Amobarbital

The laws of nature are constructed in such a way as to make the universe as interesting as possible.

-Dyson's Principle

DOES DYSON'S PRINCIPLE APPLY to the universe within? If the human brain were homogenous, with all parts equally involved in all intelligent activities, then its manner of operation would be inscrutable and not very interesting. (Such a suggestion has been made under the heading "mass action," but it has been rejected-see Vernon Mountcastle's essay in this volume.) If it were instead composed of discrete components (or "modules") like a machine, then in principle its composition would not be hard to determine, though how such a conglomerate might have evolved would be perplexing.1 In reality the brain is more interesting than either of those blueprints suggests. It presents a paradox: It is a completely connected nerve net, and yet it is highly differentiated in its parts. How can those two characterizations be reconciled?

COGNITIVE NEUROPSYCHOLOGY

Nobody knows how the brain works. But even a staggeringly complex system can be implemented by combining a limited set of building blocks, and evolution notoriously relies on a limited inventory of tactics, variously applied. We are familiar with some of the design characteristics of the neural substrate of human cognition, largely through one hundred and fifty years of observing the effects of brain injuries on people's ability to experience, think, and act. Supplemented by the patients' subjective reports, these controlled observations indicate fault lines along which cognition falls apart, thereby offering hints as to what its components, or primitives, might be. This branch of neuroscience is called cognitive neuropsychology. Additionally, many of these case studies contribute to a mapping of cognitive functions on the brain via evidence about the localization of the causative lesions (brain-behavior relationships). The process involves zeroing in on the cognitive operation-brain localization pairing: The patient with a focal injury fails at certain tasks; the investigator hypothesizes about the aspect of the tasks that presented difficulty; he then designs novel tests that more selectively probe for the hypothesized mental operation; and finally he determines whether a more reliable behavior-brain relationship is understood than before. Step by step, the nature of the cognitive primitive to be ascribed to the damaged location is crystallized.

I shall present neuropsychological observations-usually my own, but consistent with those of others-that will demonstrate that the human forebrain is exquisitely functionally differentiated and yet quite integrated. Next, I shall show that the way functions are organized in the brain corresponds in principle to how the cerebral nerve net is organized neuroanatomically. I shall then offer some suggestions as to how this organ, which is both integrated and differentiated, might work.

Most informative are local and specific injuries. Examples include limited areas of the brain deprived of circulating blood on account of the occlusion of an artery (stroke), areas replaced or compressed by a tumor, and areas that disintegrated under the impact of a blow or a missile or in the wake of a localized infection (brain abscess). Injured in such ways, people may no longer be able to do some things that they could do previously. They now exhibit a selective cognitive deficit, whose nature depends more on where in the brain the injury is located than what caused it. The smaller the lesion, the more likely it is to compromise only a very few of the things that a person can do, leaving the rest of the cognitive profile roughly intact. In contrast, large lesions are apt to impair a whole cognitive domain, such as language. Small lesions could compromise a subskill within that domain, leaving the other subskills functional (for instance, the ability to name objects aloud or, in another lesion location, to repeat sentences). My purpose is to illustrate what types of specializations exist, how they relate to each other at the cognitive level, and how in general they cluster in the forebrain.

FINE-GRAINED DISSOCIATIONS

Highly selective deficits have been documented in every aspect of cognition. Consider a patient with damage to the back of the left cerebral cortex, who can write but cannot read what he has written (alexia without agraphia).2 Or a patient with a more extensive posterior lesion, involving both sides, who cannot recognize common objects when they are shown to him yet can identify them by touch, copy them, and draw such objects from memory without a model (associative object agnosia).3 These are "dissociations" within a single person. This sort of patient performs normally on all but one type of task. One can also compare people with different brain injuries. When one brain-injured person fails on task A (say, spelling aloud) while succeeding on task B (say, spelling in writing), whereas another patient performs in just the reverse way, a "double dissociation" between the two activities is said to obtain.4 The same applies to the patients who can understand what is said but not repeat it (conduction aphasics) and those who can repeat speech verbatim but have no idea what it means (transcortical motor aphasics).5 Evidently the brain handles each one of the two dissociated activities separately.

The fine grain of cerebral specialization is remarkable. There are patients who can recognize faces but not letters (alexia), or letters but not faces (agnosia for faces). One patient's difficulty was limited not just to handwriting (apraxic agraphia), but solely to writing in cursive.6 A patient could identify objects, explain their use, but not retrieve their names; he could, however, name them when he felt them in his hand.7 Another patient could not name the colors or recognize them when named, yet he could recognize and name black, white, and gray (color agnosia).8 When competent investigators report a dissociation in a single case, that counts as an existence proof-an aspect of brain functioning has been revealed. Brain theory must accommodate the finding.

Right-Left

The instances that I have cited are notable for their discreteness The deficit scales up with larger lesions. To study these, group comparisons are more often used than single-case experimental designs, so one may compare right-lesioned patients with those whose injuries are on the left. Splinter deficits that result from fine-grained dissociations will be submerged in the statistical treatment of the group data, but broad distinctions emerge. We learn that most people's left cerebral hemisphere is concerned with language, their right with spatial orientation. More broadly still, the left caters to sequential analysis and the generating of action sequences, the right to setting such activities into a spatial framework.9 Most general of all, the left hemisphere controls motivated approach sequences(handling, eating, and so on), progressively focusing and acting upon the target; the right hemisphere is more involved with the person's movement through the intervening space and the spatial background of the target. The left hemisphere's activities can be context free, whereas the activities of the right are context bound. The hemispheres are complementary in their functioning.

Unsurprisingly, the left and right hemisphere mediate different types of attention, roughly focal and global, respectively.10 Patients with left-hemisphere lesions, who therefore rely largely on the right hemisphere, can perceive and reconstruct the general shape of a thing, but not in analytic detail. Patients with righthemisphere lesions can serially extract the details, but not articulate them into a coherent whole.11 Global attention encompasses the whole display and is bound to that context. Focal attention disregards context while it targets specifics.

