BEING ABOUT Chapter 8. Representing and the IPL
Until recently our knowledge of the human inferior parietal has depended on clinical studies, which are notoriously ambiguous. Because functional localization and extent and location of damage vary from one patient to another, results cannot be correlated with confidence. More recently, PET and fMRI imaging have provided better information on IPL activity during various kinds of tasks. But functional imaging has limitations too. Imaging studies must be confined to the kinds of task people can perform while they have their heads immobilized in a scanner. These tasks are often pencil and paper or mouse and monitor tasks -- that is, representational rather than real world tasks. The distinction is often ignored: 'object recognition' tasks are for instance often picture-of-object tasks, and 'navigation' tasks are always navigation-simulation tasks. Since the human IPL is particularly important to representational capabilities, conflating representational and real world functions is particularly damaging to studies of the IPL. When results for a PET study using virtual reality video are called spatial navigation results (Aguirre 1996) we are left without a means to study the contribution specific to representational involvement. There have however also been results, both lesion and scan based, specifically concerned with representational activities. In studies of these kinds, the IPL has been found important to every kind of notational activity -- to textual reading and writing, to musical sight reading, to mathematics, and to the use of pictures. Each of these sorts of representation evokes characteristic combinations of cortical activity. None evokes only IPL activity, but the IPL is important to all of them. The IPL predates these representational forms, and its more basic sorts of spatial function may be a precondition to their development. This chapter considers the evolution and development of the inferior parietal area and its role, first in basic spatial aboutness, and then in the representational tasks it has more recently assumed. Related to the question of IPL function is the question of hemispheric specialization in humans. Areas that are the evolutionary precursors of the IPL are symmetrically contralateral in function, as is the human IPL itself in the early years of its development in an individual. In adult humans both cerebral hemispheres are specialized. Asymmetrical functional lateralization is particularly notable in the IPL and particularly notable in relation to representational skills such as language and mathematics. Basic spatial skills mediated by the IPL and the asymmetrical lateralization of functions particular to this area both seem to be keys to the role of the IPL in representing. Newest association cortexThree large patches of association cortex are unique to human cortex: they are found in the frontal lobe, at the temporal pole, and in the lower portion of the parietal. The area at the temporal pole is important to object recognition, naming and memory, as described in Chapters 4 and 6. The area in the frontal lobe, described in Chapter 5, is important to deliberate maintaining, shifting and sequencing of attention, whether in presence or in simulation. These areas are also important to representation tasks, but in this chapter I will focus on the IPL. The human IPL is wedged into an area below dorsal action streams and above object recognition circuits in temporal cortex. Like the SPL, it connects to premotor cortex in the frontal lobe. It spans a large fiber bundle connecting two areas essential to speech comprehension and articulation, Wernicke's, near primary auditory cortex in the temporal lobe, and Broca's, near premotor mouth areas in the forebrain. Like the SPL it shows multimodal sensory and motor response; it may also be involved in progressive act coordination. Its ventral position brings it into contact with magnificatory sensory cortex of several kinds. It intercepts the forward projection of the ventral visual stream, with its many sorts of parvo resolution. Its somatosensory contact is with primary hand, mouth and throat areas, all of which have massive projections to corresponding areas in the motor strip on the far side of the central sulcus. That it is also in contact with high resolution auditory cortex is suggested by the fact that the IPL of the language hemisphere is active in speech comprehension and production at the level of the phoneme and the word. There are two inferior parietal gyri, the supramarginal gyrus (Brodmann's 40) and the angular gyrus (Brodmann's 39). The cellular architecture of the two gyri is similar. Both have a laminar structure suggesting phylogenetic recency. Both show late myelinization, that of the supramarginal gyrus beginning at the end of the first month prenatally and that of the angular gyrus continuing until the eighth month (Critchley 1953). Like homotypical association cortex everywhere in the brain, the structure of these areas is customizable in experience, and shows unusual morphological and functional variability across individuals. Precursor structures and functionsThe history of IPL precursor function is largely the history of eye- and mouth-related manual skill. Leroi-Gourhan begins the story with the first vertebrate fish, which is also the first bilaterally symmetrical animal. Like all subsequent vertebrates the fish has a front and a back end, the latter specialized for locomotion and the former what Leroi-Gourhan calls an "anterior field of responsiveness", specialized for prehension, ingestion and perception (1964/1993, 31-36, 414). Primate alterations of the anterior field are preceded by a branching within the mammals, Levi-Gourhan says. Among the walkers, like horses, pigs, and goats, the anterior field includes only the face, the forelimb being barely individuated in somatosensory cortex -- most individuated in the goat, where a small area on the inner surface of the front leg is somewhat sensitive. In the other division of the mammals, the graspers, forelimbs become part of anterior system, along with mouth, whiskers and other facial structures. There is a further division in the graspers between animals that use forelimbs only with the face, like rodents, and those whose somatosensory cortex segregates forelimbs and face so they can be used separately as well as together. In some graspers, forepaws may be restricted to symmetrical motions; in others, squirrels for instance, paw motion may be complementary to a degree that allows them to turn something they are grasping. Clawed tree-climbing animals like raccoons or bears are able to grasp with one paw and lift something to the nose for recognition before they carry it to the mouth. In graspers of these kinds there is the rapid development of the parietal noted by Critchley (1953). All primates have hands rather than paws. Arboreal prosimians, 40 million years ago, already had opposable thumbs, along with front-facing eyes and bichromate color vision (Corballis 1991, 60-61). Among primates the burst of interconnected parietal-premotor development serves to integrate these important innovations with other motor and perceptual skills, as described in Chapter 4. Large-brained primates show intensive development of somatosensory and motor cortex -- fifty percent of which is mouth-, throat- and finger-related. In diurnal non-human primates, brain size has been found to correlate reliably with three things: fruit eating, social group size, and the ratio of parvocellular to magnocellular visual cells in LGN (Barton 1998, 1934-35). The three are interrelated, and all are relevant to the refinement of hand-eye skill. With their developing parvo vision, large-brained monkeys are better able to see each other's faces and gestures. With their high-resolution binocular stereopsis, they are able to shape the hand more accurately so that precision grip becomes practicable; precision grip is found among Old World monkeys but not the smaller-brained New World monkeys, who have power grip, in which the fingers act together, but not individual control of fingers (Corballis 1991, 60-61). Mutual grooming and berry-picking both require precision grip. Tool use comes next. Chimpanzees, particularly the females, will modify blades of grass, segments of vine or branches to use in ant-dipping and termite fishing. In skilled work such as ant-dipping the hands have different roles: one hand holds the wand steady and upright in a power-grip while the other hand sweeps the length of the wand in a loose precision grip, catching up the ants in a jumble. When one hand is needed for suspension above the ant's nest, then a foot may substitute for the power-gripping hand, so that the process involves three limbs working complementarily. McGrew 1992, 117 Chimpanzees will also use tools cooperatively. Whoever has control of the nut cracker, an anvil or hammer stone, will accept nuts from another chimp and share some of them after they are cracked. One chimpanzee will also use a stick to clean another chimp's teeth. These convergences of sociality, tool use, and hand-eye coordination bode well for representing practices to come. At this stage tool use is not systematically lateralized. Either hand may be used for precision grip; although individuals have hand preferences, there is no population-level preference for the left or right hand (McGrew 1992, 69).
