['Animal Thought' © Stephen Walker 1983]
gif pdf file of this chapter

5   The functional organisation of the vertebrate brain



Concepts of phylogenetic improvement in the organisation of the vertebrate nervous system have been (Bastian, 1880; Romer, 1949) and are (Razran, 1971; Romer and Parsons, 1977) rampant in accounts of comparative neuro-anatomy. This may be because such concepts are accurate reflections of fact, but the reason may have more to do with the visibility of a topographical progression of structures in the plan of the vertebrate brain, which has in the past drawn evolutionary hypotheses irresistibly towards it. The human brain can be said to consist of the hindbrain, the midbrain and the forebrain, and almost every account of brain evolution incorporates to some extent the notion that each part was physically or functionally added at successive stages of phylogenetic development. A pleasing orderliness would prevail if the ancestors of vertebrates made do with only a spinal cord, to which a hindbrain was added for the benefit of the primitive Cyclostomes, needing the assistance of a midbrain for the control of higher fish behaviour; the forebrain arriving in a diminutive and rudimentary aspect for the purpose of life on land, and showing only a partial elaboration in the warm-blooded but instinct-dominated birds, before fulfilling its destiny with the flowering of the cortex in mammals, which has been permitted to enshroud and overwhelm the now-redundant former stages by the rationality of man.

One still has a sense of regret that this charming and convincing tale must be discarded. The weight of evidence is now if anything more in favour of the unhelpful suggestion, bravely canvassed on some occasions (e.g. Heier, 1948), that all of the fundamental parts of the vertebrate brain were present very early on, and can be observed in lampreys. In fact the strictly additive story has always had the disadvantage that all the physical divisions of higher vertebrate brains


appear in the lower classes, and the most popular hypotheses incorporate their presence in the notion of encephalisation of function (Weiskrantz, 1961; Jerison, 1973). The essential doctrine of encephalisation of function is that the superficial layout of all vertebrate brains was present in the original versions, but that there has been an evolutionary trend which moves psychological functions further and further forward in the central nervous system, from the spinal cord through the midbrain to the cerebral hemispheres. Although this has often been accepted without question, it is best to be suspicious of all purported evolutionary trends of this type (Simpson, 1953). The evidence which can be put forward in favour of encephalisation is of two kinds: first, the relative size of different brain structures; and second, more detailed experimental information bearing on their psychological importance. As far as size goes, there is no doubt that the relative size of different parts of the brain varies from species to species, and it is also beyond dispute that the relative size of the cerebral hemispheres is greater in the 'higher' vertebrates, and that the cerebral hemispheres of the human brain are exceptionally large. Jerison (1973) suggests that we should regard these facts as evidence for encephalisation of structure, rather than for encephalisation of function. These are two separate theories, since encephalisation of structure simply means that the forebrain tends to expand its size, while encephalisation of function implies that the reason for the expansion of the forebrain is that it takes over jobs previously done by lower brain divisions. One of the earliest and most explicit accounts of encephalisation of function refers to the 'migration of the dominant controlling centre' (Bastian, 1880). Before going into these ideas any further it would be as well to establish the layout and terminology for these areas of the brain which are to be distinguished (see Figures 3 and 4). Brief inspection of the outer surfaces of the brain frequently permits the identification of the cerebral hemispheres, the optic lobes, and the brainstem, and these boundaries may be confirmed by large-scale dissection. But, as reference to standard works will show, more detailed dissection and microscopic examination reveals a bewildering number of separately identified nuclei and fibre tracts (Gray, H. 1967; Pearson and Pearson, 1976; Kappers et al. 1936). In giving convenient labels to brain structures such as the cerebellum, the hippocampus, or the pituitary gland, it should not be forgotten that the names mask several orders of complexity—the named organ does not necessarily have only one sort of function but may incorporate


dozens of separate nuclei, cell types and neural pathways. The nomenclature I shall use here is fairly conventional.

The spinal cord

There is no difficulty in identifying the spinal cord as a distinct structure in the vertebrate nervous system, which serves to transmit information between the head and the body, but which also plays some part in the organisation of bodily activities. Even in higher vertebrates, we know that the running movements of a chicken may be coordinated by the spinal cord in the absence of a head, and that many reflexes can be elicited from the spinal cord of mammals when the connection with the brain has been lost (Sherrington, 1906). Leg withdrawal, away from irritating stimulation, can be observed in cats whose spinal cord has been severed above the relevant limb ('spinal' cats: Groves and Thompson, 1970) and a full range of sexual reflexes including erection, pelvic thrusting and ejaculation can be produced by tactile stimulation of spinal dogs (Hart, 1967). Similar spinal reflexes exist in all mammals, including man. Usually, higher brain centres are considered to modulate spinal activity by inhibitory influences as well as by taking active control. One can, if pressed, voluntarily inhibit the built-in spinal reflexes which would otherwise produce the withdrawal of one's hand from a flame. Motor activities such as movements of limbs are referred to as the skeletal functions of the spinal cord; vegetative activities such as digestion are referred to as autonomic processes. Our concern with thought might at first sight imply little interest in the autonomic control of digestion, perspiration and the like, but autonomic responses are often important indicators of inner mental states. Emotional feeling is intimately related to the racing heart and moist palms quantified by psychophysiologists, and the effect of external stimuli on glandular secretions of the stomach led Pavlov to the study of conditioning (see Chapter 3).

The brainstem and cerebellum (hindbrain)

The initial swelling out of the spinal cord when it reaches the brain, variously called the brainstem or the medulla, contains many nuclei


known to have particular sensory and motor functions. Information about sounds, motion detected in the inner ear, tastes and visceral states enters the brain here, and relevant motor reactions may arise directly from brainstem nuclei. A very obvious addition to the brainstem is the cerebellum, which is usually visible as a large external bulge. There is almost universal agreement that the cerebellum is specialised for the automatic fine co- ordination of fast movements. There is also a consensus that the hindbrain in general and the cerebellum in particular has undergone relatively little change 'in the course of evolution'. We should regard this assumption of lack of phylogenetic change in the hindbrain with some suspicion—it is not so much that there is no evidence of elaboration and specialisation in hindbrain structures but rather that a contrast is drawn with more radical developments which are presumed to have taken place in the forebrain. In particular the cerebellum is large and well differentiated in higher vertebrates, with strong connections (via the pons in the brainstem) with the cerebral hemispheres.

Apart from specialised sensory and motor nuclei, the brainstem contains a non-specific network of neurons of a relatively simple shape (Ramon-Moliner and Nauta, 1966) which act in concert without depending on any particular sensory modality. This network is known as the brainstem reticular formation, although it extends down into the spinal cord and forward through the midbrain into the thalamus. The function of the reticular formation appears to be to determine a general level of sensory alertness (something interesting in one modality thus engages the others) and also, by the descending rather than the ascending part of the system, to influence the overall intensity of muscular activity. It has a pervasive role in the degree of activation or 'arousal' of the brain, but also uses particular channels for the sensitisation of perceptual mechanisms and for the control of such things as muscle tone and respiration. Although both the form and the function of the reticular system are phylogenetically primitive, this brainstem mechanism has to be taken into account in sophisticated discussion of human attention and awareness (Dixon, 1970). In combination with other brainstem circuits, the reticular formation controls states of consciousness and the sleep-wakefulness cycle.

The spinal cord and the brainstem together provide the site for the lower motor system—a pool of interneurons which, though not themselves directly connected to muscles, serve as the 'final common path' for several sources of brain output to muscles.