Comparable observations have been made at the level of concepts. Patients whose right hemisphere was temporarily inactivated (soon after electroconvulsive shock treatment for depression) could accept and respond to questions (syllogisms) that assume counterfactuals, such as "Canada is on the equator. Is it hot in Canada?" They could disregard what they knew to be the case and assume the hypothetical. When the same patients' left hemispheres were inactivated, they protested the factual misstatement and refused to respond. They were context-bound to their knowledge base and could not decontextualize even conceptually. (What happens in the intact individual, in whom both hemispheres are available for use? A compromise. The patients responded more consistently to the counterfactual syllogisms when operating only on their left hemispheres than when operating on both.)12

The above conclusions are drawn from studies of patients with damage varying in extent but falling short of implicating the whole hemisphere. Total removal of a hemisphere is rare. But it is possible and sometimes clinically appropriate to simulate hemispherectomy for a short time, measured in minutes. One technique is one-sided electroconvulsive shock (ECS). This can relieve depression that resists medications. Another is the injection of intracarotid amobarbital (ICA). The hemisphere on the injected side (save the rearmost portion) is inactivated for about three minutes. The dysfunction that results affords a preview of what would occur if surgery is performed on that hemisphere. This procedure is most commonly used to screen for possible adverse effects on memory from a temporal lobectomy for intractable epilepsy.

The use of ECS and ICA reveals that people remain conscious when only one hemisphere is active, regardless of which one it is. Therefore, an indispensable "consciousness module" cannot exist in either hemisphere. Cerebral lesions in various locations restrict the richness of the contents of consciousness, but no one focal lesion can abolish it outright. Consciousness appears to be a property of the global network, not of an elite localized conscious awareness system. With respect to language, we learn that one in five left-handers programs language in the right hemisphere, and just as many use both.13 What is more, ICA performed on patients rendered aphasic by left-hemisphere strokes reveals that in some cases the right hemisphere takes over the language role.l4 The language function must have arisen from variations on the properties of the global network. Right-hemisphere compensation would not be possible if it relied on unique, recently evolved circuitry. Correspondingly, we know from anatomical studies that the language cortex relies on evolutionarily ancient cortical circuitry.15

Front-Back

An equally sweeping contrast obtains between the front and the rear (anterior and posterior) cortical regions. The prefrontal cortex exerts control over posterior activities by means of corticocortical connections that transmit along separate parallel lines in both directions.l6 It thereby brings influences to bear on the local posterior cortex-from the physical and social environment, and from the person's wishes, beliefs, preferences, and apprehensions (arising in the limbic cortex).17 Correspondingly, rather than impairing a specific skill, prefrontal lesions handicap mental agility, for example, the ability to change one's mind-set rather than persevere with an unsuccessful strategy. In this way the prefrontal cortex, and specifically its dorsal subdivision, enables the setting of priorities depending on the situation.18 In contrast, orbitofrontal lesions release impulsive responding, the uncensored drive-directed activity of the remaining intact brain. The lateral prefrontal cortex does cost-benefit accounting of the proposed act while the individual hesitates. Absent this accounting, responses are quick, incorrect, or hazardous. Phineas Gage, the famous prefrontally injured railroad worker, is a classical example. Premorbidly a conscientious worker, after an iron bolt tore through both his prefrontal cortices he became feckless and disorganized, unreliable and unemployable. His intelligence remained intact.19

Yet the prefrontal cortex is not a general "central executive." It is comprised of a set of control systems, each with its separate target.20 As their connectivities indicate, different prefrontal areas control different activities that are separately localized posteriorly. In each case, the prefrontal contribution enables the individual to overcome primitive, preprogrammed responding when that would be maladaptive. It also applies in the temporal domain to action sequences. For the brain, unlike for a machine, ceasing to do something is a positive act, in the same way that starting it is. Processing sequences do not just run dry in the sands, and some patients exhibit run-on behavior. An example is the jargon aphasic, who rattles on well after he has, at least to his own satisfaction, responded to the question. He may even put a stop to his harangue by saying "stop" (Luria's verbal control of behavior).20 In broad outline, left prefrontal lesions impair planning and right prefrontal lesions impair interpersonal relating (leading to so-called pseudopsychopathy). Both describe Phineas Gage. Again we see an overall network-this time prefrontal, with a particular potential for steering cognition-that is partitioned into subnets that exert the same general type of influence in different contexts and toward different goals.

GLOBAL IMPAIRMENTS

Intact people are able to do more than one thing at a time; how well they do so depends on whether both activities demand attention (i.e., are not automatized). If they both demand attention, then whether they can be performed in parallel depends on what I have called the "functional cerebral distance" between them.22 The less neurally interconnected the two areas that guide the activities in question are (i.e., the further apart they are), the more easily the activities can be performed at the same time. Activities that are based on highly interconnected areas interfere with each other to the point that a person can only do one at a time. This illustrates the fact that how the brain is organized constrains what people can do. Vernon Mountcastle's essay in this issue refers to the ongoing study of functional space at the neural level.

If the global network is extensively depleted of neurons, as in the dementias and to a lesser extent even in so-called normal aging, regressions occur that are domain-general.23 Modular discontinuities are not respected. The demented person increasingly relies on familiar, well-rehearsed, or biologically prepotent routines. The brain states that remain available exhibit inertia. Once implemented, a brain state is unduly persistent, which causes the person to exhibit a rigid and inflexible personality. He relies on well-trodden paths of attitude and action; in effect, he becomes like himself, only more so. The inertia of state transitions plays out not only as a difficulty in adapting a mental set to changing conditions, but even as a general slowness in processing information. At its extreme, dementia restores constraints that bind the infant, as when primitive reflexes of infancy resurface-movement patterns (synergisms) that normal adults automatically suppress. The restriction in cognitive state space is compounded by restriction in movement state space.