Leroi-Gourhan reminds us that apes are four-handed animals; the human foot, adapted to walking long distances, is very different from the primate hand, both in its form and in its capabilities. The lower limbs of humans nonetheless retain the relative cortical unimportance of the rear or locomotory limbs of the other primates. In humans 80% of motor and somatosensory cortex serves the head and upper limbs, and, as in chimpanzees, 50% the mouth, throat and fingers. Going bipedal allows many other developments to be built off human front end sensitivities -- carrying, one-handed throwing, and the use of new sorts of tools, to start with (Leroi-Gourhan 1964/1993). Being able to carry allows the invention of food containers, which allows an expansion of diet to include cereals, which in turn allows agriculture and settlement. Somewhere in the relatively rapid evolution of the human branch of the primates there comes the expansion of high order association areas that gives the human brain its present appearance. New gyri in the lower parietal and at the temporal pole are often attributed to language development: The inferior displacement of ventral stream areas and the superior displacement of dorsal stream areas both may be related to the emergence of phylogenetically newer areas specialized for language in the posterior perisylvian cortex. Similarly, the posterior displacement of ventral stream areas away from the temporal pole may be related to the role the latter cortex plays in semantic and lexical knowledge about objects. Ungerleider et al 1998, 883 This sort of description suggests that the IPL area was linguistic from the first. The paleological evidence is, however, that homotypical high-order association cortex, including the IPL and middle prefrontal gyrus, preceded language by several million years. Squared frontal lobes and the rear shift of the macaque lunate sulcus which signals the presence of the IPL are found in H.habilis, 1.8-2.5 million years ago, whereas evidence having to do with the development of the larynx suggests speech is more recent (Wilkins and Wakefield 1995). Because of its current representation-related sensitivities, Milner and Goodale describe the human IPL as outside the ventral and dorsal streams as defined in the monkey (1995). It is probably more accurate to say that the IPL is between them. Similarly, human area 46 in the middle frontal gyrus can be thought of as having expanded between areas that are the frontal termini of dorsal and ventral streams in the macaque. This betweenness suggests that both the IPL and its forward connections in premotor and prefrontal cortex probably are bridges between dorsal touch-vision-action substreams and ventral object knowledge networks. Much of the connective structure of the present-day IPL is probably culturally installed: ways of using the existing resources of these bridge areas, once discovered, are culturally transmitted, so that any human brain with access to cultural record can be modified to current usage without further genetic alteration.
The syndrome which follows removal of the parietal lobes in the monkey is contralateral and symmetrically equivalent on the two sides. Mountcastle 1975, 873 Before considering basic function in the IPL, I will need to say something about hemispheric lateralization in human brains. Control of high-cultural practices such as language and tool use is profoundly asymmetrical. The asymmetry is not in the structure of the hand, mouth or throat but in their use, as mediated by one or the other cortical hemisphere. When used for many kinds of skilled deliberate action motion of these structures is organized from the left hemisphere; when used in exploratory, environmentally sensitive ways, it is controlled from the right hemisphere. This very general description is truest when we are talking about linguistic motion. In adults the hemispheric lateralization of language can be so complete that loss of left hemisphere language areas results in almost total functional loss. Small genetic differences and the plasticity of high order association cortex result in many anomalies of functional lateralization in individuals. I will discuss some of these anomalies later. The standard practice is to ignore them and call the hemisphere not specialized for language the right hemisphere. I will adopt that convenience in what follows. There are many ways to organize sensory-motor through-function in an animal body. These include ipsilateral control, in which muscles are controlled through cortex on the same side of the body, contralateral control, in which muscles are controlled through the hemisphere on the opposite side of the body, and bilateral control, in which muscles on both sides of the body are controlled from one hemisphere or the other. There may also be control by means of a recurrent bihemisphere network. Human brains probably use all these possibilities, and they add one more. Asymmetrical lateralization, in which different cortical hemispheres control different kinds of function, is rarely found in non-human brains. In the human SPL, nonfocal perception and behavior tend to be bilaterally cross-integrated below the level of the cortex, or may use the bilateral fields of the magno-dominated dorsal stream (see Chapter 4). Vision of motion particularly in the far periphery, proprioception of arm and leg position, perception of whole body displacement, and tactile perception of large surfaces, all are not lateralized. Limb and whole body motion are integrated in areas that may also respond to the visual periphery; movement of either limb, to either side of the midline, may be controlled from either hemisphere. In contrast, the kind of vision we bring to bear on objects we handle and talk to -- central near vision -- is highly magnified and crisply contralateral. It is not eye response that is contralateral, however; front-facing eyes have overlapping fields, so there would be no advantage in having eye-specific cortical segregation. Instead, a very large area of the right occipital responds to the left side of the visual field, while a large area of the left occipital responds to the right side of the visual field. [8-4 Visual field contralaterality] Precision hand movements and the manipulation of small objects is like central vision in being organized from magnificatory sensory and motor cortex. Magnificatory hand areas are also segregated into opposite hemispheres. When the hemisphere that responds to the left side of the visual field also controls the left hand, visual contralaterality ensures the most direct possible connections between visual and motor response to events on the left side of the body. Networks organizing the contralateral hand in the contralateral visual hemifield can include action for purposes of perception, tactile perception, and self-monitoring during hand motion. In pre-primate mammals the contralateralized areas of the two hemispheres, still symmetrical, are equipotential. Among the primates, full asymmetry occurs only in humans. Like the development of the IPL itself, the development of functional lateralization is thought to predate speech. The evolution of asymmetrical lateralization occurs at around the same time as the evolution of IPL structures; as with developed IPL gyri, structural asymmetries are already present in H.habilis (Wilkins and Wakefield 1995). Basic function of the IPL in humansIf IPL structure predates language, what was its basic or original function? And what is it about IPL structure and function that allows it to be rebuilt for language and the many other kinds of representing practice? As we saw in Chapter 4, the human superior parietal seems to perform functions performed by the monkey inferior parietal -- is any part of the human IPL nonetheless a functional homologue of the monkey IPL? Are the representation-related possibilities of the human IPL based on abilities present in non-human primates? We do not have access to human brains that have not been culturally modified. The best ways to approximate a guess at basic IPL function seem to be to look at non-representation-related function in the right hemisphere, to look at the cortical topography of the area, and to consider the function of the IPL's precursor structures in the monkey. These approaches do not support certainty, but they offer suggestions.
The evidence we have says that in humans both parietal lobules, the SPL and the IPL, "play a basic role in the organization of complex simultaneous (spatial) syntheses" (Luria 1973, 147). In earlier chapters I described wide networks active in maintaining and shifting attentional axes. These networks have foci in subcortical areas, in sensory areas, in medial saliency areas, in frontal cortex, and in the SPL. An IPL focus is often also found. We do not know exactly what the difference is between the roles of the IPL and of the SPL, in these instances. Ungerleider believes, based on lesion studies (Ungerleider et al 1991, 1625), that the SPL is more involved than the IPL in the ability to disengage and shift the spatial focus of visual attention; but IPL damage nonetheless results in characteristic kinds of spatial difficulty. Different kinds of spatial defect can be distinguished. What they have in common seems to be, very generally speaking, a difficulty establishing, maintaining, shifting or recovering an attentional axis, or maintaining more than one axis at a time. There are many subvarieties of all of these deficits, and they occur in different combinations and to different degrees. It is not clear exactly how they are related. When there is bilateral damage to inferior parietal areas, patients are able to see and identify single objects if they are given enough time to fixate them, but they are not able to see more than one object at a time. They cannot fixate one object in an array. They are unable to say that one object is nearer or above or to the right of another. They cannot track moving objects. When they are walking they have difficulty seeing. They often cannot find their way either inside their home or in the street (Levine, Warach and Farah 1984). This collection of symptoms, or some subset, is called dorsal simultanagnosia. Other difficulties may include visual mislocalization, in which patients can see an object but cannot tell where it is, and left neglect syndromes in which patients do not notice objects in their left visual hemifields (Kosslyn et al 1997). Many of these difficulties also occur with unilateral damage, if the hemisphere damaged is the right hemisphere. Left parietal lesions normally do not result in spatial difficulties of these kinds -- which is evidence that the right parietal is more bilateral than the left. People with the sorts of visuospatial difficulties mentioned above also have trouble with visuospatial imagining and memory. An involvement of the right IPL in simulation has been notable in PET/MRI studies. Milner finds the right supramarginal gyrus active with short and long term location memory tasks (1996, 17). Learning and remembering a reach target position activates a right hemisphere network that includes the IPL along with superior parietal, premotor and motor areas. Left neglect for actual locations is paralleled by left neglect for imagined locations. Left neglect also occurs for covert attention (Corbetta 1993). The simultanagnosic's inability to find his or her way around a room or a neighborhood is paralleled by an inability to imagine the layout of the room or the street (Mellet 1996).