The midbrain

It is at the level of the midbrain that the battle lines of phylogenetic theories are usually drawn, since the hindbrain is taken as common to all vertebrate classes, with regions above competing for the privilege of dominating it. In the lamprey, shark, cod, frog, alligator or pigeon there is a distinctive midbrain feature in a pair of clearly visible optic lobes (collectively referred to as the tectum, or optic tectum). In the lamprey and in many teleost fish the optic lobes are larger than the cerebral hemispheres. In these and in all other non-mammalian species the midbrain tectum is the main (but not the only) recipient of optic nerve fibres. A common conclusion drawn from these two anatomical facts is that the midbrain is a visually driven 'highest correlation centre' in fish (e.g. Romer, 1962). The optic lobes receive auditory and somatosensory inputs from the brainstem, and send back massive outputs to hindbrain motor nuclei in non-mammals—it is thus assumed that the midbrain supervises hindbrain functions in these cases. In mammals the tectum is hidden from external view: it is renamed (the superior and inferior colliculi for the visual and auditory centres) and sometimes forgotten. One of the clearest examples of the encephalisation of function theory is the supposition that in mammals the relative reduction in size and the change in appearance of the midbrain has been accompanied by a sweeping degeneration of its function, connected with the substitution of the forebrain as the chief destination for visual information. If the forebrain, and in particular the cerebral cortex, performs in mammals the analyses, associations and directions left to the midbrain in lower species, then the mammalian midbrain should indeed be redundant. However Weiskrantz (1961), among others, voiced doubts about the evolutionary logic of developing a new organ to do the same job as an old one, and subsequent anatomical and behavioural evidence suggests that the difference between mammalian and non-mammalian midbrain function is not as clear cut as the encephalisation of function doctrine proposes.



Figure 3 The external appearance of vertebrate brains. A selection of vertebrate brains drawn to the same scale. The cerebrum, or forebrain, is labelled, and can be seen to increase in size through the series, although due allowance should be made for the overall body size of the animals involved. The opossum, for instance, would weigh at least four times as much as the pigeon, while its brain is only a little larger than the pigeon's. On the other hand a cat and a macaque monkey may weigh about the same amount (say 4 kg) overall, but the cat's brain would be under 30 gm, and the monkey's brain over 6o gm. The spinal cord is visible in all cases and the cerebellum, just above the spinal cord and finely convoluted, can be seen clearly in the pigeon brain, and in the larger examples. The optic lobe, or tectum, of the pigeon midbrain can be seen as an oval below the cerebrum. (From 'The Brain' by D. H. Hubel Copyright 1979 by Scientific American, Inc. All rights reserved)

The thalamus and hypothalamus (diencephalon)

The next step on the linear progression from the spinal cord to the nose end of the central nervous system is the region of the forebrain known


as the diencephalon, which includes the thalamus and hypothalamus. These two structures give a further illustration of the multiplicity of purposes served at each stage of the anatomical progression. While the thalamus (in the upper part of the diencephalon) contains sensory projections and association circuits, the hypothalamus underneath it is related to visceral and metabolic functions (among other things, the hypothalamus controls the pituitary gland). The effects of surgical damage to the hypothalamus, and the behavioural responses produced by its electrical or chemical stimulation, support the general designation of the hypothalamus as a centre of drive and motivation.

There are a number of opinions as to how the diencephalon fits into the pattern of encephalisation. A rather unusual assertion is made by Rose (1976), who says 'It is the THALAMUS which is dominant in amphibia. ... In evolutionary terms, the thalamus was not to remain dominant for long—just enough for the regions of the thalamencephalon to sprout the pineal gland, the hypothalamus and the pituitary' (Rose, 1976, p. 167: original capitals). Others hold quite different views: 'In fishes and amphibians the tectum appears to be the true "heart" of the nervous system. . .. The thalamus proper is in lower vertebrates an area of modest importance' (Romer, 1962, p. 44). Rose's assertion carries the implication that the hypothalamus, the pituitary and the endocrine (hormonal) system governed by it had to await the arrival of land-going vertebrates. It is a safer bet that the traditional characterisation of the hypothalamus as the 'head ganglion of the autonomic nervous system' applies to all vertebrates. We may note that an important aspect of the relationship between the hypothalamus and the pituitary gland in mammals—the secretion of hormones by the hypothalamus itself—was first described in a teleost fish (Scharrer, 1928). Comparative study of the pituitary and the endocrine system has tended to follow, rather than lead, comparative neuro-anatomy, and has not given rise to such strong theories of phylogenetic progression (Bentley, 1976). It is not the case that hormonal control is a primitive characteristic from which higher vertebrates become emancipated, although this has sometimes been suggested (Beach, 1947): the pituitary may be said to increase in complexity and organisation in higher animals with as much justification as most other brain structures, and possibly the most singular achievement of mammals—lactation—has required its own endocrine specialisations. Oddly, the gonadal (sex) hormones appear to be identical, from fish to man, despite the astounding range of sexual


practices which they sustain. Even the development of lactation in mammals did not require a unique mammalian hormone—the same chemical which induces milk production in mammals (prolactin) is found in other vertebrates. A few non- mammalian species, notably pigeons, manufacture milk-like substances with which the young are fed, in response to prolactin (Chadwick, 1977). It seems clear that the functions of the hypothalamus and the pituitary do not fit neatly into the doctrine of progressive encephalisation.


Sensory representation in the thalamus

In man, the thalamus is often referred to as a 'sensory clearing house', 'sensory relay station', or 'antechamber to the cerebral cortex' (e.g. Elliot, 1963). This is because all sensory information, with the single exception of olfactory input, gets to the forebrain only after passing though the thalamus. It may be convenient at this point to look at Figure 4 (p. 175) which illustrates, for instance, that cutaneous sensation (from the skin) must pass successively from the spinal cord through the brainstem, midbrain and thalamus before reaching the mammalian cortex. As the reader might by now have come to expect, this linear sequence of sensory stages has been rudely transformed into a theory of phylogenetic progression and thus the thalamus is likely to be regarded as phylogenetically advanced. One theory (supported by Rose, 1976) is that the thalamus originally succeeded the midbrain as the dominant correlation centre in amphibians or reptiles, but then became reduced to merely passing on sensory information to the superior cortex of mammals. A contrasting suggestion is that the thalamus in lower vertebrates is merely an extension of the dominant midbrain serving in a rudimentary way to connect audio-visual matters in the tectum with smells in the cerebral hemispheres. On this view the thalamus only really developed to relay detailed sensory projections to the cerebral cortex of mammals (Diamond and Hall, 1969; Romer, 1962).

The latter used to be the main theory: lacking cortex to relay to, the non-mammalian thalamus is diffuse and underdeveloped and works together with non-cortical parts of the hemispheres as a controller of reflexes. But in the last ten years or so anatomical evidence has mounted against it. It is now apparent that the thalamus in reptiles and birds contains sensory projection nuclei comparable in organisation and complexity to those in the thalamus of mammals (Nauta and Karten, 1970; Webster, 1973; Hall and Ebner, 1970; Karten, 1979).


This is not to say, of course, that the thalamus is identical, or entirely equivalent, in fish, amphibian, reptile, bird and mammal. Differences between the classes there certainly are (Riss et al., 1972) but they need to be interpreted cautiously, and the characterisation of the dorsal thalamus as a uniquely mammalian acquisition, which was added to primitive ancestral thalamic nuclei which suffice for lower vertebrates, is becoming increasingly outmoded.

The cerebral hemispheres (cortex, corpus striatum and limbic system—the telencephalon)

The forebrain is traditionally divided into the diencephalon, which I have just discussed, and the telencephalon, which is more conveniently referred to as the cerebral hemispheres. It is the evolution of the cerebral hemispheres which is given most weight in theories of how higher forms of intelligence emerged from the shadows of lower vertebrate life. The hemispheres are clearly visible as paired bulges at the front of all vertebrate brains, but show an undeniable trend of increasing size in some degree of correspondence with the notorious phylogenetic scale (see Figure 3, p. 150).

