Читать книгу Learning Bodies - Группа авторов - Страница 23

Let’s play the devil’s advocate and make the following argument against the empirical examples of embodied cognition given above:

It may be that mental transformations are based, in part, on the brain’s input (sensory) and output (motor) systems. But these operations are not really cognitive in the true sense of the word. Cognition only takes place when knowledge is either accessed during sensory processing or forms the basis of motor output. Neither requirement is really fulfilled in the examples of mental transformations because these operations apply to material that has no meaning at all. Mental transformations are simply too close to the processing carried out by the brain’s sensory and motor systems to really count as cognitive operations. Moreover, mental transformations are only demonstrated in very artificial situations with artificial stimuli that do not in any way resemble what normal human beings would naturally be exposed to.

While we do not agree with such a narrow definition of what would count as a cognitive operation, let us for the sake of argument accept it and see if we can come up with more “convincing” evidence. Given that the narrow definition places “knowledge” as the core of cognition, we will seek evidence suggesting that even knowledge organization in the brain (termed semantics) is to some extent determined by motor functions.

In 1984, Warrington and Shallice presented four patients who exhibited a curious breakdown in their semantic knowledge. While these patients were generally able to identify and understand pictures and words referring to objects from the realm of artefacts (objects made by human beings, such as furniture, buildings, tools, etc.), they were markedly impaired when they had to identify pictures and understand words that belonged to the category of natural objects (such as animals, plants, etc.). To illustrate the specificity of the impairment, consider some of the definitions produced by one of the patients for the following pictures:

Umbrella: A tube used to provide protection against the rain.

Compass: Device to tell you the direction you are heading.

Shovel: Metal made tool for digging holes in the earth.

Giraffe: Bird, not sure what it is used for.

Frog: Bird with no arms.

Rhinoceros: Animal, can be used for eating.

Poppy: Plant/tree of some sort.

From these examples, it is clear why the deficits were termed “category-specific semantic impairments” by Warrington and Shallice. On their own, deficits of this type do not necessarily tell us much about the organization of semantic memory. They could arise if natural objects for some unknown reason just happened to be more difficult to identify in general. While this may in fact be true (see, e.g., Gerlach, Law et al. 2004), it does not amount to an exhaustive explanation, since the reverse impairment is also on record (for reviews, see Humphreys & Forde 2001; Mahon & Caramazza 2003). These patients have marked difficulties in recognizing pictures and understanding words referring to artefacts concurrently with relatively preserved knowledge of natural objects. This double-dissociation strongly suggests that knowledge may be categorically organized in the brain or, at the very least, that different categories are not processed in the same way. To understand what this has to do with embodiment, we need to consider how category-specific impairments for artefacts might arise. A clue to this can be found when we look at the site of the lesion(s) that are associated with this type of deficit.

One of the most agreed-upon models of visual perception divides visual processing into two general streams (Milner & Goodale 1995). The ventral stream, which originates in the primary visual cortex, involves the ventral parts of the occipital and temporal cortex. This stream is primarily responsible for object recognition, that is, the matching of the visual impression with representations stored in visual long-term memory. The dorsal stream, which also originates in the primary visual cortex, involves the dorsal parts of the occipital cortex and the parietal lobes (see Fig. 3). This stream is responsible for visual action, that is, the operations necessary to locate objects in space in order to manually interact with objects. Perhaps, needless to say, this stream is intimately related to/forms a part of the system-subserving, goal-directed motor action.