The clinical evidence of IPL involvement in spatial function is very miscellaneous, and there is as yet no established theory to make sense of it, but we may be able to take another sort of clue from the cortical topography and evolutionary antecedents of the IPL. Many of the skills mediated by the IPL are not found in monkeys; nonetheless the most important clue to the basic function of the IPL may be the approximate coincidence of IPL-frontal connections with at least the more anterior portion of the monkey grasp stream, which is more ventral and more connected to central vision than the dorsal reach stream. Recall that anterior area AIP in the macaque intraparietal sulcus has been found to be predominantly grasp related, while VIP and MIP further back in the intraparietal sulcus have been found to be reach related. Does the human intraparietal sulcus have some of the same subfunctions? We cannot do single-cell studies on human cortex, and lesion and functional imaging evidence is imprecise about localization, so we do not know. Functional imaging and lesion evidence suggest that there is some parallel between what happens in the human supramarginal gyrus and what happens in AIP in the anterior end of the monkey ips, and between what happens in the human angular gyrus and what happens in VIP and MIP in the posterior end of the monkey ips. Although the two gyri of the IPL are architectonically similar, clinical evidence shows them to be functionally somewhat different. In this section I will mention only non-representational functions of these areas.
Cells in the macaque grasp stream include hand-related act constants, as well as cells responding to the orientation and volume geometry of graspable objects. The supramarginal gyrus, Brodmann's area 40, lies under the anterior end of the human intraparietal sulcus and, like areas AIP, 7b and F5 in the macaque grasp stream, it seems to be involved in grasp-related spatial judgments made on the basis of touch or stereoscopic central vision. The supramarginal has been found to be active during stereoscopic depth perception in near space: PET studies have found IPL response to three-dimensional orientation of objects (Faillenot, Decety and Jeannerod 1999). The anterior border of the supramarginal abuts expanded hand and face areas of somatosensory cortex. Like other parietal association cortex it is probably integrating visual and tactile perception for purposes of object manipulation with mouth and hand. Other studies have found IPL involvement in judgments of the three-dimensional form, length, size, and distance as well as the orientation of graspable objects. PET studies of object prehension in humans have shown foci mostly in left hemisphere SPL but when grasp is more consciously performed -- when the SPL is damaged and people are relearning visuomotor grasp, or when they are learning complex hand behaviors -- the IPL is also active (Rizzolatti et al 1996b, 250). A PET study by Decety and others (1994) found foci of activity in the supramarginal gyrus (and in premotor and prefrontal cortex) also when subjects imagined grasping an object.
A second basic function of the IPL seems to be discerning, imagining and remembering configurations of objects, or the location of individual elements within such a configuration. The angular gyrus, Brodmann's area 39, lies under the posterior end of the human ips. It abuts dorsal area 19 -- which is secondary visual cortex involved in high level form perception -- posteriorly. At its lower edge it borders areas in the temporal lobe important to speech and audition. Its rear border is close to motion vision areas active when we track moving objects or anticipate a trajectory. Like areas LIP, VIP, 7a and F4 in the macaque reach and saccade streams, it seems to be necessary to judgments of object position. Lesions of the angular gyrus often result in difficulty with positional relation -- right-left judgments, for instance -- especially when there are several objects in different positions. We have seen that in monkeys reach and grasp subsystems are segregated because they use different muscles, but also because they use the senses in different ways, reach using peripheral vision and shoulder and arm somatosense, and grasp using central vision and hand somatosense. In humans reach and saccade areas in the SPL also respond to object position on the basis of somatic sense and/or peripheral vision, but bilateral lesions of the IPL are found to disturb reach when it uses central rather than peripheral vision (Sakata and Taira 1994, 848), perhaps because they disturb aim toward one object in an array of objects. The two aspects of spatial function that I have described -- a volumetric sense of object form required for handshaping, and a sense of object configuration required for reach guided by central vision -- seem to be compatible with the little that is known of their continuing environmental and representational functioning. Neither spatial function can be assigned exclusively either to the supramarginal or to the angular gyrus, but the former is more often named in relation to visual and tactile three-dimensionality, and the latter in relation to horizontal left-right position. Presence hemisphereTo understand the ultimately very different tasks of right and left inferior parietal areas in humans, it is necessary first to know something about functional differences between human right and left hemispheres as wholes. In the culturally modified brain the right hemisphere seems to keep more basic or environmental functions, while the left hemisphere is rebuilt for language and language-like cultural tasks. This description oversimplifies: the left hemisphere retains some basic environmental functions, at least some of the time. At the same time, since it must take over environmental monitoring that would have been performed contralaterally by both hemispheres before specialization, the right hemisphere must at least sometimes or to some degree perform environmental functions for both sides of the body. This in itself changes the structure and function of the hemisphere. And the right hemisphere also has representation-related specializations that are complementary to those of the left hemisphere. Right cortex and left cortices are normally differently shaped. Besides being wider across the frontal lobe, the right hemisphere has a Sylvian fissure that rises more steeply, so there is less tissue in the IPL, above the fissure, and more in occipital and temporal areas below it. Right hemisphere connections are more diffusely organized (Solms 1997, 162-3). Within the hemisphere, axons are longer, so they connect more distant neurons. Probably as a consequence, the right hemisphere is slower. There are also more connections to subcortical structures such as the thalamus and basal ganglia; these connections make the right hemisphere more important to whole-body functioning, with conjoined visceral and musculoskeletal response (Damasio 1999, 211, 354). PET studies confirm that the right hemisphere is less contralateral, more bilateral, than the left. A study by Corbetta and others found two distinct foci active in the right superior parietal during spatial attention tasks, and only one in the left superior parietal (1993, 1224). The evidence is that the single focus on the left directs attention and action within the right visual hemispace, while the two foci on the right are directing attention and action within the whole visual field. Hemispheric dominance for complex hand movements has been found to vary according to act type, and this has been found for both overt behavior and simulation. The right hemisphere was found to organize the use of both hands in both visual hemifields when they are being used for active exploration, while the left hemisphere controls the use of both hands when they are being used in non-exploratory ways, for instance during skilled use of tools (Gitelman et al 1996, 176). Corbetta's finding of two foci active in the right hemisphere for visual attention in both left and right hemifields, and Mesulam's discovery of bilateral right hemisphere control for both motor and sensory aspects of exploratory behavior, suggest why right hemisphere lesions might be particularly damaging to integrated spatial function. Prefrontal areas on the right are also more involved in spatial attention to present physical environment, and in memory for spatial locations. As is discovered from classical right hemisphere syndromes such as face recognition deficits (prosopagnosia) and place deficits (topographic agnosia), the right hemisphere's ventral stream dominates object perception and recognition. Right temporal areas are more important to complex visual pattern perception and memory. Secondary auditory areas on the right are more responsive to environmentally significant sounds (like coughing or laughing) and to tonal qualities. The right hippocampus is more often active in place memory and path integration. The right hemisphere is also more involved in emotional response, and is more important to recognition of emotion in others; this is so particularly for painful emotions like fear and sadness. The right hemisphere is sometimes said to be specialized for 'visuospatial cognition' or 'concrete spatial cognition' as opposed to 'symbolic cognition'. This form of the contrast is quite unclear. Right hemisphere specializations include representational media such as pictures, which are 'symbolic' but also concretely spatial. Moreover, any representational form is spatially located and can only be used if it is perceived by the usual spatial means, that is, by orienting and by axial attention. The right hemisphere might more accurately be called the presence hemisphere, since it seems to be the hemisphere more involved in staying in touch with where we are. Roman Jakobson (1980) calls it the business hemisphere -- as opposed to the syntax hemisphere -- since it seems to respond to aspects of the real world environment that the left hemisphere ignores. Language and the left hemisphereIn monkeys, the left hemisphere has basic contralateral importance for actions of the right hand -- the same sort of contralateral importance the right hemisphere has for actions of the left hand. In humans this contralaterality of hand control