It is equally undeniable that the cerebral hemispheres form a substantial part of the brains of even the lowest fishes and that they are especially large in sharks. The direction of theory about divisions of function within the hemispheres has thus tended towards the general principle of finding important parts in the hemispheres of mammals, and less important parts in the hemispheres of lower vertebrates. In man, almost the entire surface of the hemispheres is covered by a thin cladding of grey matter, the neocortex or cortex, which is composed of six layers of neurons. Rather similar types of nerve-cell are seen on the surface of the cerebral hemispheres of reptiles and birds (Webster, K. P. 1973), and to a lesser extent in fish and amphibian brains (Pearson and Pearson, 1976). But the mammalian cortex provides a special kind of neural organisation, and the outside, or pallium of the hemispheres does not have much significance in nonmammalian classes. Within the hemispheres there are of course bundles of nerve axons (fibre tracts) connecting various regions of cortex one with another and with other forebrain structures. These connecting pathways form the white matter, which takes up a considerable proportion of the internal volume of large mammalian


brains. But the interior of the hemispheres also contains solid nuclei of neurons which can be grouped into either the corpus striatum (the basal ganglia) or the telencephalic components of the limbic system—a heterogeneous collection of structures which includes the hypothalamus and other parts of the diencephalon.


Mammalian cortex

Although the mammalian cortex is considered as a single entity for the purpose of comparison with other vertebrate classes, it is composed of nerve-cells which differ in shape, size, and dendritic structure, and this provides a basis for distinguishing its several laminations and for mapping slightly different kinds of neocortex that are distributed over the brain surface. The identification of six separate layers of neocortex is reasonably clear, although the boundaries between the layers are not always particularly sharp. Mammalian cortex can be divided into categories of complexity, with the six-layered neocortex said to be phylogenetically newest, 3-layer archi-cortex or 'allo- cortex' the oldest, and 4- or 5-layer paleo-cortex or 'juxta- allo-cortex' intermediate. The differentness of mammalian cortex from the reptilian version has been a matter of some dispute, but the detailed comparisons made by Poliakov (1964) led him to emphasise the similarity between reptilian cortex and the simpler kinds of mammalian cortex as they occur in, for instance, hedgehogs.

The distribution of various types of cell-predominance in the cortex of many mammalian species was mapped by A. W. Campbell (1905) and Brodmann (1909) and found to follow family relationships between species so that the map of the human brain is very similar to that of the chimpanzee (Kappers et al. 1936).

The advantages of spreading out neurons in a two- dimensional array on the brain surface would appear to lie in the ease of access thus given to underlying connecting fibres; the same plan of neuronal elements on the surface, with connections radiating out from inside, appears in the hindbrain cerebellum and optic lobes of birds and lower vertebrates (and mammals). The disadvantage of such a system is that when overall brain size expands, the surface area increases with only the square of linear dimensions while internal volume increases with the cube, and thus the relative amount of space available on the surface will become unduly limited. A numerically convenient example is to imagine constructing a hypothetical brain or micro-computer by Placing 1 mm-square memory element plates on the exposed five


surfaces of a 1 cm cube, this 1 cm cube being filled up with little 1 mm cubes for internal connections. This would give 500 surface memory elements and exactly twice as many (1,000) internal programming elements. Now suppose you decide to double the size of this imaginary device, keeping to the same plan, and using the same components. Choosing a 2 cm outer cube would allow you 8,000 units for internal programming and, according to the original proportions, there should be one memory plate on the surface for every two of these. Thus 4,000 'cortical' elements are called for. But on the surface of your 2 cm cube there is only room for half that number (2,000) on the five sides (400 to each 2 cm square side). To keep to the original proportions more surface space is needed, and one obvious solution is to corrugate the sides of the cube.

Something very like this solution seems to have been adopted in all orders of mammals, including monotremes and marsupials. In every order, small animals have smooth-surfaced cerebral hemispheres, but larger animals of the same type, who have larger brains, have an increasingly convoluted surface of the hemispheres, allowing them not to get too far away from the proportion of cortical area to brain volume used by the smaller members of the order. In fact, one of the older findings of comparative anatomy, the 'Law of Baillarger and Dareste' (see Kappers et al. 1936, pp. 1518ff.), is that the folding of the surface of the hemispheres does not quite maintain the ratio of surface area to volume, so that larger brains usually have more internal volume, and more white matter, per square inch of surface cortical grey matter (Baillarger, 1845; Dareste, 1862). A quantitative confirmation of this finding has more recently been published by Elias and Schwartz (1969).

The whole question of progressive increases in brain size, and their possible implications for intelligence, is exceedingly complicated, as we have already seen. It is at first sight very odd, however, that the layout of the mammalian forebrain, with seminal interchanges between an outer crust and inner nuclei, would be more directly applicable to a small-sized system, while the arrangement which appears in the other higher vertebrates, the very much smaller birds, relies on connections between solid, non- surface neuronal structures, a plan which creates fewer problems of scale when used in larger brains. It is tempting to suggest an historical account for this discrepancy, since the fossil evidence has always been interpreted as showing that mammalian features first appeared in very small animals, and that the birds on the


other hand arose from a line of very much larger reptiles, the same line that bred gargantuan dinosaurs. (There has been speculation that dinosaurs were warm-blooded, and were the immediate ancestors of birds, and Jerison concludes that dinosaurs had a normal relationship between brain weight and body weight: they were not exceptionally small-brained, and the absolute volume of a large dinosaur brain would have been several hundred times larger than that of a small mammalian brain.) It would be a pleasing illustration of the role of accident and conservatism in brain evolution if this is indeed the case and large mammals, such as ourselves, have had literally to go into convolutions to adapt what started as a small-brain design, whereas the plan originally used by large reptiles is now seen to best advantage in a sparrow.


The corpus striatum (the basal ganglia)

In mammals the floor of the hemispheres is occupied by several nuclei surrounding the thalamus and collectively called the corpus striatum or the basal ganglia. The 'striatum' in the name derives from the striped appearance of the cross-section of these structures, but this striping is on a somewhat different scale from the fine layering of the cortex. However, in both cases it is alternations of layers of cell bodies with layers of fibres that produces the striping. The main role assigned to the corpus striatum in mammals is as a stage of the 'extrapyramidal' motor pathway, which sends efferents down through the midbrain and brainstem reticular system to the spinal cord and the cranial nerves. This is as distinct from the 'pyramidal' pathway which connects pyramidal cells in the motor cortex directly to the lower motor neurons, without any relays at intermediate levels. When in higher mammals the cortex is supposed to control voluntary, intelligent and skilled activities via the pyramidal pathway, the corpus striatum has been left stranded with instinctive, stereotyped and mechanical functions. Put crudely, the cortex has been assigned the thoughtful and the corpus striatum the unthinking aspects of motor control.

It suits the principle of encephalisation to allow for stability of function in this instance: since the hemispheres of reptiles and birds contain a large identifiable corpus striatum and very small amounts of rudimentary cortex, it would follow that the activities of both birds and reptiles should be largely instinctive, stereotyped and mechanical, even though the hemispheres of a bird may be as big as those possessed


by a mammal with the same body weight. However, while the psychological capacities of reptiles remain obscure (Burghardt, 1977), the performance of birds on the standard tests of learning ability does not always suffer by comparison with mammals. And neuroanatomists have now discovered that connections to parts of the striatal mass of the hemispheres of reptiles and birds are at least analogous and possibly homologous to the connections between the thalamus and the neocortex of mammals (Nauta and Karten, 1970; Webster, 1973; Ebbesson, 1980). Cells in these sensory projection areas of the avian striatum react in the same way as the corresponding cells in mammalian cortex in electrophysiological experiments (Revzin, 1969; Pettigrew and Konishi, 1976; Karten, 1979) and lesions to the visual projections in the bird striatum can be shown to affect performance on behavioural tests of visual perception (Stettner, 1974; Macphail, 1975; Hodos, 1976).