NESTED COGNITIONS, NESTED NETWORKS

We emerge with an understanding of cerebral organization as an architecture of nested functional units. Within an overall sequential analytic mode, the left hemisphere generates word sequences in one of its parts, action sequences in another, and sequential identification (for instance, of letters) in yet another. Within subareas of the respective areas, it enables both recognition and expression. In subareas of those subareas, it enables material-specific processing, for example, of action words, color names, names of animals, or the connotations of "right" and "left." Within its overall spatial relational mode, the right hemisphere enables orientation in ambient space in one of its parts and pattern perception in another. In subareas of these subareas, it permits subskills like map reading, face recognition, or identifying sketched shapes ("perceptual closure"). A recursive organization emerges; superordinate processing modes differentiate into distinctive domains of functioning, and these again into different specific applications. I envisage a relatively undifferentiated whole that progressively differentiates (in child development? in evolution?) into the rich set of specifically human potential skills and alternative strategies.

The recursive organization of the "cognitive profile" finds its counterpart in the recursive organization of the cortical neuronal network. The cortex is conceptualized as a recursive network or "net of nets," constructs that are vividly realized in cerebral organization: The global network is composed of separate parallel cortical "trends" (and patches of multimodal cortex).24 Trends are sequences of processing stages. One instance is the "dorsal visual stream" that interfaces with visual input, especially from the peripheral fringes of the visual field in visual area one (V1). V1 is the first cortical relay for visual input; its more internal anchor is the less differentiated and earlier-evolved parahippocampal gyrus. Another trend, the ventral stream, also relates V1 input but more from the focal center of the visual field to less differentiated and earlier-evolved stages, culminating in inferotemporal cortex. Anatomical minutiae apart, the implication for brain organization is that trends are composed of individual sequentially connected processing units called stages (with each stage consisting of three parallel subsections). The latter in turn consist of sets of columns of cortical cells. So I (that is, the brain that is me) am a network (global) of networks (trends) of networks (stages) of networks (columns). The recursive organization of cognition arises naturally from the recursive organization of cortical nets.

The term "stream" is misleading in that it implies a unidirectionality of flow, which is not the case. Trend is a more neutral term that can accommodate the fact that activation can flow in both directions: from the outside in, signaling a perturbation of the system, and from the inside out, preforming the anticipated sensory consequence of the act in progress.25 The two waves of activation meet, forming a standing wave across the elements of the trend. The conscious content of the activation pattern that results is an inextricable amalgam of represented anticipation and represented perturbation.

It used to be believed that much of the human cerebrum was dedicated to higher mental functions, and only a small part to sensorimotor input and output. We now know that all of the cortex originally evolved as a series of adaptations to the requirements for sensation and action. Our ability to manipulate and abstract is superimposed on this wealth of parallel sensorimotor representation, not localized in a separate part of the brain. Again, evolution pours new wine into old bottles.

REALITY CHECK

Are the localizations reliable? Can one directly interpret cognitive deficits in terms of the preexisting skills that the affected areas of the brain normally contribute? In short, are the lesion effects "transparent" with respect to the underlying mechanisms? Given the vagaries of localization, of plasticity, of compensatory activity, of premorbid strengths and weaknesses, can they be?

Lesion site and type of deficit do not exactly correspond in different patients. But neither does a given "splinter deficit" result from lesions in widely different locations in different patients. We are not mistaking the electric cord for the light bulbcutting the cord, seeing the light go out, and supposing that the light source was where we cut. If we were, we would observe the same narrow deficit as a result of damage in many different parts of the brain. This does not happen. The fine grain of human cortical specialization is unquestionable, as demonstrated time and again by the controlled study over years and decades of highly selected, focally injured patients.

The brain operates in terms of interactive groups of neurons that excite or inhibit each other, two-way connections between neuronal assemblies, and neurohumors that transmit coded information (neurotransmitters) or adjust the set point for change in the rate at which particular cell assemblies oscillate (neuromodulators). We are far from being able to identify the neurocognitive primitives, the particular patterns of firing for particular cell assemblies that underlie any human activity at all. But we can propose generalizations about how the system works as well as claims about how it does not work. I base the latter proposals on what I call "nonexistence proofs."

NONEXISTANCE PROOFS

The nonexistence proof is an essential corrective to the extravagances of inductive reasoning. By carefully selecting his instances, a theorist can support any one in too wide a range of possible solutions. The nonexistence proof invokes the negative instance. It enables us to weed out solutions that are nonstarters and alerts us to some ways in which the brain might not work.

In neuropsychology the nonexistence proof specifies conceivable deficits that do not in fact occur. But how can we prove a negative? How can we ever be sure that a particular form of malfunction does not, and could not, occur? By all means, we are constantly surprised by new findings, but the following remains a viable working hypothesis: If a particular malfunction has not been discovered after a century and a half of diligent search by neuropsychologists the world over, it probably does not occur. Our preferred model of brain function should then be one in which such malfunctions would indeed not occur.

The nonexistence proofs that I offer point in an interesting direction. Rather than being assembled piecemeal and glued (conjoined, integrated) together, the percept, construct, utterance, or intention gradually differentiates out of the preexisting brain state.26 Diversity is continually being carved out of the existing unity. The operative question is not "How are the details assembled into a whole?" but rather "How is the whole reshaped to incorporate the details?" What follows are some mistakes patients do not make.

