Ivane S. Beritashvili (1884-1974): from spinal cord reflexes to image-driven behavior.

Ivane Beritashvili ("Beritoff" in Russian, and often in Western languages) was a major figure in 20th-century neuroscience. Mastering the string galvanometer, he founded the electrophysiology of spinal cord reflexes, showing that inhibition is a distinctly different process from excitation, contrary to the concepts of his famous mentor, Wedensky. Work on postural reflexes with Magnus was cut short by World War I, but he later demonstrated that navigation in two-dimensional space without vision is a function solely of the vestibular system rather than of muscle proprioception. Persevering in his experiments despite postwar turmoil he founded an enduring Physiology Institute in Tbilisi, where he pursued an ingenious and extensive investigation of comparative memory in vertebrates. This revealed the unique nature of mammalian memory processes, which he forthrightly called "image driven," and distinguished them unequivocally from those underlying conditional reflexes. For some 30 years the Stalinist terror confined his publications to the Russian language. Work with his colleague, Chichinadze, discovering that memory confined to one cerebral hemisphere could be accessed by the other via a specific forebrain commissure, did reach the West, and ultimately led to recognition of the fascinating "split brain" condition. In the 1950s he was removed from his professorial position for 5 years as being "anti-Pavlovian." Restored to favor, he was honorary president of the "Moscow Colloquium" that saw the foundation of the International Brain Research Organization.

Hemispheric interaction, metacontrol, and mnemonic processing in split-brain macaques.

These experiments explored the interactions remaining between the cerebral hemispheres in two split-brain macaques. The 'split' was earlier confirmed by showing that one hemisphere was incapable of identifying visual images seen by the other. The critical tests for residual interactions were intermingled with control trials in a continuous recognition task. These tests were of two kinds: 'parallel processing', to determine how simultaneous viewing by both hemispheres affected subsequent recognition by one of them alone; and 'conflict', where opposite responses were demanded from the two hemispheres, thus assessing the issue of metacontrol. Two types of stimuli were also employed: ART, in which each hemisphere saw essentially the same image; and BIPARTITE, in which images were entirely different for each hemisphere. Since, with either type of stimulus, performance was best when viewed by both hemispheres at both encoding and retrieval, 'parallel processing' was highly efficient. However, when both hemispheres viewed initially and only one was subsequently queried, performance was significantly worse than when each hemisphere acted alone on each occasion. It is thus reasoned that when both hemisphere view together, the resultant memory trace somehow reflects the bilaterality, a conclusion concordant with observations of Marcel on blindsight. Processing different images (BIPARTITE) was somewhat more disruptive in this regard than if the same image was viewed by each hemisphere. This was particularly true in the conflict situation, where for one hemisphere the item seen was NEW and for the other it was OLD. A response of 'OLD' was, at first, consistently rewarded. When this well-established protocol was changed, the hemispheres in each animal were gradually able to revise their joint behavior. This, together with the effect of disparate images, and the deficiency evoked when the animals were forced to recognize unilaterally an image first viewed under bilateral conditions, all manifest considerable, and complex, interaction between the hemispheres despite absence of the forebrain commissures. The superior colliculus seems a likely focal point for such interhemispheric effects.

Temporal cost of switching between kinds of visual stimuli in a memory task.

These experiments measured the extra time required to respond when the type of stimulus was changed from one trial to the next. Heretofore, switching costs have been measured for a change in task, but we wished to isolate the cost of changing the sensorial component per se and its necessary analytical processing, as distinct from changing the question being posed or the type of response to be given. Thus, the task was identical throughout the experiments: continuous recognition in a single, 240-trial session, in which the subject was required to distinguish initial from repeat appearances of a stimulus, the single repetition of each stimulus occurring after 1-31 intervening trials. There were two categories of 200-ms stimuli, linguistic (words and non-words) and images (multiple-colored or gray scale panels, human faces, or butterflies); and two conditions of switching, predictable (alternating on each trial) or unpredictable, in which the switch occurred after three to eight trials of one kind. In the majority of cases, there was a robust switching cost, from 24 to 92 ms. The similarity of costs in the predictable and unpredictable modes suggests that this cost is derived at least as much from terminating the modus operandi for the preceding type of stimulus as from a reconfiguration of processing for the new type of stimulus. In the switch between words and images, the costs were "paradoxical" (asymmetrical), in that the switching from an image to a word was more costly than the reverse, that is, changing from the more to the less difficult required the greater time. This, too, is compatible with the idea that termination of the previous mode of processing is a major component of the cost. Thus, in contemplating the neuronal/cognitive events underlying visual memory, consideration must be given to the inertia of pre-existing linkages.