occurs for some kinds of action but not others, and it is combined with hand preference for skilled action -- that is, with specialized lateralization. The right hand is more skillful with complex rhythms than the left hand. It is the hand preferred when we use a hammer or a pencil: any sort of tool. The categories of action for which the left hemisphere has bilateral competence are those requiring fast fine motor control. Gitelman reports PET results showing that overlearned or stereotyped movements activate the left hemisphere whether they are performed by the left or right hand, provided that they occur in the right visual hemispace. He also reports similar results with fractionated finger movements of either the right or left hand. For stereotyped representational acts, or skilled use of tools, it is the left hemisphere that controls both hands (1996, 176). PET studies find language and bilateral control of fine manual skill to be lateralized to the same hemisphere in the majority of people. That is, the language hemisphere is usually (but not always) also the hemisphere most involved in skilled manual and oral action with focally seen objects in near space. For speakers of manual languages like American Sign Language the two categories coincide. ASL uses both hands in a communicational subspace at the midline of the body. We know the language hemisphere is organizing the signing of both hands because lesions affect linguistic and nonlinguistic gesture differently. Referential uses of gesture by deaf signers -- nodding, looking, pointing with hand or chin, turning, and pantomime -- doubly-dissociate from syntactic uses of gesture. Concrete or 'topographic' uses of sign are damaged along with other concrete spatial abilities; syntactic uses of sign are damaged along with other left hemisphere linguistic abilities. Signing fluency and grammaticality survives right hemisphere damage but not left hemisphere damage (Poizner, Klima and Bellugi 1987, 211). In oral speakers the linguistic effect of left hemisphere damage is the same. There is almost complete loss of linguistic fluency and grammaticality, spoken and written, with damage of left hemisphere language areas. The fact that skilled use of both hands is organized by the hemisphere that also organizes language begs explanation. There are a number of puzzles. Are skills used in focalized action with small objects somehow relevant to language? Why is the contralaterality of right hand control in monkeys replaced by a bilaterality that depends on the nature of the action being performed? If left hemisphere control for skilled action is bilateral, why is the right hand more skilled? What relation is there between manual skill and the skills of mouth and ear that enable speech? Why is the correlation between hand preference and hemispheric specialization found in the majority of people but not in all?
One sort of attempt to answer these questions looks at the significance of spatial and temporal scale. Consider first the very fine scale of phonological articulation, which requires precise, contextually constrained cyclic patterns at the 200 ms time scale of the vowel-consonant unit. The motor system must prepare to say the next phoneme while it is still in the process of articulating the present one, since speaking a b after a k needs a different motor sequence than speaking a b after an a. Similarly there will be differences in hearing a b after a k or an a. Like reach and grasp, the fine coordination of articulatory pattern is learned over several years. It begins as babble, which Emmorey and Reilly describe as motor sterotypies with a syllable-like structure (1995, 13). Gesture babble in deaf babies begins at the same age as vocal babble in hearing babies. The coincidence of manual precision and linguistic precision in the deaf suggests something about a left hemisphere specialization that could support both. The muscular run of the sentence, the elegance of practiced handwriting, the smooth facility of gestures or facial postures used socially -- these phrased combinations of communicational elements do have something in common with the smooth, cyclic and contextually constrained motions of tool use. Kimura suggests that the left hemisphere is well adapted not for the symbolic function in itself but for the execution of some categories of motor activity that happen to lend themselves readily to communication (Kimura 1993). There is something else language and tool use have in common -- both are high cultural skills. Both are deeply social and both are performed in what could be called a technical state of attention. Both are socially learned, and this learning goes beyond instruction to empathy; the discovery of mirror function in manual and oral grasp areas of premotor cortex suggests that when we watch skilled manual and oral action in another creature, we are covertly rehearsing the action. And both require a technical or instrumental attitude, which generates action and perception in sustained, focused, deliberate ways (Corballis 1991, 197). A technical attitude is not completely uninterested in the surrounding environment; an instrumental or communicational task is itself part of an environment. But technical attention is often focused on very restricted aspects of an environment that must be perceived precisely and at speed. Tool use and language are demanding tasks. Someone attending to either is usually having to ignore much of the situational given. This relative attentional absence from physical surroundings increases when either tool or language is used simulationally. Someone using a tool is often guided by imagining a desired product. Language is often used to evoke simulational structure in perceptual areas which then are not available for environmental perceiving. The technical hearing, seeing, remembering and acting organized by left hemisphere means tend in these ways to be dissociated from basic environmental presence.
The human cortex does not completely dispense with symmetrical contralaterality after infancy, but it rebuilds both hemispheres to accommodate asymmetrical specializations. There are structural asymmetries already present in the fetal brain (Geschwind 1992, 196), and young children do show hand preference, but they have been found to use both hemispheres for language. If a child's left hemisphere is lesioned, the right hemisphere is able to take over linguistic function almost without remaining deficit. This suggests that the specialized left hemisphere structures serving language are custom built during linguistic development. If an adult's right hemisphere is damaged, the left hemisphere is not able to take over the sorts of spatial function that had been performed by the right hemisphere; this suggests that once it has been rebuilt for language, the language hemisphere cannot revert to earlier structure. Deacon believes left lateralization for language to be a "dynamical functional result of processing demands imposed by language performance during development". One of these demands is for fast mapping of phoneme to morpheme; at 15 months it takes the baby a second to recognize a spoken word, at 24 months, 600ms. An adult, Deacon says, can get elephant from eleph in 400ms (Deacon 1997, 292). The development of phonemic fast mapping happens at the same time as the switch from bilateral to left hemisphere control for grammar; response to closed class forms (prepositions, conjunctions, and articles, etc.) begins to shift into the left side by the end of the third year (Neville, Mills and Lawson 1992). Bellugi brings confirming evidence from studies of language acquisition in deaf children, who start by using pointing gestures the way hearing children do. In the year between one and two they begin to switch to syntactic uses of point, and by two and a half they have made the transition from prelinguistic gesture to a fully linguistic system (Poizner, Klima and Bellugi 1987, 22). This shift is accompanied by a shift from bilateral to left hemisphere control. There are several ways of understanding why it is the left rather than the right hemisphere that is rebuilt for language. In those who are left lateralized for language the left hemisphere planum temporale, a secondary auditory structure particularly sensitive to speech sound, is larger than the planum on the right. Marion Annett (Annett and Manning 1990) believes there is a gene, one allele of a gene pair, that delays development of the right hemisphere planum during the sensitive period for language acquisition. According to this theory the left hemisphere would take on language functions because it is in a growth spurt at just the time when children are learning to talk. Alternatively, it is possible that the right hemisphere, which begins to develop earlier, has already been preempted for more basic and more bilateral perception-action skills that have been developing since birth. The left hemisphere may have become the tool and language hemisphere because, for purposes of bilateral motion integration, it is to some extent a spare. Some of the environmental functions retained by the right hemisphere are themselves forms of linguistic function. Deacon thinks a wide range of competing language-related functions might have displaced each other into separate hemispheres because they use similar cortical structure in different ways. Phonological and pragmatic aspects of speech audition, for instance, use similar areas of temporal cortex and must operate simultaneously. [8-7 Deacon's map of R-L displaced functions] Left lateralization of skilled right hand action may follow as well as preceding left lateralization for language. It makes sense to control communicational behavior and manual skills learned and practiced in communicational contexts from the same hemisphere. Skilled actions are often taught with the use of language. The left hand is at a disadvantage with verbal instruction, because the organization of motor control for the left hand must flow from language areas on the left to the left premotor area and then across the anterior corpus callosum to right premotor and motor areas. The right hand, controlled through more direct connections between language areas and motor areas in the same hemisphere, is in fact faster in linguistically instructed action.
