to demonstrate that the time taken by subjects to reach a decision increased as a linear function of the angular difference between the figures: the greater the angular difference, the longer the reaction time (50%, to be precise). At first, this finding might not seem surprising. If you were presented with two such figures (on separate pieces of paper), you would probably rotate one of them until it visually matched – or turned out not to match – the other.

Figure 1

The interesting part of Shepard and Metzler’s experiment is that the subjects were not actually manipulating the figures but merely looking at them presented on a screen. The interpretation seemed quite straightforward: the subjects were rotating the figures in their heads. Hence, the title of their paper came to be: “Mental Rotation of Three-Dimensional Objects”. According to Shepard and Metzler, the subjects were presumed to encode one of the figures as a mental representation which was then “virtually” rotated to test whether it could overlay the other. That is, the subjects were hypothesized to be performing a mental visuospatial transformation. Recent studies have cast doubt on the completeness of this explanation. If visuospatial transformation is all there is to rotation, we would expect this task to be mediated by areas in the brain that are normally associated with visuospatial processing. To be sure, such areas do “light up” when subjects are scanned while they perform the task, but they are not the only active areas during the task and, perhaps, not even the areas critical for the “rotation” component in the task. As it turns out, rotation tasks also recruit areas of the brain that are normally involved in motor functions (the primary motor cortex and the premotor cortex) (see, e.g., Richter, Somorjai et al. 2000; Windischberger, Lamm et al. 2003; Wraga, Thompson et al. 2003). In a sense, the table is turned: we now have a situation in which activation of motor areas is found even though no actual manual manipulation is performed! Before jumping to the conclusion that mental rotation is actually achieved by covert motor processing (using motor areas of the brain to simulate real movement), we need to consider the possibility that the motor activation observed could be an epiphenomenon. Just because an area lights up during execution of a given task, there is no guarantee that the area is critical for performing the task (Price, Mummery et al. 1999). In the present case, it is conceivable that motor areas become active because they automatically prepare for action even though no action is required (a motor priming effect). There are two lines of evidence against this possibility. Firstly, if motor areas are temporarily made unusable (by means of transcranial magnetic stimulation), performance during at least some types of rotation tasks deteriorates (Ganis, Keenan et al. 2000). Secondly, if subjects are required to perform a task requiring actual motor behaviour concurrently with a mental rotation task, performance on the rotation task deteriorates (Wohlschlager & Wohlschlager 1998). This would only be expected if the two tasks recruit the same resources; otherwise, there should be no interference. This has perhaps been most elegantly demonstrated by Wexler et al. (1998). They had subjects perform a task requiring them manually to move a joystick either clockwise or counterclockwise to the rotation required in the mental rotation task. As would be predicted if the mental rotation task actually requires processing by motor areas in the brain, mental rotation was faster and more accurate when the direction of the mental and manual joystick rotation were compatible than when they were incompatible. Moreover, the speed of the mental rotation correlated with the speed of the manual joystick rotation (when they were compatible) for the individual subjects even though the speed varied considerably between subjects. Importantly, this latter finding is not simply due to a general (underlying) visuospatial processing speed that would be manifest in all tasks. If this were the case, mental rotation times would also correlate with the time taken to perform other tasks requiring visual-motor processing, and they do not (Pellizzer & Georgopoulos 1993).

From the findings considered here, it would appear that the motor system is not only a simple output device for cognition but may actually drive some cognitive operations in the first place. Interestingly, this state of affairs may not be limited to transformations based on rotation. It may also apply to transformations of size. Consider the two pairs of figures presented in Fig. 2.

Figure 2

Imagine that you have to decide whether the figures are identical to each other, apart from differences in size (Fig 2a), or whether they are mirror images of each other (Fig 2b). How do you do this? Probably in much the same way as when you have to rotate mentally the figures presented in Figure 1. You encode one of the figures as a mental representation, but this time you make a “virtual” transformation of size, rather than a virtual transformation of orientation to test whether it could overlay the other. As is the case for mental transformations of orientation (rotation), mental transformations of size also yield reaction times that increase as a linear function of the difference in size: the greater the difference, the longer it takes to figure out whether they are identical or not (Bundesen & Larsen 1975). What happens in the brain when people perform mental size transformations? Well, quite a lot in fact. Here we will concentrate on one particular finding: area V5 gets activated (Larsen, Bundesen et al. 2000). This area is located posteriorly in the brain in the visual cortex and is believed to be specialized in analysing visual movement (objects that move in space) (Zeki 1993). Why should this area be important for solving a task in which there is no visual movement in the stimuli (e.g., in a scenario resembling mental rotation in which the motor cortex gets activated even though no movement is performed)? A possible explanation has been given by Bundesen et al. (1983). They presented people with stimuli similar to those presented in Figure 2 but one at a time in succession (rather than simultaneously). By adjusting the interval between the presentations of two stimuli, they were able to induce an impression of (apparent) visual movement; in other words, to the subjects, it appeared as if there was only one figure moving back and forth (depending on whether the first stimulus was bigger or smaller than the one presented next). For this illusion to work, the interval between the stimuli presented had to be of certain duration. If it was too short, the subjects would just see flickering. If it was too long, the subjects would get the impression of one figure being replaced by another. Critically, the optimal interval between the presentations of the stimuli needed to induce an impression of visual movement depended on the difference in size between the stimuli. The greater the size difference, the longer the interval had to be and, as the reader might have guessed by now, the critical inter-stimulus interval increased as a linear function of the size ratio. The similarity between induced visual apparent movement and size transformation of simultaneously presented figures suggests that subjects resolve differences in size between simultaneously presented figures as if they were differences in depth. That is, subjects seem to imagine (simulate) what would happen with one figure if it came closer to them. This is probably the explanation of why the mental size transformations recruit area V5 in the brain, which is specialized in processing visual motion, even when no movement is present. Subjects “simply” use the machinery in area V5, developed for the processing of motion, to transform the size of objects mentally. This is very similar to what people do when they use their motor cortex to simulate rotation¹⁴. Hence, we conclude that the sensory system is not only a simple input device for cognition but may actually drive some cognitive operations in the first place: cognition is embodied.