Patients with focal brain lesions do not violate the rules that governed their premorbid performances. They remain guided by preexisting parameters but become less specific, slower, and less stable in their responses. Preexisting constraints become more stringent, and the boundary conditions for the impaired behavior become more limiting. Properly scrutinized, apparently qualitative differences turn out to be quantitative differences. Patients retain high-frequency responses (e.g., names of familiar things) while losing low-frequency ones. Patients continue to operate within the relevant domain even if they mistake or misspecify the exemplar, with errors that conserve the implicated domain being the most common. No real focally damaged patient mistook his wife for a hat. He might, however, mistake her for another person or mistake his hat for his scarf. This point is illustrated by "deep dyslexia," in which words are often misread, not for words that look or sound similar, but for words with similar meanings.27 In the domain of action, an apraxic person who fails to make a fist does not do handstands instead; he does something of the same general nature as the act that was requested. The patient with a visual-recognition deficit (agnosia) may know that he is viewing an animal, but not which one. He may even distinguish possible from impossible pictured objects, without knowing the identity of the possible objects. In short, in the impaired domain, the patient's cognitions are simplified, primitive, even rudimentary, but not fanciful, perverse, or wildly irrelevant. These principles befit a recursive organization. The totally lesioned area, itself incorporated, is nested within a broader area that deals with the same domain as the lesioned "module," though in a less differentiated manner.

Within a specialized network, individual responses are not further structurally differentiated and segregated; rather, the full set of responses arises from patterned activation of the entire network unit. Thus within the affected domain the deficit applies to the type. Individual tokens do not become selectively unavailable. A patient with word-finding difficulty (anomia) does not selectively have trouble with all words that begin with the letter f or are bisyllabic or have an unpleasant connotation, let alone with one particular word, sparing the rest. I conclude that the same network handles all tokens of the implicated type. What constitutes a type, however, is determined by the brain. The special facility for face recognition makes faces a type; within that type, there are no selective losses. The hand in action is perhaps also a type. But apparently no comparable aggregation of circuitry deals with bodies, or autographs, or infants that do not yet walk. There is no general rule by which one could predict what aggregates and what dissociates in the brain.

The principle of domain-specific dedifferentiation is well illustrated in the way in which aphasics generate sentences. Among the conceivable subtypes of lesion-induced language disorder, one is conspicuously absent: a jumbling of the sequence of uttered words. Words may be omitted or simplified, for instance, resulting in lost inflection, but no aphasic "order the himself wrong in expresses." I conclude that one way that sentences may not be formed is by lining up words like beads on a string. Instead, I infer that sentences differentiate out of less-specific preconscious precursor states, with the word order implicit in the precursor state. The brain models and remodels until the utterance is perfected in its analytic detail.

Order errors do occur in impaired verbal performance. Order errors in spelling (a specifically learned skill) characterize a particular syndrome of left-parietal impairment.zs When people, normal or not, repeat arbitrary word sequences, they regularly make order errors. But word-order errors do not characterize neuropsychological impairments in naturally developed language.

In perception, again, one does not see "assembly errors." Percepts do not appear to be strung together, because braindamaged patients do not report them to be misstrung (say, a cat with paws protruding from its head and its ears on its tail). Analogous regularities characterize sequential action. In ideational apraxia, in which the ability to execute familiar sequences is impaired, the action sequence becomes shortened by the omission of one or more components. But the components are not reversed or jumbled in their sequence.

Also eloquent in its absence from the literature is the partial disruption of a "mental model" of the surrounding world or of one's own body. When incoming information is obstructed by a lesion somewhere between the retina and primary visual cortex, patients have blind patches in their visual fields (scotomas). But they do not have holes in mental models; they do not complain of systematically being unable to discern objects in a particular egocentric relative location, however they look at their world. Deficit in a visual mental model should be apparent regardless of retinal angle of view, but such mental models do not appear to exist. The same reasoning applies to the so-called body schema. If there were a representation (a model) of the body parts in the brain, then focal damage should in one case implicate one part (the right elbow) and in another case, another part (the left hip). Putting together localizations from many patients, we could then assemble a cerebral topography for the body schema, like assembling a jigsaw puzzle. But such regional imperfections in the body image (partial asomatognosias) have not been reported.29 As Andy Clark in this issue infers on other grounds, we do not build up models of the world and our bodies for reference purposes. Instead, since we carry our body and our world around with us, we simply refer to them when necessary.

The nonexistence of a deficit illuminates another issue of current interest: the alleged "binding" of visual features into perceived objects, of objects into scenes, scenes into episodes.30 I assume that the brain can be stopped from doing anything it customarily does by an appropriately placed lesion. Some patients certainly report that objects look distorted (metamorphopsia) or persist in view (like afterimages) after the patient looks away (palinopia). But how would a patient describe his failure to bind, for instance, a color to a shape? Given the limitations of iconic memory, anyone might misremember which color belonged to which shape when briefly shown multiple, diversely colored, different shapes.31 But primary failure to bind should express itself differently, as in "I see some colors and I see some shapes, but they do not go together." No such free floating of the colors of shapes or the shapes of colors is on record.

Another form of presumptive binding that lacks neuropsychological reality is cross-modal binding. No case had been reported in which within-modality perception pursued a normal course, but the percepts could not be combined across modalities. In a centered brain we would expect that there is convergence to a highest level in each modality, the activity of which incorporates the unified experience within that modality. But no such "summary module" is to be found.32 On the contrary, it is the earliestevolved cortical sensory areas that tend to be multimodal, and no single one of them encompasses all the modalities.33 Earlier in brain evolution, the modality of input did not differentiate out. The trick is not to combine modalities, but to differentiate them. The nonexistent deficits all send the same message: experience is not a composite assembled out of its parts. The contrary position-that experience is carved out of a less differentiated whole-gains plausibility. While no truly apt metaphor for how the brain works comes to mind, "crystallizing out" seems more fitting than "assembling together." Interestingly, brain development proceeds according to similar principles. The newborn has a full complement of neurons; further development proceeds by selective cell death and elimination of synaptic connections. The biological chisel prefigures the microgenesis of brain states.34

ATTENTION

The traditional model of brain organization in effect consigns attention to a central overseer-a central processing unit, central executive, a homunculus or femincula. This is unavoidable if information is integrated by multiple stepwise convergences to a central decision point (derisively pinpointed by William James as the "pontifical cell"). Decisions would then be implemented along a conversely diverging series of stages.35 Carl Wernicke set the agenda a century ago: "To find the route, the telegraph line, by which the telegram is conveyed."36 Attention would resemble Jeremy Bentham's synopticon-a neural "beam" that emanates from a central "lighthouse" and illuminates the action anywhere in the brain at will.