Figure 3

In general, patients who suffer from lesions in areas belonging to the ventral stream may experience difficulties in recognizing objects they may nevertheless be able to locate in space and interact with (to some extent); this disorder is called visual agnosia. For patients with lesions in the areas belonging to the dorsal stream, the opposite is true. They may experience difficulties in handling objects which they may very well recognize (a disorder termed apraxia) or fail to locate the position of objects in space (Balint’s syndrome). In normal subjects, these two streams clearly interact. Goal-directed action is much easier when you know what you are reaching for. Likewise, it may also be easier to recognize objects when you can tell how they might be handled – a central proposition we will return to below. While it is not entirely clear how or where this cross-talk between streams occurs, it is likely that the two streams converge in the frontal lobes. A region where the two streams might blend is the premotor cortex, situated just in front of the primary motor areas. As the name implies, this region is involved in motor operations and, more specifically, in the planning and comprehension of complex goal-directed actions (Rizzolatti & Arbib 1998; Rizzolatti & Fadiga 1998; Binkofski, Buccino et al. 1999). However, cells in this area discharge not only when objects are grasped, but also when objects are simply observed (Murata, Fadiga et al. 1997). Hence, cells in this region are driven by both vision and action and especially in combination.

All this talk of dorsal and ventral visual processing streams and their likely functions would, of course, be irrelevant to the present context if category-specific disorders did not to some extent map onto these streams, which they do. By and large, category-specific disorders related to natural objects seem to follow lesions to areas associated with the ventral pathway (the occipital-temporal region); whereas category-specific disorders related to artefacts seem to follow lesions to areas associated with the dorsal pathway (the frontoparietal region) (Gainotti 2002). On the one hand, this suggests that perceptual properties may be more important for the comprehension of natural objects compared to artefacts or, alternatively, that natural objects may be harder to differentiate perceptually than artefacts. On the other hand, it also appears that the comprehension of artefacts to some extent relies more on access to motor knowledge (knowledge of how things can and should be handled) than the comprehension of natural objects. This latter suggestion clearly has a bearing on embodiment as it suggests a link between bodily-rooted knowledge of object utilization and the (cognitive) act of comprehension. Before accepting that the existence of category-specific disorders for artefacts argues in favour of cognition being embodied, we need to address a few issues concerning the effect of category.

Firstly, we would like to know whether category-specific disorders for artefacts reveal anything about the cognitive architecture in normal subjects or whether they are simply a curiosity created by brain damage? The answer to this question is relatively clear: category effects can also be found in normal subjects. One of the most consistent findings from functional imaging (brain scans) of category specificity in normal subjects is that the processing of artefacts causes significantly greater activity in the premotor cortex than the processing of natural objects (Martin, Wiggs et al. 1996; Grafton, Fadiga et al. 1997; Grabowski, Damasio et al. 1998; Chao & Martin 2000; Gerlach, Law et al. 2000; Gerlach, Law et al. 2002a). Moreover, this activation seems to reflect that the area is somehow involved in the act of comprehension (a semantic aspect) rather than just a motor priming effect arising because the motor system is preparing to act. If the premotor activation reflects a motor priming effect, we would expect this area always to light up when subjects are shown pictures of artefacts. This appears not to be the case as it does not occur during tasks in which subjects process the shapes of artefacts but only in tasks that necessitate access to semantics (e.g., picture naming and categorization) (Gerlach, Law et al. 2000; Gerlach, Law et al. 2002a). Having said this, it must be acknowledged that it is not entirely clear which type of semantic operation is handled by the premotor cortex (Gerlach, Law et al. 2002b). What does seem clear is that the premotor activation is not specific for artefacts per se, since the area also gets activated when people semantically process vegetables and fruit (Gerlach, Law et al. 2002a). Accordingly, the premotor area is probably recruited when objects are manipulable, and this just happens to be more frequent for artefacts than for natural objects.

Even though it remains to be established exactly what role the premotor area plays in the processing of manipulable objects (Gerlach, Law et al. 2002a), evidence from both brain-damaged patients and functional imaging with normal subjects suggests that action knowledge, which is mediated in part by the premotor cortex, does contribute to comprehension. This is a clear case in favour of embodied cognition as it suggests relatively direct links between bodily-based knowledge and high-level cognition (semantics). Thus, even faced with a very narrow definition of what counts as cognition, we are still forced to conclude that at least some cognitive operations are embodied. That this appears to be the case even for the act of categorization is especially interesting, given that the received view in linguistics for more than two centuries has been that the categories by which we describe the world are transcendental, that is, beyond the vulgarities of the specific organism doing the categorizing (Lakoff 1987). This is clearly not the case if they are occasionally embodied.