Long-term reversal of hemispheric specialization for visual memory in a split-brain macaque.

Accuracy of response and pattern of ocular fixations in three split-brain macaques were used to evaluate performance of each hemisphere in a continuous visual recognition task. The animal indicated by ocular fixation upon response points whether a displayed image or face was 'NEW' or 'OLD'. An inadvertent lesion of cingulate gyrus severely reduced contralateral fixations and impaired performance of the affected hemisphere in one animal, confirming the inferred relation between hemisphere and laterality of fixations. The hemispheres in the other two animals were initially remarkably similar in accuracy with human faces and with images; but the right hemisphere was significantly superior to the left for macaque faces. Parallel to this, in the one animal tested while simultaneously using both eyes/hemispheres, fixations were made primarily on the left half of human and macaque faces (right hemispheric control), whereas for images the ocular fixations were predominantly focused on the right half. However, after further, extensive training the left hemisphere performed with significantly greater accuracy than the right on all material and this shift was accompanied and further corroborated by a reversal of the fixational pattern to favor the right half of faces, as continued to be the case with images. Thus, over the long term both the pattern of ocular fixations and the accuracy of performance demonstrate a migration from right to left hemispheric dominance as familiarity with the task increased. Performance of the initially superior hemisphere actually diminished with this shift, presenting a uniquely puzzling question of hemispheric balance in the absence of the forebrain commissures.

The five mysteries of the mind, and their consequences.

While Western man has recognized for almost 2500 years that mind derives exclusively from brain, clothing this fact with explanatory detail still proves elusive. First, is consciousness per se, created by processes demonstrably limited to certain, but still unspecified, neuronal arrangements and activities. Then there is perception, its ineffable qualia, and the fact that it arises from neuronal activity widely dispersed in space and time within networks of vast complexity. Voluntary control is equally dispersed as to neuronal participation, and nescient as to origin. An often overlooked mystery is the unity of mind and behavior that prevails despite the potential for bihemispheric duplication of processes and experience. Finally, there is memory, which while credibly within grasp of understanding as a synaptic alteration maintained via activation of the nuclear genome, still wholly defies comprehension when viewed as commanded recall of myriad, randomly selectable details of the past, a largely effortless and 'instantaneous' flood of memories. For two centuries science has endeavored to demonstrate how these mysteries proceed from physics and chemistry, as indeed they do; but viewed from this direction alone, mind is but the babbling of a robot, chained ineluctably to crude causality. In a bold and revolutionary stroke, Roger Sperry has conceived a more credible paradigm, that the totality of neuronal action, as a richly intercommunicating system, gives rise to effects transcendent to the individual physicochemical elements that compose it. A major achievement of this position is that it is immediately consonant with everyday human experience and belief. While neither Sperry's vision. nor the reduction of the mysteries to a dance of ions can yet be proven, the vast advantage of Sperry's thesis is that it again imbues human thought and action with responsibility, and opens morality to the light of science, while the long wait for certainty unfolds.

Commonality of processes underlying visual and verbal recognition memory.