Contemporary neuropsychology steers us in a different direction. There is no forebrain lesion that selectively abolishes the ability to attend, across domains (nor one that selectively ablates consciousness).37 Some lesions of the network result in altered behavior that the residual intact network takes cognizance of. Other lesions simply remove the affected domain from the sphere of consciousness, and the patient experiences neither function nor malfunction in that domain. It follows that the individual is aware of some types of deficits; of others he is not aware. He is aware, and complains vehemently, of difficulty with word finding (anomia), calculating (acalculia), recognizing faces (agnosia for faces), or printed words (alexia). The brain self-monitors, and mismatch between anticipated and actual outcome obtrudes into awareness.38 In contrast, awareness is absent or incomplete when a right-posterior lesion causes "neglect" of the left side of space and person, leading to unawareness of blindness, paralysis, or anesthesia on the left.39

Deficits without awareness are distortions of attention. It is not that there is a gap or lacuna in attention; rather, attention is occupied elsewhere. In left neglect, attention and intention are biased rightward. When one attends selectively, one is largely oblivious of what remains unattended. This is even more pronounced with people who have attentional neuropathology. One can only be aware of how constricted one's attention is if one can internally represent (imagine) the possibility of attending some other way. This is done by initiating, but then aborting, such an attentional shift. If the brain substrate for such imaging is nonfunctional, one cannot experience the fact that one's attention is curtailed.

Remarkably, neglect patients' attentional capability can temporarily be normalized by the simple maneuver of stimulating the vestibular (balance) system by irrigating the opposite ear canal with lukewarm water.40 Stimulating the vestibular system with water below body temperature directs ascending activation to the injured opposite hemisphere. It becomes clear that there had been an activation imbalance between hemispheres such that the uninjured hemisphere preempted attention, targeting it according to its natural bent to the opposite side of space and of the body. When the imbalance is corrected, neglect is no longer apparent. In this normalized state the neglect patient is not aware that his attention had previously been restricted. When the normalizing effect wears off, the patient again does not remember that his attention was different moments earlier.41

Unilateral neglect teaches us that awareness is a property of the ongoing pattern of forebrain activation, and its contents are those that the activated cell assemblies represent. I have called this controlling pattern of activation of the global network the "dominant focus."42 Attending to a side is controlled by the reciprocal (negative feedback) interaction of a right-sided facility that directs selective orienting toward the left extreme of the viewed display, and a corresponding left-sided facility that directs it rightward.43 If the right-sided facility is inactivated, attention that is now exclusively controlled by the disinhibited left hemisphere swings rightward-in any modality, in action and anticipation as well as in perception, and also in mental imagery. The patient is not aware that anything has gone wrong. He can no more think about attending to the left than do it (or imagine doing it).

I have referred to the distributed functioning outlined above as the "uncentered brain."44 In the uncentered brain there is no location at which a lesion can abolish consciousness and yet leave intact the processes that compute what we are conscious of. When consciousness lapses, cognitive processes lapse. No lesion strips awareness from the functioning brain like a layer off an onion. Consciousness must be inherent in the working machinery; it cannot be its product.

Disorders of awareness restrict the contents of consciousness. The lesions that cause them are extensive; they distort and curtail the flexibility of the global network. But even limited lesions influence the functioning of the network as a whole.

LESION EFFECTS AT A DISTANCE

"Pure" or "specific" deficits are something of an abstraction. Lesions exert remote effects on destinations to which their sites are heavily connected. The downstream area can be deprived of excitation or released from inhibition. Moreover, any lesion can interfere with cognition in general. Patients perform less well than matched control subjects even on tasks that are quite unrelated to their specific deficits (as complex as problem solving or as simple as reacting to a flash or a click). Selective impairments are disproportionate rather than pure, and any impairment of the network handicaps the network as a whole. Modularity falls short.

When one hemisphere suffers major injury by stroke, the other hemisphere has been found to have become hypoactivated.45 While ICA is in effect, not only the anesthetized hemisphere but to a lesser extent the opposite hemisphere also is underactivated, as judged by EEG slow waves and reduced blood flow tracked by a radioactive tracer (both of which indicate decreases in metabolic activity).46 Even the opposite cerebellum is hypoactive during ICA. Each hemisphere must normally be activating the other side of the brain at a distance.

Striking distance effects are revealed by the dramatic and well publicized consequences of surgical section of the corpus callosum, the split-brain state, in which the two hemispheres are partly uncoupled. When stimulation is limited to one of the disconnected hemispheres and withheld from the other, the unstimulated hemisphere is typically unable to respond meaningfully (though there are exceptions). It is customary to invoke interrupted interhemispheric flow of information to explain hemispheric independence.47 But much, if not all, split-brain phenomenology can alternatively be attributed to the loss of transcallosal cross-activation.48 In the intact brain, the activated working hemisphere coactivates its partner, safeguarding its readiness to respond. Callosal section disrupts cross-activation, letting the unstimulated hemisphere lapse into sluggish unresponsiveness. In the intact brain, when all or much of one hemisphere is underactivated, this leads to some underactivation also of the other. This is not because both necessarily collaborate in tasks, but because when one is active the other must be ready also to be called upon at any time. The underactivation at a distance may not manifest itself behaviorally as a cognitive deficit but perhaps rather as a sluggishness on the part of the hemisphere to assume control when the need to do so comes its way.