Apparently mammals have adopted a particular form of construction of the cerebral hemispheres, relying heavily on the spreading out of neurons in the cortical sheet over the surface, while in birds functionally similar neurons are compressed into a more restricted region, previously identified as the hyperstriatum or 'Wulst' which can often be seen as a bulge somewhere on the top of their cerebral hemispheres. The Wulst is confined to birds, and there are many variations of terminology applied to the thalamic projections of non-mammalian hemispheres, but there is a measure of agreement that the telencephalon of reptiles and birds (and possibly of fish and amphibians also) can be divided into an 'internal striatum', which corresponds roughly with the corpus striatum of mammals, and an 'external striatum' plus remaining bits of primitive cortex (which tend to be fused with the external striatum in birds but occupy a more separate location in most reptiles). On the anatomical evidence to hand it is not unreasonable to suppose that the external striatum in non-mammals has some functions in common with mammalian cortex. Some such degree of similarity between the cerebral hemispheres of mammals and other vertebrates is assumed in Figure 4 (p. 175).


The limbic system (hippocampus, septal area, amygdala, etc.)

One of the constancies of vertebrate brain geography is that the olfactory sense projects directly into the cerebral hemispheres without the intervening relays in the brainstem, midbrain and/or thalamus


which apply to all other sensory modalities. Either the entire cerebrum, or some part of it, can therefore be identified with olfaction. For lower vertebrates (fish, amphibians and reptiles) the telencephalon has frequently been called the 'smell brain', with the explicit assumption that it has only a marginal role in any other function. The endearing simplicity of this assumption has recently been shattered by a single blow: when olfactory pathways are picked out by modern histological methods, it turns out that these projections are confined to only a small part of the cerebral hemispheres, even in sharks (Ebbesson and Northcutt, 1976). The structures which have a relatively close connection with olfactory pathways are part of what is known as the limbic system. In birds (which have rather little sense of smell), higher mammals, and probably in vertebrates generally, the limbic system is the locus of brain circuits which are fundamental to bodily needs and drives, and to motivation and emotion.

It is usually held that the limbic system has had a long and fairly stable evolutionary history, and the limbic system is sometimes characterised as the 'reptilian core brain' or the 'proto-reptilian brain' (Isaacson, 1974). This may be misleading: it is certainly wrong to say that the mammalian brain contains within it a limbic system similar to that of a reptile, on which the mammalian cortex has been superimposed. A palliative is supplied by extensive evidence of evolutionary changes within the limbic system, which is involved in human emotion and feeling and also in the emotional aspects of human communication and language (Lamendella, 1977). However, it should be remembered that there is no simple phylogenetic sequence of modality dominance, with smell a primitive vertebrate specialisation and hearing and sight later refinements, which would support the theory that the limbic system began as a set of structures exclusively concerned with olfaction but gradually and continuously changed in its functions as the importance of olfaction declined.

Sight was one of the earliest vertebrate inventions and specialisations, and this fact is used in connection which the argument for midbrain dominance in fishes and amphibians. Mammals, as a class being originally nocturnal, first discounted vision, and emphasised smell and touch very much more than reptiles or birds. It is arguable that both mammals and birds made a quantum leap in techniques for hearing, by re-assigning bones which are part of the jaw in reptiles to a sound-transmitting job in the middle ear. The need in mammals for brain apparatus to codify the additional auditory information thus


obtained is given great play by Jerison (1973), but touch and smell are better candidates for modalities which led to mammalian distinctiveness, as they are not well developed in birds. The continuation, or possible re- emergence of vision as a dominant modality in tree-living insectivores, and primates, is a special factor: lower vertebrates and birds generally have good colour vision whereas mammals other than primates generally do not have colour vision—having lost it, we assume, during the early phases of nocturnality. The main point is that in so far as the limbic system is concerned with olfaction, there is no reason to suppose that it is of great importance to highly visual species, and these include the majority of non- mammalian vertebrates, and in particular the majority of reptiles. (One well-known reptile, the arboreal chameleon, has such well-developed eyes that it is referred to as a 'living microscope': Walls, 1942.)

Perhaps it is a mistake to group the various limbic structures together too firmly in the first place. The hippocampus is cortex, actually on the surface of the hemispheres (dorso-medially) in most non- mammals, but submerged down at the bottom of hemispheres in most mammals. There are only three cellular layers to the hippocampus, which perhaps makes it second-rate cortex in mammals, but this is as good as cortex comes in other classes. From the hippocampus, the fornix provides extensive two-way connections with the hypothalamus (terminating in the mammillary bodies). Non-cortical limbic components in the telencephalon include (1) the amygdala, which is continuous with the basal ganglia or internal striatum and, like the hippocampus, is well connected with the hypothalamus (via the stria terminalis); (2) the septal area, which has anatomical pathways both to the hippocampus (via the fornix) and to the hypothalamus (via the median forebrain bundle). As well as the hypothalamus, other parts of the diencephalon such as the anterior thalamus and the habenular nuclei are also considered to be part of the limbic system. The above description is for mammals, but homologous components and connections of the limbic system are ascribed to most other vertebrates (Kappers et al. 1936).

It is obvious that there are extremely rich interconnections within the limbic system, and the effects of electrical stimulation and lesion damage also provides justification for the belief that the limbic structures act in concert in a manner which suggests emotional regulation of one sort or another. The pattern of interconnections, along with analyses of cell types, is what has led to the identification of


limbic structures across vertebrate classes. But of course it is extremely unlikely that what is identified as the hippocampus in lampreys, the septal area in a frog, or the amygdala (archistriatum) in birds, has the same functions as the corresponding brain regions in a mammal. There is certainly no reason to regard the limbic system as especially reptilian.

The two halves of the brain—bilateral organisation and decussation

One almost universal feature of vertebrate brain organisation, which I have so far conveniently ignored, is the existence of separate circuits for inputs and outputs to and from the left and right sides. Bilateral symmetry of the body is a fact of life for all vertebrates but the importance of separate consideration of the two halves of the bilaterally symmetrical vertebrate brain arises partly because of the special role apparently played by an asymmetrical assignment of duties in the human brain. The human left hemisphere seems to dominate the right in the control of speech and language, skilled manipulation, reasoning, and conscious experience, while the right hemisphere holds sway over emotion, intuition, and unconscious thought (Gazzaniga, 1975; Corballis and Beale, 1976; Dimond, 1972). This kind of division of labour with a line of lateral demarcation between the hemispheres is thought by some to be one of the dimensions on which the human brain differs from that of other vertebrates (Levy, 1969, 1977), but the general problems associated with categorising information into left and right are by no means confined to the human species.

In all vertebrate classes, there is separate neural input from the left and right sides of the body, from the left and right eyes and ears (or lateral pressure sensors), and from the left and right olfactory bulbs. Similarly there is separate neural output to muscles on the left and right sides of the body. We may ask, what would be the simplest and most primitive way of fitting together these four lines of information —-- left and right input, and left and right output? Anatomically it would be most straightforward to keep all left-side lines of information on the left of the brain, and all right-side lines on the right of the brain, but this by itself would have the rather serious disadvantage of isolating one half of the body from the other. In order


to co-ordinate the whole animal we should want both sides of sensory input to eventually get to both sides of motor output, and this would clearly require some kind of interaction between the two sides of the brain. The simplest solution would seem to be to put left-side information in the left half of the brain, and right-side information in the right half of the brain, but to cross-connect the two brain halves. In the typical vertebrates solution the two halves of the brain are indeed cross-connected, but in addition left- side information is put on the right side of the brain and vice versa.