Memory for visual and verbal material engages widely distributed systems, to a large degree focussed in different hemispheres. It might thus be expected that these disparate neuronal populations should display significantly different characteristics in regard to mnemonic performance. Visual memory, fundamental to all human beings, and whose characteristics are largely shared with macaques, was assayed using unique non-objective, colored images lacking ready verbal description and was contrasted with memory for four-letter non-offensive English words. The effects of memory loading, stimulus duration and long-term test intervals (1-2 weeks) were studied in regard to accuracy and reaction times for recognizing initial versus re-exposure to these two types of items. No effects of memory loading were apparent despite the incrementing memory load in the 240-item, running recognition sessions. Words were better remembered than images, both in the long and short term, but the detailed characteristics of reaction times and accuracy in relation to number of intervening items, and in long-term memory were strikingly similar. Given the wide and well-established disparity in cerebral loci participating in linguistic versus image analysis, these multiple similarities in the pattern of mnemonic performance indicate that the underlying neuronal processes must be comparable for remembering either images or words. Furthermore, the strong link manifested between individual items across a varying number of intervening intervals and added items suggests that a phenomenon highly similar to the "stimulus specific adaptation" (SSA), displayed by units in macaque inferotemporal cortex, occurs for each item to be recognized. Finally, the significant augmentation in accuracy both in short- and long-term memory for images when viewing time permits saccades is explained if each saccade and fixational pause recruits additional neurons into the pool displaying SSA, or its equivalent, for the item being viewed.

Habituation of ocular following reflex requires corpus callosum for interhemispheric transfer.

In cats, unanesthetized following transection of the brainstem at a level precluding painful sensation, and limiting ocular motility to a vertically oriented course (the pretrigeminal preparation), habituation of the orienting reflex, consisting of ocular fixation and smooth pursuit, readily transferred between moving visual stimuli directed first at one and then the other cerebral hemisphere. Under the same conditions, when the corpus callosum had been transected 2 weeks prior to the habituation, interhemispheric transfer was absent. Thus, despite substantial brainstem involvement and bilateral coordination of ocular motility the neocortex plays an essential role in this habituation, just as it does in the interhemispheric transfer of visual discrimination learning. This suggests that habituation is a fundamental form of learning in the mammalian forebrain.

Interhemispheric sharing of visual memory in macaques.

(1) In macaques with the optic chiasm transected, and forebrain commissural communication limited to the anterior commissure or the posterior 5 mm of the splenium of the corpus callosum, visual patterns viewed initially by only one eye (hemisphere) are subsequently recognized by the other with normal accuracy. (2) The efficiency of these commissural paths is further indicated by the fact that even when as many as six "target" images are presented for memorization to only one hemisphere, it makes essentially no difference as to accuracy or latency of performance which hemisphere is then required to distinguish "target" from "non-target" images. (3) By electrically tetanizing structures in one or the other temporal lobe at various times in relation to visual input and/or mnemonic testing it could be shown: (a) that a memory trace restricted in its formation to a single hemisphere was available to the other via either forebrain commissure, and (b) that the memory is formed bilaterally despite unilateral input. (4) When the chiasm is split but the commissures are intact, simultaneous presentation of disparate images to each hemisphere severely perturbs performance, suggesting that the callosal system operates continuously to unify visual percepts; but when only the anterior commissure is intact, the two hemispheres accept incongruent images without perturbation. (5) In the fully "split-brain" condition, when one hemisphere cannot access memories held in the other, the accuracy of performance by each hemisphere is nevertheless burdened by the memory load of its neocortically disconnected partner. It can thus be inferred that the brainstem plays a critical, unifying role in this mnemonic process.

Time is of the essence: a conjecture that hemispheric specialization arises from interhemispheric conduction delay.

Tomasch (1954) and Aboitiz et al. (1992) found the majority of the fibers of the human corpus callosum are under 1 micron in diameter. Electron microscopic studies of Swadlow et al. (1980) and the detailed study of LaMantia and Rakic (1990a) on macaques show the average size of the myelinated callosal axons also to be less than 1 micron. In man, the average-sized myelinated fiber interconnecting the temporal lobes would have a one-way, interhemispheric delay of over 25 msec. Thus, finely detailed, time-critical neuronal computations (i.e., tasks that strain the capacity of the callosum and hence could not be handled by just the larger fibers) would be performed more quickly via shorter and faster intrahemispheric circuits. While one transit across the commissural system might yield tolerable delays, multiple passes as in a system involving "setting" would seem prohibitively slow. We suggest that these temporal limits will be avoided if the neural apparatus necessary to perform each high-resolution, time-critical task is gathered in one hemisphere. If the, presumably overlapping, neural assemblies needed to handle overlapping tasks are clustered together, this would lead to hemispheric specialization. The prediction follows that the large brains of mammals such as elephants and cetaceans will also manifest a high degree of hemispheric specialization.