Tools for the hand, language for the face, are twin poles of the same apparatus. Leroi-Gourhan 1964/1993 Speech needs auditory fast mapping of phonemes to morphemes and precisely articulated movements of mouth and throat, and various kinds of technical skill need fast, precise articulation of hand micromovements within phrased sequences of manual action. In culturally trained people both are organized from the left hemisphere because that hemisphere has been modified for perception and motion at that spatial and temporal scale. There are reasons to suspect that profound structural/functional connections between hand and mouth underlie these similarities between speech and manual skill. Grammatical and acquisitional similarities of signed and spoken languages are one kind of evidence. Kimura describes others; she finds that, in hemispheric lateralization tests, speech and gesture during speech are the most reliably correlated of all behaviors. She also finds correlations of language difficulties (aphasias) with difficulties in handling objects (apraxias). There is also the fact that experimental subjects asked to read aloud while tapping with their right hand have difficulty doing so, but left handed tapping does not interfere (1993). Recall that mouth and hand both belong to what Leroi-Gourhan calls the anterior field of responsiveness, and that, in early mammals, when the hand has begun to have its own magnified sensory and motor areas in parietal and prefrontal cortex, the forepaw is the mouth's auxiliary in grasping food. At this stage hand and mouth work together and presumably are coordinated from the same or overlapping cortex. Recall also that in parietal and frontal grasp areas in the macaque there are cells that respond during kinds of prehension (grasping, tearing, manipulating, etc.) whether performed by mouth or hand. That there are remnants of these sorts of overlap or conjunction in humans, and that they are also related to the eye, is suggested by the way, in people performing difficult kinds of hand-eye coordination (a child learning to print, an adult threading a needle), the tongue can be seen shadowing manual action. Mouth, throat, tongue are like the hand in having extensive cortical area in both somatic and motor fields -- as though the mouth were the fovea of the facial somatosensory system (Colby, Duhamel and Goldberg 1996, 49); and both kinds of highly magnificatory primary cortex happen to be bunched together at the ventral end of somatosensory and motor strips. We have seen how, in the macaque, the grasp stream dips through this ventral-most lateral region on its way to premotor and motor cortex in the forebrain. Mouth motion, including motion for purposes of speech, must be organized within integrative gradients at least as complex as those organizing reach and grasp in parietal-prefrontal circuits. Some of these mouth gradients are likely to be overlap parts of grasp gradients. Phonological perception and production require a left hemisphere network that includes posterior temporal and parietal regions as well as a premotor region, as described in Chapter 6. We know more about the end points of this speech gradient than we do about middle sections. Conjunction analysis of PET results shows a network of areas active in both comprehension and production of language (Papathanassiou et al 2000; Mesulam 1990). As we learn more about the network character of linguistic function and the recurrency of cortical nets, it seems likely that, like reach and grasp, linguistic perception and action are organized in sheets of tissue with both sensory and motor response. The posterior part of the planum temporale is larger in the left hemisphere and is thought to respond selectively at the temporal microscales of phonological hearing. Adjacent to the planum, near the temporal-parietal junction, is Wernicke's area, a large region of association tissue now known to be active in speech preparation as well as phonemic perception. An area near Wernicke's in the superior temporal sulcus has also been found to be active in preparing or understanding skilled hand-object interactions (Perrett et al 1990). Broca's area, situated with other sorts of premotor cortex in the left forebrain, lies at the anterior end of the speech stream, adjacent to premotor areas preparing non-speech movements. It is known to be crucial to fluent speech articulation, but it also has recently been found to be active during speech comprehension. Broca's seems to be one of the places where mouth and hand organization may partially overlap. We have seen that in macaques the grasp stream terminates at premotor area F5. Caminiti (1996) describes the macaque F5 as having two functionally distinct areas, one of which might be the precursor of Broca's area. The area in question is a premotor area with visual response also to hand and mouth actions performed by others -- the area with mirror cells. It has long been noticed that hearing speech is helped by seeing speech, and that both seeing and hearing speech seem normally to have subvocal participatory involvement. Especially if it has mirror cells, Broca's area would be one likely place for this sort of covert motor participation. Functional imaging evidence leads Zatorre to propose that subjects making phonetic judgments must access articulatory circuits included in Broca's (Zatorre et al 1992, 848). Broca's has more auditory (and possibly less somatosensory) sensitivity than the macaque F5. That it does have visual and mirror sensitivity is suggested by the fact that people with Broca's lesions are not able to lip-read. If mirror cells in Broca's are responding to observed speech motion the participation of Broca's in language comprehension may include motor microsynchronization of hearer with speaker. (If so, learning to imitate accents may be an inherent part of hearing speakers with accents.) The motor entrainment of hearer with speaker may be necessary to learning to speak as well as to understanding speech; something similar may occur in songbirds, who are unable to sing if they have not heard other birds (Nottebaum et al 1990). The IPL, lateralization and representationSo far in this chapter I have described precursors to IPL function and structure, including basic perception/action functions of the IPL in humans, and then lateralized specialization of cortical hemispheres as wholes. In this section I describe representation-related IPL functions as they are found after lateralization of function in the two cortical hemispheres. The question to be kept in mind is this: what is the relation of representational uses of the IPL to earlier non-representational uses? We saw earlier that the basic pre-representation function of the angular gyrus seems to be something about locating objects relative to other objects, as in right-left judgments. This locational function probably has something to do with object manipulation guided by central vision. The supramarginal gyrus, on the other hand, seems to be necessary to perceptual and simulational judgments of object form for purposes of grasp, these judgments of length, size, distance and orientation being made on the basis of touch or three dimensional central vision. Can all representing practices be understood as built around these basic spatial functions -- even mathematics and language, the hardest cases?
In the majority of people the Sylvian fissure is less steeply angled on the left. There is more tissue in the area of the occipital-temporal-parietal junction above the fissure than on the right, and less in the occipital and temporal areas below it. Areas enlarged on the left include both gyri of the inferior parietal lobule. This enlargement of the left IPL is not solely a result of practice; it is present at birth (Geschwind 1992, 196). The arcuate fasciculus between Wernicke's and Broca's areas underlies the left IPL, and both gyri of the IPL also show speech sensitivities. When left hemisphere SPL and premotor areas are damaged, or when we are learning complex hand motions, the left supramarginal participates in skilled prehension. It seems likely that the left IPL is also involved in organizing the fast, fine-scale, routinized motor cycles required for speech articulation; if this area is lesioned there is difficulty switching from one speech movement to another (Rushworth et al 1997, 1261). The left inferior parietal is like the left SPL in being predominantly contralateral rather than bilateral. Where PET studies find the supramarginal on the right responding to visual events in either the right or left hemifield, the supramarginal on the left is active only with visual activity in the contralateral right hemifield (Gitelmann 1996, 174). This contralaterality is particularly apparent in all sorts of notational activity. If we are right-handed the left hand is almost entirely uninvolved. When we write from left to right, motion is directed into the right visual hemifield. Since the right central hemifield is perceived with expanded occipital areas of the left hemisphere, and right hand motion is being also being organized from the left hemisphere, sensory-motor aspects of writing and calculating can occur quickly, integrated by an almost exclusively intrahemisphere network. Lesions of the left IPL can result in Gerstmann's syndrome, a constellation of deficits including inability to write and inability to calculate. The relation of these difficulties to basic spatial functions of the IPL is suggested by the fact that Gerstmannn's usually also involves difficulty distinguishing fingers and difficulty with horizontal right-left attention in general (Gold et al 1995). As we have seen, left hemisphere representational specialization may be a specialization of scale. The various forms of reading and writing are like speech in requiring speed and tight, hierarchically nested cycles; and the smallness of notational forms like written letters, musical notes and mathematical symbols requires the resources of cortical magnification. The left IPL, which has access to central vision and to magnificatory areas of somatosensory and motor cortex, may be structurally retrofitted to just this spatial and temporal microscale. Kosslyn describes connectionist models best suited for syntactic rather than spatially coordinative functions as "networks with small, less overlapping receptive fields" (1992, 562). The left hemisphere with its larger IPL area and shorter axons may be more able to organize itself into a congeries of habituated small subnets, more like switches than like integrational fields. Activity is found in the angular gyrus, area 39, when we name something presented visually, or when we read aloud. Activity is also found in the angular gyrus when we name something presented acoustically (Geschwind 1992, 189). In professional musicians, and in people with perfect pitch, the left hemisphere is more active than the right during any music tasks. Areas of the left supramarginal gyrus active in sight reading music are distinct from, but adjacent to, areas active in reading and writing words (Sergent et al 1992, 108). Left hemisphere IPL lesions destroy ability to name numerals (nine for 9), along with the ability to read and write mathematical words and symbols. Although arithmetic calculation strongly activates the homologous area on the right, calculation fails outright only with left hemisphere lesions (Dehaene 2000, 995). Dehaene suggests this failure may result because calculation routines make use of subvocal speech.