The basic switch, which means for instance that the left hemisphere tends to control the right limbs, is referred to as 'crossed-lateral control' and no one seems to know why it takes place. If the two halves of the brain are going to be cross-connected anyway, it seems to be an unnecessary complication. Sarnat and Netsky (1974) suggest that crossed-lateral control arose in the first place because of a primitive need for left-side input to be converted to right-side output, and vice versa. In Amphioxus, the swimming chordate supposed to represent a vertebrate ancestor, there is a powerful coiling reflex produced by the contraction of muscles on the side opposite to the one prodded. But this does not take us very far in explaining why, in vertebrates, both left-side tactile input and left-side motor output tend to be localised in the right side of the brain.

By comparison, the reason for direct connections between the two sides of the brain is obvious. If an animal is to act as an integrated whole, and not as two independent halves, shunting circuits across the brain are clearly required. Pathways strung across from similar points on either side of the central nervous system could do this job. Usually such tracts are called commissures, and Kappers et al. (1936) make a sharp distinction between these and 'decussations' which occur when fibres from a certain point on one side of the brain cross over to a quite different location on the other side. However, it seems possible that 'partial decussation' could serve as a sort of diagonal commissure. In partial decussation connections up and down in the nervous system are to both the same and the opposite side, and thus information could be diagonally distributed across the midline as an alternative to the up-and-across method.

For whatever purposes, all vertebrate brains make very extensive use of both commissures and decussations. Ventral and dorsal commissures occur within all vertebrate spinal cords. Similarly, commissures directly traversing the brain are found in the brainstem,


the cerebellum, the midbrain and the forebrain, with some degree of regularity in the various vertebrate classes. Emphasis is usually given to forebrain commissures, partly because a major new tract, the corpus callosum, evolved to interconnect the neocortex in the two hemispheres of placental mammals. It should not be forgotten, though, that other forebrain commissure can provide for hemispheric cross-talk in the absence of the corpus callosum. Monotremes and marsupials possess mammalian hemispheres but lack the corpus callosum, and rely on having large anterior and hippocampal commissures. The forebrain anterior commissure is phylogenetically very stable, being present in lampreys and sharks, and retained in higher mammals and man adjacent to the front end of the corpus callosum itself (Putnam et al., 1968). Other forebrain cross-connections vary more from class to class and also vary within classes. Lampreys and teleost fish are assigned a dorsal forebrain commissure, and sharks have a number of commissure-like paths including the superior telencephalic commissure associated with 'hippocampal' areas. In amphibians the superior forebrain commissure has rather different terminations, and there is a separate hippocampal commissure, giving three main links between the cerebral hemispheres altogether. All extant reptiles have a hippocampal commissure in addition to the anterior commissure, and lizards and snakes have a third one, again considered to be associated with the hippocampus (Pearson and Pearson, 1976).

Interhemispheric connections are still somewhat mysterious in birds, since functional interaction is more obvious than the anatomical routes, but the anterior commissure in birds, like that in mammals, contains a section that terminates in the amygdaloid ('archistriatal') regions (Kappers et al., 1936; Pearson, 1972; Cuenod, 1974). Mammals retain a separate hippocampal commissure in the psalterium (composed of crossing fibres from the fornix) as well as the anterior commissure, along with the new corpus callosum.

Apart from direct transverse commissures, there are many decussations or diagonal crossings, especially in the course of sensory input and motor output, in all vertebrates. In general decussations are partial, involving same-sided as well as opposite-sided transmission. Given the profusion of commissures and decussations in all vertebrate brains, it is reasonable to assume that much information is in principle transferable from side to side. When I deal with other aspects of brain design, looking primarily at relations between hierarchically organised 'up—down' stages of processing, interaction between the two


halves of what is through and through a paired system will be taken for granted. However the functional effectiveness of lateral co-operation (or in the case of cerebral dominance, lack of co-operation) is subject to experimental test. In general animals appear to experience little difficulty in giving universal application to information received from one side of the body. For instance, visual stimuli received initially by one eye, are normally recognised by the other eye, even in animals (such as the goldfish) where there is a complete decussation of the optic nerves (Ingle and Campbell, 1977). Phylogenetic theory would suggest that the channel of transmission in this case should be the midbrain commissures of the optic tectum, but it would appear that even in the goldfish forebrain decussations are involved in the transfer of visual information.

The replication of visual information is a clear example of the necessity of bilaterally available representations: an object seen with one eye, or in one half of the visual field, needs to be registered in both sides of the brain if an animal is not to be constantly surprised by each turn of its head. But as well as the requirement for duplex representation, there is also a need for some separation, so that the animal codes whether a seen object is on the left or the right. Perfect duplication of information would lead to trouble in distinguishing left and right. To a degree, such difficulties occur, especially in animals and children (Corballis and Beale, 1976). The importance of separating left and right is more obvious in the case of motor instructions than it is in perception—we must be able to lift one leg or the other, and not attempt to lift both at once. The presence of some degree of equivalence between motor commands to our own right and left limbs is apparent in the traditional problem of making circular movements on the stomach with one hand while doing non-circular pats on the head with the other. But in general limbs must work in conjunction, doing different things at the same time, even in the case of the movements of fins in fish. The brain must be able to label left and right in terms of motor commands at the same time as being able to recognise equivalences between left and right perceptual input. Conceivably, the need for left/right differentiation is the reason behind the peculiar crossed-lateral layout of sensory and motor pathways. Since any neuron which crosses the midline of the brain must violate strict mirror-plane bilateral symmetry, brains organised on the crossed-lateral plan with left-sided input and output going to the right half of the brain (and vice versa) have an additional source of


structural asymmetry by comparison with the more obvious design of keeping all left-side information in the left half of the brain (Walker, 1981). This might explain a curious exception to the crossed-lateral plan. The vertebrate rule is that sensory and motor pathways tend to cross from one side of the body to the other side of the brain but a major exception to this rule is that smells received by one nostril go into the cerebral hemisphere on the same side. If it is generally important to code sensory and motor information according to whether it pertains to left or right peripheral organs, but not important to code smells according to whether they are received by the left or right nostril, and the crossed-lateral layout serves the purpose of left/right differentiation, then the absence of crossed input from the nostrils makes sense.

The optic chiasma

The best-known example of the general rule of sensory decussation is that in the majority of vertebrate species the retinal output from one eye goes to the midbrain or thalamus in the opposite half of the brain. Most non-mammalian species have their eyes pointing sideways, and this means that a large part of the left visual field goes to the right side of the brain and vice versa. Some mammals, especially carnivores and primates, have less panoramic visual fields because both eyes point forward to survey more or less the same scene. An advantage of this is that for an object in the near distance the slight differences between the images projected onto the left and right retinas can be used to give the impression of depth. Perhaps in order to capitalise on this possibility, most mammals have partial decussation of the optic nerves, so that the left visual field, as seen by both eyes, goes to the right side of the brain and the right visual field, as seen by both eyes, goes to the left half of the brain.