Role of the forebrain commissures in bihemispheric mnemonic integration in macaques.

A serial probe recognition task was used to examine the interhemispheric exchange of visual data in macaques. Each block of trials began with the memorization of one to six visual target images. The monkeys then had to determine, in tests that followed immediately, whether probe images were or were not members of the learned target set. Previous work with both humans and macaques has shown that the time required for the evaluation of probes generally increases, while response accuracy decreases, as a function of the number of targets, the "memory load". By testing animals with bisected optic chiasm, it was possible to direct visual information to only one hemisphere at a time, simply by occluding the opposite eye. In this fashion, the quality of intrahemispheric evaluations (in which a monocular probe was a match for a target previously viewed through the same eye) was compared with that of interhemispheric evaluations (in which a probe was a match for a target previously designated through the opposite eye). A key question was whether division of the target list between the hemispheres modified the relationships between reaction time, response accuracy, and memory load. Provided that either the anterior commissure or the splenium of the corpus callosum was intact, interhemispheric processing was only subtly less efficient than intrahemispheric processing. The ability to perform interhemispheric evaluations was selectively and completely disrupted if all forebrain commissural fibers were transected. In this latter split-brain condition, the time required for probe evaluations was, as expected, determined solely by the number of target items memorized by the probed hemisphere. Accuracy, however, was always a function of the total memory load, regardless of the distribution of targets between the hemispheres. This implies, first, that accuracy and latency do not reflect identical mnemonic factors, as frequently held, and second, that in mnemonic processing, the two hemispheres draw upon a unified, shared resource, probably allocated by the intact brainstem.

Bi-versus monohemispheric performance in split-brain and partially split-brain macaques.

Experiments comparing binocular with monocular abilities of monkeys working on visual mnemonic tasks were performed. First, it was shown that even in split-brain monkeys performance was more accurate when both hemispheres were utilized than when the task was performed with only the single (better) hemisphere. Some form of noncommissural integration is thus possible. However, when the forebrain commissures are present, as in four other animals (with only optic chiasm transected) it was shown that integration occurs via callosal mechanisms as well. This was demonstrated by the fact that here, too, binocular performance was normally more accurate than monocular performance, but when different images to be remembered were presented concurrently to the two eyes, the binocular advantage was lost. Finally, in three monkeys with only the anterior commissure allowing interhemispheric communication the superiority of binocular assessment remained even when the two hemispheres simultaneously received such differing images.

Schizophrenia: a disease of interhemispheric processes at forebrain and brainstem levels?

The evidence is convincing that each human cerebral hemisphere is capable of human mental activity. This being so, every normal human thought and action demands either a consensus between the two hemispheres, or a dominance of one over the other, in any event integrated into a unity of conscious mentation. How this is achieved remains wholly mysterious, but anatomical and behavioral data suggest that the two hemispheres, and their respective bilateral, anatomical-functional components, maintain a dynamic equilibrium through neural competition. While the forebrain commissures must contribute substantially to this competitive process, it is emphasized in this review that the serotonergic raphé nuclei of pons and mesencephalon are also participants in interhemispheric events. Each side of the raphé projects heavily to both sides of the forebrain, and each is in receipt of bilateral input from the forebrain and the habenulo-interpeduncular system. A multifarious loop thus exists between the two hemispheres, comprised of both forebrain commissural and brainstem paths. There are many reasons for believing that perturbation of this loop, by a variety of pathogenic agents or processes, probably including severe mental stress in susceptible individuals, underlies the extraordinarily diverse symptomatology of schizophrenia. Abnormality of features reflecting interhemispheric processes is common in schizophrenic patients; and the 'first rank' symptoms of delusions or hallucinations are prototypical of what might be expected were the two hemispheres unable to integrate their potentially independent thoughts. Furthermore, additional evidence suggests that the disorder lies within, or is focused primarily through, the raphé serotonergic system, that plays such a fundamental role in consciousness, in dreaming, in response to psychotomimetic drugs, and probably in movement, and even the trophic state of the neocortex. This system is also well situated to control the dopaminergic neurons of the ventral tegmental area, thus relating to the prominence of dopaminergic features in schizophrenia; and the lipofuscin loading and intimate relation with blood vessels and ependyma may make neurons of the raphé uniquely vulnerable to deleterious agents.