When we are engaged in representational tasks, the right hemisphere IPL may use its basic environmental functions to stay in touch with the context of the representational event, and/or it may use them in specifically representational ways. I will first outline basic environmental functions of the right hemisphere IPL, and then describe its specifically representational specializations. Someone with right hemisphere lesions of the IPL has difficulty seeing the left visual hemifield or initiating movement to the left. The right supramarginal, with the right SPL and forebrain, is part of a network that controls exploratory action by both hands in both visual hemifields (Gitelmann 1996, 174). Experimental subjects learning the positions for objects on a screen were found to be using the right angular gyrus along with motor, premotor and SPL areas ( Kawashima et al 1995, 111). The right IPL has been found to have relatively wide-spread connectivity, in contrast with the tighter, smaller, more discrete areal organization of the left IPL. Steven Kosslyn thinks of the right IPL as specialized for holistic judgments, judgments of overall spatial organization of the sorts best modeled by connectionist networks with large overlapping receptive fields (Kosslyn 1992, 562). In adults who have had callosal commissurotomies, each hemisphere's competencies can be tested in isolation. Subjects in whom an isolated right hemisphere is being tested are better able to see complex wholes, and are better at understanding three-dimensional facts in terms of two-dimensional representations. They are better at connecting two dots or matching oriented lines. They are better at disembedding hidden figures, finding contours, drawing from life. They are better at copying drawings, especially if they are complex or abstract. They are better at making and understanding diagrams, maps and plans. They are more likely to be able to match faces photographed from different angles or in different light (Geschwind and Galabura 1987). Specifically representational IPL function includes many of these right hemisphere graphic functions. Object agnosia for drawings -- an inability to name or recognize drawn objects -- is a result of right IPL damage (Kandel 1991). Along with locations in the superior parietal and prefrontal, the right IPL is active during working memory tasks with perspective drawings of cubes; along with medial areas it is also active in long-term memory tasks with line drawings. Giftedness in spatial reasoning abilities important in chess, mathematics, physics and music has long been correlated with unusual right IPL development. In keeping with its holistic bias, the right hemisphere seems to be necessary to geometrical or topological intuition. Dehaene (1997, 2000) credits the right IPL with subitization, which is the ability to say how many things there are without having to count them in sequence. The right IPL is also active when we count dots. Dehaene finds significant right lateralization for number comparison tasks, no matter which hand is used to respond. Early phases of a calculation task (as measured by event related potentiation) are sensitive to the form of arithmetic notation and are bilateral; in the second phase there is stronger activity in the right IPL, and this activity goes on longer if the calculation involves large magnitudes. Dehaene therefore thinks of the right hemisphere as performing a sort of spatial visualization of quantity comparison. This guess is in keeping with right hemisphere talent for rapid quantity approximation. The right IPL is also known to be better with geometric visualization. A developmental theory of left lateralization for language and language-like representation is supported by the fact that the right hemisphere has been found to have linguistic and mathematical abilities characteristic of an early developmental stage. An isolated right hemisphere is more able than the left with the earliest names learned, and retains language abilities at the level of the young teenager (Papathanassiou 2000). Dehaene finds both hemispheres competent in basic mathematical skills: visual and tactile object size estimation, perception of small sets, numeral recognition, and quantity comparison using numerals. In both hemispheres these tasks activate the intraparietal sulcus at the top of the angular gyrus. In addition to its specialization for environmental sound and for tonal aspects of speech, an area of the right angular gyrus is found to be specialized for appreciation of melodic line, timbre and harmony in music (Zatorre and Sampson 1991, Zatorre et al 1992, Zatorre et al 1994, Tramo and Barucha 1991). It is also more tuned to the melodic possibilities of language; von Stein et al found the right hemisphere but not the left responding to nonsense names, spoken or written (1999). An isolated right hemisphere is better at understanding and producing other nonsyntactic aspects of language. If the right homologue of Wernicke's area is damaged, there are difficulties producing and understanding pragmatic aspects of language like tone, loudness, timing, and emotional expression (Sheilds 1991). The right hemisphere is also better at appreciating emotional aspects of the communicational situation and of the communication itself, and better at understanding gestures accompanying speech. The left IPL is more involved than the right in notational behavior, but right hemisphere neglect and simultanagnosia can also include reading and writing difficulties, since words must be read in sequence and placed in relation to each other on a page. The participation of the area around the right hemisphere Wernicke's homologue has been found to be important to comprehension and production of linguistic units at the level of the sentence or the text, as opposed to the syllable or the word. This may be related to the right hemisphere's involvement in basic situational aboutness, since comprehension and production of sentences and texts often involve situational simulation. Attention to the spatial layout of the communicational event is also necessary to certain categories of linguistic form, and the right hemisphere IPL is necessary to this sort of language use for both spoken and signed language. (More in Chapter 9.) It is also necessary to linguistic description of spatial layouts (McNeill and Pedelty 1995, Emmorey et al 1995). The environmental involvement and situationally holistic character of right hemisphere response also makes it important to linguistic play -- to metaphor, to humor, and to the sense of simultaneous or alternative meanings (Brownell et al 1990).