The crossing of the optic nerves before they enter the brain is called the 'optic chiasma' in both mammals and non- mammals. Usually the optic chiasma is a total decussation in non-mammals, and a partial decussation in mammals, but there are some exceptions to this. At least two kinds of mammal, with little else in common, have total crossing of the optic nerves. These are the dolphins and whales, and the guinea pigs, which both have their eyes so much on the side that there is almost no overlap in the independent fields. By and large, mammals with side-directed eyes have almost completely crossed optic nerves,


and those with both eyes pointing forward, pre- eminently primates, have a relatively even partial decussation at the optic chiasma. The mammalian type of partial decussation at the optic chiasma is occasionally seen in lower vertebrates: in some amphibians; snakes and lizards; the sole living representative of the immediate precursors to teleost fish (Northcutt and Butler, 1976); and adult lampreys (Kennedy and Rubinson, 1977). Birds, turtles, teleosts and sharks all seem to conform to the non-mammalian pattern of total optic nerve crossing (Ebbesson, 1970). Whether or not the partial decussation of the optic chiasma that sometimes occurs in lower vertebrates is utilised for stereoscopic vision has not been behaviourally tested, but in the case of the teleost precursor, the long-nose gar, its ecological niche as a fast predator, chasing and catching fish that swim in front of it, implies that binocular distance perception would be a help.

Whether or not stereoscopic depth perception is possible in birds (which have totally crossed optic nerves) has been tested. As long as the individual fields of view of the two eyes overlap, and these fields can be compared at some stage, for instance by the use of midbrain or diencephalic commissures, there is no reason why stereopsis could not be accomplished without the mammalian convenience of partial decussation at the optic chiasma. That stereopsis occurs in many birds has often been suspected, but an ingenious and conclusive test has been performed on the falcon, by fitting a bird with goggles which allowed the placement of red and green filters over the separate eyes, and presenting it with the well-known Julesz type of array, by which three-dimensional subjective effects are produced in human observers due to disparities between red and green elements. The bird was successfully trained to fly only to the 'three-dimensional' displays, thus demonstrating that it possessed a mechanism for the perception of depth and distance by binocular disparity. Such a mechanism would undoubtedly be of real use in chasing and stooping on other flying birds. Cuenod (1974) and Pettigrew and Konishi (1976) have demonstrated that what happens in the owl and the pigeon, and probably in many other avian species, is that although the optic nerves themselves are totally crossed, bilateral projection from each eye to both cerebral hemispheres is brought about at the 'supra-optic decussation', which goes through the midbrain just above the optic chiasma itself, but is a visual projection pathway from the thalamus to the telencephalon. This means that the Wulst (hyperstriatum) of the owl hemispheres is able to react to binocular stimulation in roughly the


same way as the cat's visual cortex, even though the binocular information has come via a different route. Like the cat, the owl's eyes are pointed straight ahead, and no doubt depth perception is as useful to the owl, when pouncing on a mouse, as it is to the cat.

Apart from the use of binocular vision for depth perception at short distances by birds, it is likely that many of them assess long distances by bobbing their head up and down to take successive looks at objects with the same eye from different positions. The cocking of the head to look at the same object with either eye alternately is another strategy which may expand the possibilities of binocular perception with side facing eyes. Thus, although mammals possess a peripheral splitting of visual input, and a cortex for the analysis of binocular comparisons, it certainly does not follow that mammals are the only vertebrate class capable of seeing in three dimensions.

Brain asymmetries and human speech and handedness

The vertebrate brain, like the vertebrate body, is superficially remarkable for its anatomical symmetry. One half of the central nervous system is generally taken to mirror the other although marginal violations of mirror-plane symmetry must occur at the midline in cross-connections. A notable exception to the bilateral symmetry of visible gross anatomy is common in lower vertebrates in the habenular nuclei, which are an otherwise insignificant part of the diencephalon, usually classed as olfactory centres and components of the limbic system. The habenular nuclei on the left and right are markedly asymmetrical in lampreys and hagfish, sharks, and some teleosts and amphibians (Braitenberg and Kemali, 1970). The largest side is the right in the lowly cyclostomes, the left in the sharks and some frogs, and variable from species to species in the teleosts.

Not surprisingly, rather more attention has been given to the fact that anatomical asymmetries are measurable for certain areas of the cortex of the human cerebral hemispheres. Cunningham (1892) observed that the upward turn of the Sylvian fissure is more acute in the right hemisphere of the human brain than it is in the left. The Sylvian fissure divides the temporal lobe below from the frontal and parietal lobes above, and Geschwind and Levitsky (1968) examined a roughly triangular portion of the upper surface of the temporal lobe, inside the Sylvian fissure, termed the 'planum temporale'.


Measurements of the longitudinal extent of this area gave average figures of 3.6. ± 1.0 cm in the left hemisphere and 2.7 ± 1.2 cm in the right hemisphere, the left figure exceeding the right in 65 of the 100 brains studied. On its own this is a less than compelling variation in physical structure, but of course the favouring of the right hand over the left for writing and other manual skills has been a fact of human life throughout recorded history (Hardyck and Petrinovich, 1977) and the restriction of the more sophisticated mechanisms governing speech to only one hemisphere, which is usually the left, has now become equally well established (Gazzaniga, 1975; Warrington and Pratt, 1973). That there are implications for theories of cognition is clear, although exactly what the implications are is somewhat less than clear.

In the context of animal thought it is obligatory to consider the extreme possibility that all forms of cognition are completely dependent on an exclusively human degree of anatomical asymmetry and lateralisation of function in the cerebral hemispheres. We must therefore examine briefly how asymmetrical functioning in the human brain might be related to the more general questions of symmetry and duplication of function in the vertebrate nervous system. I will deal with handedness and speech separately, and ignore for the present the many other interesting but subtle specialisations of the two halves of the brain that have been detected (Walker, 1980; Denenberg, 1981).



The fact that each human limb tends to be served by the opposite side of the brain is not, of course, remarkable. What is distinctive about human handedness is that one side of the brain, and therefore one hand, appears to be better than the other. Vertebrate species apart from ourselves seem to be wonderfully ambidextrous, both in the sense that individuals are capable of performing most skills necessary to them with either side of the body and in the sense that when individual side preferences can be found, they are distributed very evenly within the species. It should be noted that in a given population of rats, or monkeys, most individuals will prefer a particular forelimb for a simple task such as reaching out for food, but there will be the same number of 'left-handers' as 'right-handers' (Peterson, 1934; Lehman, 1978; Collins, 1977).

What can have induced the human species to become predominantly right-handed? There is no shortage of theories, but a specialisation of some sort would seem to be the most obvious


advantage of manual asymmetry. Reserving the left hand to place over one's heart is one of the oldest and least plausible theories, but incorporates an interesting feature—the task of the 'non-preferred' hand. For purely one- handed activities the usefulness of not being ambidextrous is obscure, given the assumption that the massive human corpus callosum should enable skills to be transferred from one limb to the other. If, however, there are important activities in which the two hands do different things then a degree of isolation of the separate skills might be helpful. The manufacture of stone tools is a distinctively human, two- handed and asymmetrical skill which was sufficiently important during the period of human evolution to have supplied a unique selection pressure. As the nature of the paleontological evidence suggests that it began at least two or three million years ago and continued while the brain grew from about 500 to about 1300 gm, the manufacture of tools seems on obvious candidate for a selection pressure for a 'holding and hammering' specialisation.