Distance effects help the differentiated network maintain its precarious functional unity. Unity is not inherent in the network. It is secured by the neural architecture and dissolves in certain pathologies and under certain conditions, even in normals.49

AFFECT CONTROLS REASON

As was exemplified by unilateral neglect, when a specialized network is lesioned it may relinquish its role to another (opponent) network that propels behavior in the opposite direction. The symptoms that result are of two kinds. The first is those that derive from the loss of the role of the damaged or underactivated system; these are deficiencies, referred to as negative symptoms. The second is those that derive from the overactivity of the opponent system, which has been released from inhibition by the damaged network. It contributes positive symptoms. For instance, in neglect, unawareness of the left is a negative manifestation; excessive orienting to the right is a positive symptom. Many such opponent couplings are represented in the central nervous system. They underlie contrary behaviors that cannot be combined, between which the individual must choose. Sometimes the choice is automatic: turn left, to the door, and not right, into the wall. But often it requires cost accounting to determine the prospective benefits and costs of a given behavioral choice. This is where affect comes in.

Motivation drives behavior; absent motivation, the organism, however well equipped, remains inert. Positive emotions, such as elation, satisfaction, and gratification, earmark a particular action sequence that had a successful outcome as worth repeating in the future. Conversely, negative emotions attend mismatch between a goal and the actual outcome; this failure, highlighted subjectively by fear, revulsion, disgust, and other disagreeable feelings, calls for a halt in the action plan and a reappraisal of the situation. Attention broadens, additional factors are taken into consideration, and ideally a new improved strategy is devised. The action plan, narrowing towards the goal, is largely driven by the left hemisphere. The "suspend operation" instruction, accompanied by the surge of negatively charged arousal, is more right hemispheric in origin.50

Faulty monitoring of the outcomes of action plans may result in psychopathology. There might be an imbalance between the network that registers "match" and the network that registers "mismatch." A tendency to register spurious "match" outcomes would appear as a denial of deficit. Spurious "mismatch" results would send the individual back over and over again to repeat obsessively activities that others would consider completed. Failure to plan ahead (in left prefrontal dysfunction) would lead to an oppressive frequency of failure and interrupt experiences, engendering dysthymic moods and a sense of helplessness. A disinhibited arousal response to mismatch would present as mood swings or "hysterical overreaction." Conversely, too rigorous a correction of such arousal (inhibition of right posterior by right prefrontal cortex) would produce the emotional impoverishment and lack of capacity to enjoy (anhedonia) of major depression (melancholia). Cyclical swings of activational predominance of the left and right hemisphere (i.e., of the continue and interrupt function, respectively) could be related to the alternation of mania and depression in bipolar disease. In each case, a strategically located cortical injury could mimic a psychiatric disorder. It does so not by creating qualitatively abnormal behaviors but by causing one type of behavior to be used to excess, and its opposite not enough. In turn, the psychiatric disorder (whatever its cause) plays out in part by creating imbalance between cortical processing units, as described above. This imbalance is thought to result from imbalances between neurotransmitter systems.

The fact that a neural system is in place does not guarantee that it will participate in the control of behavior; it needs to be switched on, or activated. Powerful ascending activating systems hold the cortical units ready to respond. Different activating systems are powered by different neurotransmitters.51 Either a deficiency in the action of a few neurotransmitters, or even a single one, or an excess in the action of another is held responsible or suspected for several psychopathologies. These chemical dysfunctions differ from structural lesions in that they can potentially be corrected with psychoactive drugs that either assist or impede the efficacy of particular neurotransmitters.

The overall lesson is that reason is the instrument of affect, not its master. Well-tuned affect enables reason to be deployed to best advantage. Both involve the highest cortical level of functioning. Poorly tuned affect compromises reason, and malfunctioning reason is reflected in deviant affect.

SCULPTING COGNITIONS

The notion that the brain is a switchboard for signals (or symbols), parts of which can be disassembled without consequence for the remaining net, is outdated, though persistent. The global net is depleted of resources by injury to one of its parts. It reorganizes to the extent possible to replace, or compensate for, the impaired capability. But now it operates with a restricted computational space, and this becomes apparent when the patient is asked to do something novel or difficult. In short, the injured brain operates on a skimpier neuronal and cognitive base, with fewer differentiated activity patterns available. In the absence of the specialized generator, another area may take over its role. The right hemisphere assumes language function when it is relinquished by the damaged left. The brain reacts as a whole to local impairment, demonstrating the connectedness of the differentiated neural net. The nonlinear dynamical model of the brain (discussed by Andy Clark in this issue) conceived in terms of coexisting attractor states-or better, in view of the transient nature and restless onrush of mental states, a trajectory through state space-seems to be consistent with evidence from lesion studies. The pattern of activation is global; its details are local.

During every conscious moment, the whole forebrain is a landscape (brainscape?) of peaks and valleys of activation (an energy topology). This pattern is as much the creation of the cell assemblies that are inhibited, and thus do not fire, as of those that contribute activation maxima. It updates Sherrington's fancy of the brain as a magical loom across which lights incessantly twinkle on and off.

The poet Stanley Kunitz remarked accurately, "We are not souls, but systems, and we move in clouds of our unknowing." We are beginning to know a little about ourselves as systems, and we are as interesting as Dyson led us to expect.

ENDNOTES

Douglas Derryberry and Don M. Tucker, "The Adaptive Base of the Neural Hierarchy: Elementary Motivational Controls on Network Function," in P. Rakic and W. Singer, eds., Neurobiology of Neocortex (New York: Wiley, 1990).

2Marcel Kinsbourne and Elizabeth K. Warrington, "A Disorder of Simultaneous Form Perception," Brain 8S (1962): 461-468.