Not everyone has language lateralized to the left hemisphere, and not everyone has language and manual skill lateralized to the same hemisphere. In some, language and/or manual skill seem not lateralized at all. Annett and Manning have recently made sense of these variations by proposing a three-part categorization to replace the old right-left dichotomy (1990, 61-62). They propose an allele, one of a gene pair, which would delay the development of the posterior part of the right planum temporale relative to the development of the left planum temporale. Distribution of this gene, the rs gene, would vary within a population. In people without the rs gene, who are more often male, the left planum will not be enlarged relative to the right planum; as a consequence there will be developmental delays for language skills, unusual aptitudes for right hemisphere specializations, and an unusually strong left hand. People with two rs genes would have a relatively underdeveloped right planum; they would have a weak left hand, would be strongly left dominant for language and developmentally advanced in language skills, and would show general weakness in right hemisphere specializations. This group would include more females than males. In people with only one rs gene the right planum would not be developmentally delayed. These people would tend to be right-handed but they would have a relatively strong left hand (this could explain why identical twins don't necessarily have the same handedness), and they could have either hemisphere specialized for language; many would show standard left hemisphere specialization, but in any of these individuals dominance could go either way. The one rs gene category could include the group of gifted students Winner identified as having "more anatomically and functionally symmetrical brains, with language less lateralized to the left hemisphere and visual-spatial functions less lateralized to the right side" (1998, 161). Enhanced right hemisphere use for mathematical or verbal reasoning correlates with precocity in these areas, as shown on SAT results. Witelson (1985) finds gifted mathematical and verbal reasoners of both genders to be less lateralized for syllable listening and face perception tasks. Geschwind and Galabura (1987) also describe a group, left handed or ambidextrous, who use both right and left hemispheres for both speech and mathematics. Interhemisphere connectionWe have seen that IPL areas more often than not show marked lateralization of representation-related function. Lateralized specialization is practicable because there are large fiber tracts connecting cortical areas with mirror image structures in the opposite hemisphere. The corpus callosum is the largest of the interhemispheric commissures. The corpus callosum is like the IPL in being myelinated late in development. Its thickness increases into the twenties, through a period characterized by gradually improved bimanual coordination as well as increasing representational skill (Schlaug et al 1995, 1048). Callosal development readily shows practice effects; the size of the anterior corpus callosum, which connects motor and premotor areas in the two hemispheres, increases with early training in musical performance, which requires coordinated but highly asymmetrical motion of the two hands (Schlaug et al 1995). More symmetrical brains have more developed callosa (Witelson 1985). Imaging studies often find bihemispheric activity for representational tasks. The von Stein EEG study that found intrahemisphere temporal-parietal synchronization in all representation tasks also found interhemisphere coherence increases at the same frequency (1999, 141-144); the study reports strong interhemispheric coherence between ventral temporal areas in the two hemispheres during pictorial object recognition, for instance (1999, 144). Asymmetrically lateralized left and right intraparietal areas appear in these instances to be working together on cultural tasks. In many representational circumstances, specialized skills of both hemispheres will be needed. Sampson and Zatorre find that in a musical task the left hemisphere responds to the words of the song while the right hemisphere responds to melody and chord progressions (1991). There are reasons to think a complementary tying-together of the hemispheres is important also to language. The left lateralization of linguistic function is well established, both for spoken and for signed natural language, and we have seen that intensive left hemisphere integration must be necessary to tie anterior, posterior, dorsal, and ventral regions into a linguistic wide net, but conjunction analysis of PET results by Papathanassiou et al (2000, 348) shows contralateral homologues of Broca's and Wernicke's areas also implicated in both comprehension and production of language. Right hemisphere aspects of language already mentioned have included intonation, expressive and referential gesture during speech, emotional comprehension, linguistic play, and a sense of alternative meanings. All of these right hemisphere specializations have been described as pragmatic or paralinguistic, not part of the syntactic core of language. There is evidence however that right hemisphere involvement may also be central to making sense with language. A right hemisphere linguistic disorder called semantic-pragmatic language disorder is characterized by "copious and complex prolix hyperlexia" (Shields 1991), that is, by syntactically competent floods of nonsense. Bellugi describes what sounds like a similar right hemisphere defect: Williams syndrome is a form of congenital retardation characterized by sociability and fluent speech together with remarkable deficits in spatial competence and physical concepts (Bihrle et al 1989). Correlation analysis of PET results has shown task response in one hemisphere preceding response in the other, as if a first stage is accomplished by one hemisphere and a later stage accomplished by the other. The direction of precedence depends on the nature of the task. A study of face recognition and dot location tasks, for example, found that response in what and where tasks, both designed to minimize language involvement, happened first in the right temporal and parietal, and then in corresponding left hemisphere areas (McIntosh et al 1994, 655). Hemisphere precedence of this kind is often spoken of as if 'information' or 'code' or 'an engram' is shipped from one hemisphere to another, but we could also think of structures achieved in one hemisphere, language-related structure for instance, as instigating and organizing a different kind of structure in the other. Counterpart areas in the two hemispheres will not necessarily be doing the same things. This way of thinking is compatible both with hemispheric specialization and with interhemisphere coherence.
We have seen that the development of the inferior parietal is a precondition for every kind of representational function. When counterpart structures in the two hemispheres have specialized functions both of which are necessary to representational tasks, is the IPL, in particular, active in linking the hemispheres? Given its location on either side of the posterior corpus callosum, it seems likely. Dreaming and representational function may be somewhat alike in their use of the IPL. Solms reports that if either the left or the right IPL is lesioned, dreaming ceases altogether. Because global anoneira is reliably correlated with Gerstmannn syndrome, which is a left hemisphere syndrome, Solms believes that during dreaming the left IPL organizes the right IPL, which in turn organizes secondary visual cortex bilaterally (1997, 140-141, 166-175).
I have described our use of representational objects and events as perceptually managed simulation: we often perceive a representational form in order to imagine something else altogether. An interesting aspect of the perception/simulation duality in representational function is that the attentional axis organizing perception of a representational object can be incommensurable with the axis implicit in the simulation it evokes. Kubovy (1986) gives the example of our situation when we look at a painting hung at eye level; our actual task axis is horizontal, but the painting may have a perspective organization implying that we are looking upward. Or we may sit looking upward at a screen whose moving images imply we are falling from a plane. It can happen with language too. We may be listening to someone sitting next to us on our left who is saying There was a steep drop on the right of the track, with the river far below. We manage these incommensurabilities easily. How? When speaking or listening organizes response to our actual location, (Watch out below, says the construction worker) the right hemisphere is presumably setting up a widely integrated perception-action net in response to rapid linguistic discriminations made on the left. When we are talking about something that isn't present, could the right hemisphere be providing expansive simulational structure instead? It may be that, for representational uses, we perceive and imagine by means of different hemispheres, left hemisphere perception of words organizing right hemisphere scene simulation, or right hemisphere perception of black and white lines on paper organizing left hemisphere object recognition and naming. The lateralization of the IPL may be necessary to sustaining factive and fictive cortical structures simultaneously required by representational function. Taking speculation a step further, the ability to perceive and imagine at the same time may have begun in practical abilities to be simultaneously about two things in one place -- or about one thing at two different scales, or by means of two different senses, or for two different purposes. The development of a bifocal parvocellular vision system alongside an existing magnocellular system makes it possible to interact with nearby objects with three-dimensional exactitude, and the energetic hyperactivation and greater resolution of foveal vision sets up wider, more integrated, more energized, more synchronized, object perception networks. As seen in previous chapters, these parvo-based networks seem to be anchored in the ventral stream, and may include many kinds of temporal lobe matrices -- feature matrices, category matrices, object-constant matrices. When it is involved in organizing action or readiness in relation to a particular focalized object, a hyped, recurrently synchronized object-perception subnet of this sort might be set up predominantly in the hemisphere contralateral to the preferred hand. A creature in danger from predators or conspecifics has to keep environmental readiness even or especially while intent on plucking nits or fishing for termites. In these circumstances, general environmental readiness may amount to a second attentional axis, another, separate object-perception subnet. It too would be an extended and coherent net, and it too would be active in organizing motion, but visually it might also be more magno-derived, and thus less energized, less detailed, perceptually thinner, in effect less conscious: a background network, in other words. Since it is organizing readiness of a different kind on the basis of a different perceptual mix, the second network might best be segregated in a second hemisphere. Here, then, may be a thread that ties the evolution of parvocellular vision to the eventual development of language. The magno system, which uses peripheral and bilateral vision and controls peripheral and bilateral response, would set up a less-ventral, less consciously visual, and more somatic network in the right hemisphere. The parvo system, using the convergently directed center of both eyes, would in the meantime set up a more ventral, more visual network in the left hemisphere, which is specialized for the sorts of hand and mouth motion organized from areas of cortical magnification. In animals without lateralized hemispheric specialization, either hemisphere might respond either way. Lateralized specialization when it occurs may in fact be a specialization for network type. In humans the left hemisphere may be permanently specialized for setting up foreground object perception/action nets, and the right hemisphere permanently specialized for setting up background object perception/action nets. Environmental action with its generally symmetrical motor behavior using both sides of the body would as a rule be directed from the right hemisphere dorsal stream. Focal behavior with its generally asymmetrical specialized behavior of the favored hand would be directed from the left hemisphere dorsal/ventral grasp stream. Both systems would makes coordinational use of the corpus callosum. This division of labour could allow the development of language, in which a focal hemisphere perceives and acts at a technical and notational microscale while the other hemisphere continues to track basic level events and objects. When, as often, language is used in relation to basic level events and objects, the corpus callosum would connect grammatical with semantic aspects of language function -- the left hemisphere would discover and create components of linguistic form, and the right hemisphere would guide and follow it with perceptual or simulational activity ( -- guide during language composition, follow during language comprehension). The right hemisphere homologue of Wernicke's area is thought to be involved in understanding discourse wholes. When we hear or read extended passages, we will have to sustain situational simulation as more aspects and details are added. A sense of the layout of a location may be a good framework for sustained elaboration, since background layouts are generally stable while foreground attention shifts and foreground objects change. Isolated left hemisphere language might lack semantic coherence because it lacks the participation of basic simulational aboutness. The right hemisphere may be important to the comprehension of sentence and textual wholes because these wholes are forms of whole-situational simulation. When we write or read there is usually little perceived background. We don't see the paper, only the letters. But there is also little perceived foreground. Instead, a series of very small nearby objects is very rapidly registered in focal detail and to precise effect, but in such a way that we are not conscious of perceiving them. We have to be perceiving in great detail, but also at great speed. It may be that the speed does not allow us to accumulate object perception subnets of the kind that support sentient vision. It is more likely however that we do not see the letters because we are instead visually preempted by simulational relation to whatever we are reading about. Simulation becomes foreground, by means of a temporally sustained hyperactive subnet.