Reconstructions of tool-making techniques, and the habits of modern Australian aborigines, imply that the way to make stone tools is to hold a flint or bauxite core against the body or against an anvil stone, with the left hand, and swing at it with a hammer stone held in the right hand. Although we have difficulty in making different ballistic movements concurrently with separate arms, as in the rubbing and patting trick, we are well adapted for gripping and holding firmly with one hand while making finely controlled movements with the other. The utility of being able to do this could be a reason for the development of an asymmetry in arm use, but it does not of course explain why it is the right hand, rather than the left, which gets the more interesting work. Perhaps this was arbitrary, or due to a slight left-brain physiological superiority. But it must be pointed out that the left hand, and the right hemisphere, are not devoid of the biological capacity for carefully and finely controlled movements. The left hand of a virtuoso violinist may be said to accomplish the highest form of human manual dexterity. Usually when one hand is selected for delicate movement, we plump for the right, but if the left is forced into service it is not necessarily found wanting. The use of the left hand for the fingering of the violin and similar stringed instruments (and more recently for the fingering of the valves of the French horn) is almost certainly an historical accident, due to the preference for holding things with the left hand while making ballistic movements on or around them with the right. But such accidents serve as a useful


indication that some human asymmetries result from culture and convention as much as biological predestination. It should be remembered that handwriting, which is used as the modern index of human lateralisation, is confined to the last few thousand years at most, and to the last few hundred for all but a tiny fraction of each human generation. It is thus supremely irrelevant as a selection pressure, though perhaps not quite so irrelevant as a connection between human lateralisation and language.


Human language

The universal use of spoken propositional communication by the human (or a pre-human) species was undoubtedly a Rubicon whose crossing led to new pastures of group activity, social organisation and, eventually, civilisation, which are denied to all animals that remain on its other side. The fixation of the mechanisms which enable the use of so decisive a facility on only one side of the brain is something of a puzzle—if ever there was a case for doubling up on brain circuits as a fail-safe device, language would surely be it. A number of speculations on this theme may be presented. First, it is conceivable that language was not of such overwhelming importance at earlier stages of human evolution as it is in modern literate societies. Second, assuming that language is crucial, being able to accomplish speech with only half a brain has its points as a form of insurance. Language is not genetically restricted to the left side of the brain since damage to this hemisphere (or its removal), in childhood, means that the remaining right hemisphere does the job. There is very little firm evidence to support the contention that those left-handers who use the right side of the brain for all, or part, of speech control suffer from significant impairment of speech or other faculties (Hardyck, 1977). Needing only one side of the brain for speech might thus be supposed to be somewhat safer than needing both, but duplicating speech mechanisms on both sides would be safer still.

An argument for the non-duplication of language mechanisms in the two cerebral hemispheres is that space which would be taken up by duplication can be put to better use (Levy, 1969). Given that unilateral brain injury is not a major feature of the human condition (at least in the absence of car accidents and strokes) it would perhaps be unduly cautious to waste brain space by the duplication of any function that is not inherently bilateral. The theory that when the left hemisphere takes command of speech the right hemisphere is freed for


other duties such as spatial awareness, and non-verbal imagination and intuition thus provides an explanation for cerebral lateralisation in terms of evolutionary economy, and has been broadly supported by various kinds of psychological testing since it was first put forward by Semmes (1968) and Levy (1969).

A difficulty with this strict division of labour theory is that it is perhaps more convincing than the data requires. For performances other than linguistic ones, differences between the hemispheres are more a matter of degree (and a matter of serving the two sides of the body) than a revelation of qualitatively quite different modes of functioning. Further, theoretical advantages of strong lateralisation imply severe deficits if it is much reduced, and the absence of catastrophic impairments in the capacities of moderately left-handed individuals whose brains tend to be anatomically and functionally symmetrical (Hardyck, 1977; Lemay, 1976) indicates that brain lateralisation is not a necessary condition for the attainment of uniquely human cognition.

What then are the implications of asymmetries in human brain function for the psychological gap between humans and other animals? Man seems to share with other animals the general duplication of the nervous system, both in terms of the normal programming of two-sided behaviours and in terms of the ability of one hemisphere to take over whole-brain capacities after damage to the other in childhood. But for speech and handedness the typical adult human brain appears to follow a very distinctive strategy which involves specialisation of the left half of the brain for these functions. If the demands of human thought and intellect are so much greater than the demands placed on the brains of other species that a unique division of labour, even in an enlarged brain, has been necessary to meet them, this certainly widens the gap between human and animal cognition. On the other hand, if lateralisation of brain functions confers so much benefit on man, should we not expect that similar strategies should have been resorted to at least occasionally in other species, which may have more limited need for cognition but have equally limited amounts of available brain tissue? Unfortunately evidence which bears on this last question is not easy to come by. However, a single finding of some significance concerns vocalisation in birds.

Vocalisation in birds as in man is controlled by apparatus which does not require independent movements in its left and right halves. It


may be that the lack of necessity for equal but independent representation of the two halves of the speech apparatus was an important factor in the evolution of human brain asymmetries, as similar asymmetries appear to occur in song birds. A series of experiments by Nottebohm seems to have established beyond doubt that in several species of seed- eating passerines the left side of the brain and the left side of the peripheral nervous system is predominantly responsible for controlling song in the normal adult males (Nottebohm, 1971, 1976, 1977, 1979; Nottebohm et al., 1976). In the birds studied (the canary, the chaffinch and two species of sparrow) the organ of sound production, the syrinx, has muscles on the left and right sides, and each set responds to a separate branch of the hypoglossal cranial nerve. The left-hand set of syrinx muscles in considerably larger than the right, which is in itself suggestive, but clear results are obtained by severing the nerve to the left- hand muscles and comparing the effects of this operation with the effects of cutting the nerve to the muscles on the right side of the syrinx. If the cut is on the left, all, or almost all, of the syllables previously used by the canary (in a range of 20—40 syllables, identified by an auditory spectrogram) are lost, whereas if the right-side muscles are inactivated vocalisation even immediately afterwards is virtually unchanged, with only one or two syllables missing at the most. Since, in the canary, the left hypoglossal nerve comes from the left side of the brain, one might expect left- brain dominance of vocalisation in the canary, and other experiments confirm that this is so.

In man, there is a region of the frontal lobe of the left hemisphere which is called Broca's area because a French neurologist of that name discovered that patients in which it was damaged suffered from a form of aphasia. In the canary, there is a region in the left hemisphere which perhaps ought to be called 'Nottebohm's area', since Nottebohm claims that lesions at this place cause a major disruption of song. It is located in the external striatum (at the caudal or rear end of the hyperstriatum ventrale) which has been assumed on other grounds to be analogous to the cerebral cortex of mammals. Although lesions to this area in the left hemisphere resulted in the immediate loss of almost all song, lesions placed at the corresponding point in the right hemisphere allowed more than half of the usual syllables to be reproduced immediately after the operation in a pattern that to the human ear was indistinguishable from the pre-operation performance.


The prognosis for language disturbance (aphasia) which results from brain damage in human patients is not good, but significant recovery is not at all uncommon either spontaneously or with the assistance of speech therapy, especially after relatively clean lesions due to accidents or war wounds rather than strokes or tumours (Goldstein, 1948; Paradis, 1977). Nottebohm's canaries, even those which suffered a profound impairment of song, made a good recovery after the brain lesions in the course of the next year. Another parallel with human language mechanisms is that if the left branch of the hypoglossal nerve going to the syrinx is cut in young birds, the right-side structures very quickly take over the control of song. Whether or not auditory perception or comprehension of sounds is as much lateralised in the song-bird brain as vocal production seems to be remains to be seen. Acoustic lateralisation of a completely different kind is known in several species of owls, some of which have very pronounced physical asymmetries of the external ear, consistently in all individuals. It is presumed that the left/right differentiation in these cases is for the purpose of localising sounds in the vertical as well as the horizontal plane (Norberg, 1977).