3Janet Jankoviak et al., "Preserved Visual Imagery and Categorization in a Case of Associative Visual Agnosia," Journal of Cognitive Neuroscience 4: 119131. Martha J. Farah, Visual Agnosia (Cambridge, Mass.: MIT Press, 1990). 4Marcel Kinsbourne and Elizabeth K. Warrington, "A Case Showing Selectively Impaired Oral Spelling," Journal of Neurology, Neurosurgery and Psychiatry 29 (1965): 219-223; Marcel Kinsbourne and David Rosenfield, "Agraphia Selective for Written Spelling: An Experimental Case Study," Brain and Language 1 (1974): 215-225.

SMarcel Kinsbourne, "Behavioral Analysis of the Repetition Deficit in Conduction Aphasia," Neurology 22 (1972): 1126-1132; Morris Freedman, Michael P. Alexander, and Margaret A. Naeser, "Anatomic Basis of Transcortical Motor Aphasia," Neurology 34 (1984): 409-417.

6Marcel Kinsbourne and Bear Hiltbrunner, "A Selective Deficit in Cursive Writing" (unpublished).

Claudio Luzzatti, R. Rumiati, and G. Ghirardi, "Visuo-verbal Disconnection and its Anatomical Constraints in Optic Aphasia," Brain and Cognition 32 (1996): 199-202.

8 mARCELKinsbourne and Elizabeth K. Warrington, "Observations on Colour Agnosia," Journal of Neurology, Neurosurgery and Psychiatry 27 (1964): 296-299.

9Marcel Kinsbourne, "Hemispheric Specialization and the Growth of Human Understanding," American Psychologist 37 (1982): 411-420. "Lynn C. Robertson and Dean Delis, "Part-whole Processing in Unilateral Brain-damaged Patients: Dysfunction of Hierarchical Organization," Journal of Neurosciences 8 (1986): 3735-3769.

"Elizabeth K. Warrington, Merle James, and Marcel Kinsbourne, "Drawing Disability in Relation to Laterality of Cerebral Lesion," Brain 89 (1966): 53-82.

'2Vadim Deglin and Marcel Kinsbourne, "Divergent Thinking Styles of the Hemispheres: How Syllogisms are Solved during Transitory Hemisphere Suppression," Brain and Cognition 31 (1996): 285-307. Gail L. Risse, J. R. Gates, and M. C. Fangman, "A Reconsideration of Bilateral Language Representation based on the Intracarotid Amobarbital Procedure," Brain and Cognition 33 (1997): 118-132. "Marcel Kinsbourne, "The Minor Cerebral Hemisphere as a Source of Aphasic Speech," Archives of Neurology 25 (1971): 302-306; Christopher Code, Language, Aphasia and the Right Hemisphere (Chichester, England: Wiley, 1987). Marcel Kinsbourne, "The Right Hemisphere and Recovery from Aphasia," in Brigitte Stemmer and Harry A. Whitaker, eds., Handbook of Neurolinguistics (New York: Academic Press, 1997).

'5Heiko Braak, "On Magnopyramidal Temporal Fields in the Human BrainProbable Morphological Counterparts of Wernicke's Sensory Speech Region," Anatomy and Embryology 152 (1978): 141-169. Max S. Cynader et al., "General Principles of Cortical Organization," in Pasko Rakic and Wolf Singer, eds., Neurobiology of the Neocortex (New York: Wiley, 1988); Deepak N. Pandya, Benjamin Seltzer, and Helen Barbas, "Input-Output Organization of the Primate Cerebral Cortex," in Horst D. Steklis and J. Erwin, eds., Comparative Primate Biology (New York: Liss, 1988).

"Antonio R. Damasio, Descartes' Error (New York: Putnam, 1994). 'Karl H. Pribram, Languages of the Brain (New York: Prentice-Hall, 1971); P. W. Burgess and Tim Shallice, "Response Suppression, Initiation and Strategy Use following Frontal Lobe Lesions," Neuropsychologia 34 (1996): 263-273. 19 Damasio Descartes' Error.

2Deepak Pandya and C. L. Barnes, "Architecture and Connections of the Frontal Lobe," in Ellen Perecman, ed., The Frontal Lobes Revisited (New York: IRBN, 1987); Timothy W. Robbins, "Dissociating Executive Functions of the Prefrontal Cortex," Philosophical Transactions of the Royal Society of London B351 (1996): 1463-1472; Richard E. Passingham, "Attention to Action," Philosophical Transactions of the Royal Society of London B351 (1996): 1473-1479.

2"Marcel Kinsbourne and Elizabeth K. Warrington, "Jargon Aphasia," Neuropsychologia 1 (1963): 27-37.

22Marcel Kinsbourne and Robert E. Hicks, "Functional Cerebral Space: A Model for Overflow, Transfer and Interference Effects in Human Perfor

mance: A Tutorial Review," in Jean Requin, ed., Attention and Performance VII (Hillsdale, N.J.: Erlbaum, 1978).

23Marcel Kinsbourne, "Attentional Dysfunctions and the Elderly: Theoretical Models and Research Perspectives," in Leonard W. Poon et al., eds., New Directions in Memory and Aging (Hillsdale, N.J.: Erlbaum, 1980). 24Gordon G. Globus and Joseph P. Arpaia, "Psychiatry and the New Dynamics," Biological Psychiatry 35 (1994): 352-264; James P. Sutton and James A. Anderson, "Computational and Neurobiological Features of a Network of Networks," Computation and Neural Systems (forthcoming); Cynader et al., "General Principles of Cortical Organization"; Pandya et al., "Input-Output Organization of the Primate Cerebral Cortex."