Simulation-based accounts of linguistic semantics run into objections from people who say they do not imagine anything when they use language. There are in fact many sentences that are not easy to account for in simulational terms. Linguistic devices often seem to evoke schemata rather than situations. Linguistic effect can be very abstract, more like a calculus than like a narration. Recall Kosslyn's description of connectivity in right and left hemisphere IPL areas, right hemisphere neurons having longer, more widely connected axons, suited to widely integrative response, and left hemisphere neurons having shorter, more locally connected axons, suited to fast, habituated switch-like response. Linguistic response at the microscale of phonological perception and production would seem to require just such rapid habituated response. These mechanical aspects of communication belong to perception and action -- the part of representational function that has to do with presence. What about the part of representing that has to do with simulation? Do fast, overlearned language circuits evoke fast overlearned simulational structures? Some uses of language evoke rich simulational structure, with sensory and motor participation at many logical levels -- what I described in Chapter 7 as semantic depth. In these instances, often stories or poems, we imagine thoroughly. In other instances -- and possibly always, in some people, because there seem to be large individual differences -- we may imagine so swiftly and sketchily that we hardly seem to be imagining at all. An instance of the latter might be what happens with rapid habitual use of spatial relation words, as when we say something is above/below, right/left of, inside/outside of, connected to/disconnected from, something else -- what Kosslyn calls an abstract or conceptual sort of spatial cognition. Are Kosslyn's conceptual spatial abilities minimal versions of structures that, evoked on the right, would be part of more comprehensive situational simulation of the sort set up when we understand a story or design a garden? Are left hemisphere syntactic and semantic contrasts a sort of calculus of minimal simulation? If so, circuits making minimally connected spatial distinctions on the left might under some circumstances also set up more widely connected right hemisphere circuits via counterpart structures accessed through the corpus callosum. Or it might happen in the other direction: widely connected right hemisphere spatial simulation might evoke left hemisphere minimal simulation, which would in turn evoke a spatial name like under or beside. Left hemisphere spatial functionOnce specialized for language, the left hemisphere cannot take over the spatial and practical functions of the right hemisphere: does the lateralization evidence mean language rebuilds structures built for sensory-motor purposes? If so, does language retain evidence of such rebuilding? Are left hemisphere linguistic responses procedural metaphors?
The question particular to a cognitive theory of syntax is a question about the systematicity of grammar, the nested structure of contrasts described in formal linguistics. How is it that structures evolved in prelinguistic contexts can lend themselves to the systematic subordinations and inflections of sentence grammar? Lesions of Broca's area in the forebrain damage fluent articulation, as would be expected from its role as a premotor area. But Broca's lesions damage sentence grammar as well as phonology and word morphology. Broca's damage to grammatical competence is specific to closed class language forms such as pronouns, prepositions, connective particles, and verb inflection, all of which are lost with Broca's lesions, while open class verbs and nouns are retained. Is the distinction between syntax and lexicon in linguistics embodied in an anterior-posterior gradient along the Wernicke's-Broca's stream? We have seen that Broca's may have developed from a division of an area something like the macaque F5 grasp-related premotor area coordinating hand and mouth motion. Are closed class syntactic devices left hemisphere uses of structures that organize hand-mouth synergies? The hearing are fully able to understand speech without gesture (as when we listen to the radio), but in face-to-face speech we gesture abundantly. We even gesture when we talk on the phone. Is gesture accompanying speech evidence that language structures have colonized areas that organize the hand? We do know that the rhythmic structure and temporal scale of a gesture is similar to the rhythmic structure and scale of the sentence, that hand and mouth action is coordinated from adjacent and overlapping forebrain areas, and that Broca's lesions may damage hand praxis as well as language. We also know manual languages have the same grammatical structures as oral languages, are acquired by children at the same time and in the same way, and are cortically lateralized to the same hemisphere. We have seen that macaque parietal and premotor subnets can be understood as embodying motor decisions of different kinds and at different levels in a hierarchy. In both parietal and prefrontal areas there are cells of very general action types; cells that respond during grasping, holding, tearing, or manipulating, whether by the right hand, the left hand, or the mouth. There are cells specific to grip types: power grip, precision grip, and others. There are cells active during brief or extended temporal segments of an action. There are cells selecting for size, orientation, or kind of object. Each of these cell types can be thought of as deciding within a small set of possibilities: object positions in near space, act types, prehension types, object types, and so on. Cells active at different times can be understood as participating in nested cycles within subordinative hierarchies of action timing. Like reach and grasp, motor aspects of speech behaviors require hierarchically nested small sets of motor decisions at several time scales. For signed languages these are decisions about hand location, hand shape, hand coordination, and hand motion -- which includes rhythm, speed, repetition. Oral speech requires decisions in relation to, for instance, mouth shape, tongue position, and breath control. A syntax is a system of linguistic possibilities. Verb choice, for instance, requires decisions about tense (past, present, future), person (first, second, third), and number (singular, plural), all of them very small sets of contrasted options. Decisions required in both grasp-related and language-related hand and mouth action are motor decisions. Is there a relation between the two sorts of linguistic systematicity, motor systematicity and syntactic systematicity? Can motor schemas for mouth and hand motion also have something to do with syntactic contrasts? Is left hemisphere motor structure originally used to make systematic motor decisions now being used (act-metaphorically) to make act decisions that are syntactic decisions? Rizzolatti describes the AIP-F5 parietal-prefrontal grasp stream as embodying a sort of language of mouth and hand action related to prehension (Rizzolatti 1995, 439), with composable elements and hierarchical organization, like a grammar. He is also particularly impressed by the possibility that Broca's has mirror cells of the kind found in the macaque F5 -- cells that combine fine-grained movement analysis with social act empathy. If this conclusion is correct, the functional specialization of human Broca's derives from an ancient mechanism related to production and understanding of motor acts. From this mechanism evolved, possibly in relation with the development of a more complex social life, first the capacity to make and interpret facial communicative gestures, and the capacity to emit and understand 'verbal gestures'. Rizzolatti et al 1996a, 137 Cells in parietal and premotor areas, tested by single cell readings and microstimulation, often show combinations of sensitivity. Di Pellegrino et al found a range of kinds of response in mirror cell areas, some cells responding only to observed acts, some responding only in the process of executing an act, and some responding similarly to observed and executed acts (1992). This combinatory flexibility may in the end also support, for instance, aspects of the inflection of a verb.
Chapter 9. Kantian stories |