Forebrain commissures and visual memory: a new approach.

The primary purpose of these exploratory experiments was to determine: (1) whether the forebrain commissures can provide full accessibility of the mnemonic store to either hemisphere when the taks involves memory for 'events' (images) rather than, as in essentially all previous tests on split-brain animals, memory for 'rules' (discrimination habits); and (2) whether the anterior commissure (AC) alone is capable of such function. Macaques, with optic chiasm transected to allow limitation of direct visual input to one or the other hemisphere, were trained on tasks requiring recognition of previously viewed photographic slides. For one task, delayed-matching-to-sample (DMTS), the animal was presented with a 'sample' image, and then 0-15s later was required to choose that image in preference to a second image concurrently displayed. On the other task, running recognition (RR), a series of images was presented, some of which were repetitions of images previously seen in that session, and the animal was required to signal its recognition of these repetitions. For either task the initial presentation could be made to one eye and hemisphere, and subsequent recognition required of the other. In such circumstance, if all forebrain commissures were divided, such interhemispheric recognition was no longer possible. For the DMTS task if either the AC or 5 mm of the splenium of the corpus callosum were available, interhemispheric recognition was basically equivalent to that using the same eye and hemisphere. However, interhemispheric accuracy with the RR task, while well above chance levels, was consistently inferior to that achieved intrahemispherically when complex scenes or objects were viewed. This is probably a consequence mostly of the differing visual fields of the two eyes, since interhemispheric accuracy was greatly improved by use of images having approximately identical right and left halves. No consistent hemispheric specialization nor difference in direction of interhemispheric communication was observed despite the use of different types of material and the different mnemonic tasks. It is concluded that the AC in macaques can achieve full and continuously operative neural unification of the mnemonic traces of past experience.

Morphometric analysis of the human corpus callosum and anterior commissure.

The cross-sectional areas of the corpus callosum (CC) and anterior commissure (AC) were determined by computer-assisted morphometry in normal human brains obtained at autopsy. In addition, the shape of each CC was examined qualitatively by three "blind" observers. A two-fold variation was observed in the cross-sectional area of the CC. Surprisingly, callosal cross-sectional area was not significantly related to brain weight. Moreover, contrary to recent reports, neither simple inspection nor morphometry revealed structural variation related to sex. A striking, seven-fold, variation was observed in the cross-sectional area of the AC. However, AC cross-sectional area was not related either to brain weight or CC cross-sectional area. A trend toward sexual dimorphism in AC cross-sectional area was observed, with males having the larger AC's. Since the interhemispheric commissures are composed, to a large extent, of fibers that link the various cortical areas of the two hemispheres, these observations suggest that variation in the cross-sectional area of the interhemispheric commissures is not simply related to brain weight or sex but, rather, reflects a similar degree of variation in some aspect of cortical structure.

Perceptual "blankout" of monocular homogeneous fields (Ganzfelder) is prevented with binocular viewing.

The loss of visual perception or "blankout" which occurs when a homogeneous field (Ganzfeld) is presented monocularly is prevented when the same field is viewed binocularly. Thus, blankout cannot be retinal; and contours or transients in time and space are unnecessary for the continuous maintenance of visual perception. Experiments are reported in which blankout ensues only if the two eyes receive luminance disparities ca 0.75 log I. Furthermore, blankout is only marginally affected by stimulus intensity, nor is it dependent on stimulus hue. However, equally luminant but disparate hues presented to the two eyes produce perceptions reminiscent of blankout, with the darkness of blankout replaced with that of color. It is hypothesized that the underlying mechanisms have a commonality in the phenomena of blankout and binocular rivalry but several noncongruent features require explanation.

Comparable performance by man and macaque on memory for pictures.