25 Cynaderet al., "General Principles of Cortical Organization"; Pandya et al., "Input-Output Organization of the Primate Cerebral Cortex." 26Jason W. Brown, The Life of the Mind: Selected Papers (Hillsdale, N.J.: Erlbaum, 1988); Brian Kolb, "Brain Development, Plasticity and Behavior," American Psychologist 9 (1989): 1203-1212.

2'John C. Marshal and Freda Newcombe, "Patterns of Paralexia: A Psycho

linguistic Approach," Journal of Psycholinguistic Research 2 (1973): 175-199. 28 MarcelKinsbourne and Elizabeth K. Warrington, "Disorders of Spelling," Journal of Neurology, Neurosurgery and Psychiatry 27 (1964): 224-228.

29Marcel Kinsbourne, "Awareness of One's Own Body: A Neuropsychological Hypothesis," in Jose Luis Bermudez, Anthony J. Marcel, and Naomi Eilan, eds., The Body and the Self (Cambridge, Mass.: MIT Press, 1995). 30 FrancisCrick, "Function of the Thalamic Reticular Complex: The Searchlight Hypothesis," Proceedings of the National Academy of Sciences 81 (1984): 4586-4590; Wolf Singer, "Synchronization of Cortical Activity and its Putative Role in Information Processing and Learning," Annual Review of Physiology SS (1993): 349-374.

"Anne M. Treisman and G. Gelade, "A Feature Integration Theory of Attention," Cognitive Psychology 12 (1980): 97-136.

32Semir Zeki and S. Shipp, "The Functional Logic of Cortical Connections," Nature 335 (1988): 311-317.

33Cynader et al., "General Principles of Cortical Organization"; Pandya et al.,

"Input-Output Organization of the Primate Cerebral Cortex." 34Brown, The Life of the Mind; Kolb, "Brain Development, Plasticity and Behavior."

Norman Geschwind, "Disconnexion Syndromes in Animals and Man," Brain 88 (1965): 237-294, 585-644.

36Carl Wernicke, Der Aphasische Symptomenkomplex (Breslau, Germany: Cohn and Weigert, 1874).

37Marcel Kinsbourne, "Integrated Cortical Field Model of Consciousness," in Anthony J. Marcel and Eduardo Bisiach, eds., The Concept of Consciousness in Contemporary Science (London: Oxford University Press, 1988).

3"Eran Zaidel, "Hemispheric Monitoring," in D. Otteson, ed., Duality and Unity of the Brain (London: McMillan, 1987).

39Marcel Kinsbourne, "Hemineglect and Hemisphere Rivalry," in Edwin A. Weinstein and Robert P. Friedland, eds., Hemi-inattention and Hemisphere Specialization: Advances in Neurology (New York: Raven, 1977). 40J. Silberpfennig, "Contributions to the Problem of Eye Movements: III, Disturbance of Ocular Movements with Pseuohemianopsia in Frontal Tumors," Confinia Neurologica 4 (1949): 1-13; Alan B. Rubens, "Caloric Stimulation and Unilateral Neglect," Neurology 35 (1985): 1019-1024. 4'Vilayanur S. Ramachandran, "Anosognosia in Parietal Lobe Syndrome," Consciousness and Cognition 4 (1995): 22-51.

42Kinsbourne, "Integrated Cortical Field Model of Consciousness." 43Marcel Kinsbourne, "Lateral Interactions in the Brain" and "Mechanism of Hemispheric Interaction in Man," in Marcel Kinsbourne and W. Lynn Smith, eds., Hemispheric Disconnection and Cerebral Function (Springfield, Ill.: Thomas, 1974).

'"Marcel Kinsbourne, "Models of Consciousness: Serial or Parallel in the Brain?" in Michael S. Gazzaniga, ed., The Cognitive Neurosciences (Cambridge, Mass.: MIT Press, 1995).

45R. J. Andrews, "Transhemispheric Diaschisis: A Review and Comment," Stroke 22 (1991): 943-949.

4"David W. Loring et al., Amobarbital Effects and Lateralized Brain Function: The Wada Test (New York: Springer, 1992); D. McMakin et al., "Assessment of the Functional Effect of the Intracarotid Sodium Amytal Procedure Using Co-registered MRI/HMPAO-SPECT and SEEG," Brain and Cognition 33 (1997): 50-70.

47Eran Zaidel, Jeffrey M. Clarke, and B. Suyenobu, "Hemispheric Independence: A Paradigm Case for Cognitive Neuroscience," in Arnold B. Scheibel and Adam F. Wechsler, eds., Neurobiology of Higher Cognitive Function (New York: Guilford Press, 1990).

48 Kinsbourne "Lateral Interactions in the Brain"; Kinsbourne, "Mechanism of Hemispheric Interaction in Man"; Marcel Kinsbourne, "The Corpus Callosum as a Component of a Circuit for Selection," in Eran Zaidel, Marco Iacoboni, and A. Pascual-Leone, eds., The Corpus Callosum in Sensory Motor Integration: Individual Differences and Clinical Applications (New York: Plenum, 1998); Yves Guiard, "Cerebral Hemispheres and Selective Attention," Acta Psychologica 46 (1978): 41-61.

49Anthony Marcel, "Slippage in the Unity of Consciousness," in Experimental and Theoretical Studies of Consciousness (Ciba Foundation Symposium 174) (New York: Wiley, 1993), 168-186.

50 Marcel Kinsbourne, "A Model of Adaptive Behavior Related to Cerebral Participation in Emotional Control," in Guido Gainotti and C. Caltagirone, eds., Emotions and the Dual Brain (New York: Springer, 1989).

51 Timothy W. Robbins and Barry J. Everitt, "Arousal Systems and Attention," in Gazzaniga, ed., The Cognitive Neurosciences.

Marcel Kinsbourne is Professor of Psychology at New School University.

Copyright American Academy of Arts and Sciences Spring 1998
Provided by ProQuest Information and Learning Company. All rights Reserved