Two macaques, shown a series of pictures, recognized 79% and 85% upon re-presentation after 45 other pictures intervened. Human subjects working with the identical pictures (chosen to avoid human linguistic and experiential connotations) averaged 83% correctly recognized. The human false 'recognition' rates were lower than the macaques', hence the average human accuracy was better, but the range of accuracy among the human subjects overlapped that of the macaques.

A macaque remembers pictures briefly viewed six months earlier.

While there are several studies documenting the enduring nature of memory for acquired discriminatory "habits' or "skills' in animals, comparable data on memory for visual scenes, i.e., "events', are essentially non-existent, and difficult to obtain even in man. An opportunity to assay this question in macaques arose in the early stages of training an animal on a running recognition task. It had previously been trained on trial-unique delayed matching to sample, and its past experience with this visual material was precisely known. When some of these images which had not been seen by the monkey for at least 6 months were intermingled with comparable material during its training on the running recognition task, with a high degree of statistical reliability (P less than 0.005) it distinguished about one-third of the earlier images, many of which had been seen for a total of only 30 s or less. A medical student, who had previously trained the animals and had had more exposure to the material than did this macaque, and certainly had more precise instruction on how to perform, recognized two-thirds of these same images, also after a hiatus of 6 months. It thus appears likely that the permanence of mnemonic storage for briefly encountered scenes is comparable for the central visual systems of macaque and man.

Cryogenic blockade of the visual cortico-thalamic projection in the rat.

A simple procedure is described for rapidly inactivating areas 17 and 18 in the rat by superfusion of the immediately overlying dura mater with chilled, physiological saline solution. Unit recording indicates deep depression of cortical activity within 20 s or less, and equally rapid restoration upon rewarming. Repetition does not appear to be deleterious. During such inactivation of the "visual" cortex (VC) essentially all neurons in the visual portion of the thalamic reticular nucleus (vTRN) are significantly depressed in their background activity and/or in their response to photically or electrically elicited input over the optic tract (Table 1). Activity of neurons in the dorsal portion of the lateral geniculate nucleus (LGNd), on the other hand, is much less likely to be affected, although in occasional neurons the effects should be profound. No evidence of disinhibition was apparent. It is concluded that the vTRN in the rat is highly dependent upon the VC in its activity, and is, thus, likely to be primarily a tool of the VC in the modulation of thalamic events. Extensive work of others shows that the vTRN provides inhibitory input to the LGNd, but in the present experiments the loss of inhibition via this route seems to be balanced by a corresponding loss of general excitatory input when the VC is inactivated.

Nongeniculate afferents to striate cortex in macaques.

Horseradish peroxidase (HRP) was injected in relatively massive amounts to cover most, or portions, of opercular striate cortex in four macaques. Absence of transcallosal or circumventricular labelling, plus discrete and consistent retrograde labelling in other areas in the four cases, assured the validity and specificity of the observations. Numerous labelled cells in regions directly bordering striate cortex, however, were excluded from the analysis because of the possibility of uptake consequent to physical diffusion. With this exception, all labelled cells were counted at roughly 2-mm intervals for one case with extensive unilateral injection of HRP. Even excluding the closely circumstriate population, the totals indicate that more than 30% of the afferent input to striate cortex arises from nongeniculate sources. Four areas of neocortex together make up about one-fourth of the total afferents: superior temporal sulcus 17.1%; inferior occipital area, 6.1%; intraparietal sulcus, 0.4%; and parahippocampal gyrus, 0.3%. Other areas projecting to striate cortex include claustrum, pulvinar, nucleus paracentralis, raphé system, locus coeruleus, and the nucleus basalis of Meynert. Cells of the latter were particularly striking with their very heavy uptake of HRP, and, even in cases of minimal effective injection, were scattered throughout an extensive area from the posterior edge of the globus pallidus passing rostrally beyond the chiasm and into the nucleus of the diagonal band. On the basis of their distribution and known cholinergic affinity, it is argued that this group also includes the cells labelled in and around lateral hypothalamus and cerebral peduncle, and that as a whole the group constitutes a cholinergic counterpart of the diffusely projecting monoaminergic systems. It seems possible that the basalis projection at first follows a fornical-subcallosal pathway to reach striate cortex via callosoperforant fibers.