It would be curious indeed if of all vertebrates only song birds, owls and humans benefited from relaxations of bilateral brain symmetry. But there is increasing evidence that left/right distinctions in brain function of one sort or another are not particularly uncommon (Walker, 1980; Denenberg, 1981). Unfortunately it is still by no means clear whether lateralisation of human brain function is foreshadowed in other primates, and this is obviously a crucial point. In terms of anatomical inequalities, Cunningham (1892) found that human Sylvian fissure asymmetries had counterparts in the great apes, and large monkeys, and this has been confirmed by Yeni-Komshian and Benson (1976), Lemay and Geschwind, (1975) and Cain and Wada (1979). In terms of recognition of species-specific cries Petersen et al. 1978) claim that left-hemisphere processing in Japanese macaque monkeys is analogous to that of humans listening to speech, and Dewson (1977) has also produced data that appear to demonstrate Preferential use of the left hemisphere for complex auditory functions in monkeys. The best assessment at present is, in my view, that asymmetries of brain function as such are not a human prerogative even though one of the things that is lateralised—language—is. Cerebral lateralisation is neither a necessary nor a sufficient condition


for human mental activity and therefore we need not suppose that degree of brain lateralisation determines the extent to which an animal species may be said to possess cognitions.

Hierarchical design in vertebrate brains

I shall now ignore the question of side- to-side interactions in the brain, and return to the idea that progression along the tail-end to nose-end axis of the brain is the fundamental dimension of vertebrate brain evolution. A rough plan of the anatomical relationships between the longitudinal divisions of the brain which I reviewed earlier in this chapter-hindbrain, midbrain and forebrain—is shown in Figure 4. A speculative simplification which is shared by phylogenetic theory, human neurology and other attempts to give an overall description of brain function (Riss, 1968a; Pearson, 1972) is the identification of a hierarchy of levels of brain activity. At the lowest level, the spinal cord clearly supplies the most direct route there is between sensory input and motor output. At the highest level, some regions of the mammalian cortex receive sensory information which has been successively transformed, filtered and classified at previous stages, and other regions of the cortex initiate actions which may be given more detailed organisation by lower brain divisions. Between the spinal cord and the cortex, the brainstem, midbrain, cerebellum, thalamus and corpus striatum can be assigned particular intermediary roles. Clearly an underlying assumption is that the higher levels of brain organisation are linked with complex forms of cognition and thought while lower levels are concerned only with more mechanical and reflexive matters.

It is hard to imagine how any sense could be made of the workings of the brain without some such assumption of a hierarchy of levels or progression of forms of integration, but a number of reservations ought to be expressed before I make further use of this kind of concept. First, many important details are necessarily ignored: identifying levels of function makes things more intelligible, but the levels may in some senses be convenient fictions. Second, if divisions of function are made, it is almost inevitable that one succumbs to the temptation of imposing spurious meanings on the divisions. As a cautionary example, even the split between sensory and motor functions is sometimes regarded with suspicion.



Figure 4 Naming of parts and general layout of the vertebrate brain. The top diagram is a rough sketch of the anatomical relationships between the structures usually identifiable in any vertebrate brain. The bottom diagram shows schematically the points of entry of the major sensory nerves, and grossly simplified pathways of neural connections within the brain (see text). In mammals and birds, the forebrain is larger than indicated here, and the brain as a whole is compressed, over the long axis (see Figure 3). In mammals, post-thalamic projections are composed of the neocortex of the cerebral hemispheres. (After Nauta and Karten, 1970)


The complexity of the sensory and motor systems

By interpreting brain organisation in terms of sensory input moving upwards through hindbrain, midbrain and forebrain tiers, progressively more remote from the sense organs themselves, with planned actions, co- ordinated response sequences and mechanical reflexes cascading back to peripheral muscles, one is in danger of overemphasising simple input and output categories. To counter this, it should be remembered that something like 99 per cent of the neurons in the brain and spinal cord cannot be classified as either sensory or motor, but may be considered as an 'intermediate net' coming between strictly sensory or strictly motor nerve cells (Nauta and Karten, 1970). The implication is that what begins as the firing off of sensory cell axons, and ends as the contraction of muscles, is subject to an enormous fanning out within the nervous system before a reversal of this process converges to produce behaviour. In between sensation and action the brain of even the most primitive vertebrate has a life of its own.

The fanning out of sensory information to multiple destinations is well illustrated by the review of the visual pathways given by Ebbesson (1970). This showed that the initial projection of fibres from the retina of the eye goes to as many as five or six different points, in all vertebrate classes. (This is not counting the possible duplications involved in bilateral projections to the two sides of the brain.) Leaving aside most of these, which may have rather narrow bands of function (for instance the input to the hypothalamus, which probably assists in things like responses to changing day-length), there are still two major visual pathways. One starts with the projection from the eye to the optic tectum in the midbrain, and is usually assumed to dominate the vision of lower vertebrates, and the other begins with the retinal projection to the thalamus (in mammals to the lateral geniculate nucleus—this is the instance where the forebrain pathway is supposed to have 'taken over' vision from the midbrain pathway). All vertebrates make use of this kind of double vision, possibly because one pathway detects moving objects while the other analyses stationary patterns. But several nuclei in the thalamus are involved in each case, and exceedingly complex transformations and abstractions may take place in the cerebral hemispheres, even in lower vertebrates such as sharks (Graebner, 1980).

Using at least two parallel but complementary interpretations of sensory input for a single modality may be the rule rather than the


exception. Acoustic information from each mammalian ear goes to two nuclei in the brainstem, and from them goes either direct to the thalamus (the medial geniculate nucleus) or, by a separate route, relays in the midbrain (the inferior colliculus) before going up through the thalamus to auditory cortex. Similarly it is well established that in mammals there are at least two projections of the body surface on the cortex (Adrian, 1941; Woolsey, 1965). Motor traffic also passes down double or triple lanes before converging on to muscles, the main distinction being between the pyramidal path direct from motor cortex to the spinal cord, and the extra-pyramidal circuit through non-cortical forebrain structures and the midbrain and brainstem, with detours taken by both to pick up elaborations of assignment in the cerebellum.

Clearly, much of the richness and variety of the central nervous system is lost if we condense function down to sensation, action, and correlations between the two: one should use terms like 'sensory' and 'motor' not as labels for elementary categories but as synonyms for problems whose practical solution required the evolution of an organ as complex as the brain, and whose interpretation in neuropsychological theory is still tentative.

Hypotheses of phylogenetic change in brain organisation

Bearing in mind the limitations of the basic concepts, we can now examine more closely ideas about how the organisation of vertebrates brains may change from class to class. Most of these ideas can be subsumed under the general heading of encephalisation, but I believe that speculation under this heading covers a multitude of theoretical sins, which I intend to categorise as hypotheses concerning take-over, addition, and conservation of function.

Take-over of functions

The most common form of the doctrine of encephalisation presupposes that behaviour-controlling functions become progressively concentrated at higher brain levels (that is, towards the front of the longitudinal layout: Figure 4, p. 175). This notion has been applied to the evolution of vertebrates from invertebrates, the evolution of


successive classes of vertebrate, and to differences between and within mammalian orders. With mammals, encephalisation is taken to mean corticalisation. It is often assumed, for instance, that primates, being more advanced, make greater use of their cortex than rodents or carnivores, and similarly that man is more dependent on cortical functions than prosimians, monkeys or apes, in that sequence of other primates.

The origin of the take-over of function hypothesis is that the business of any given part of the central nervous system can be seen in broad terms as the interpretation of sensory information and the control and organisation of behaviour. Once this is done there is a tendency to want to appoint some brain division as 'in charge' of all the others, and this sets the stage for phylogenetic take-overs of the executive position. The dominance of a particular brain level is often linked with the duties of correlation, integration or association, so we may find references to the midbrain as the dominant correlation centre, or to the thalamus as the most highly developed mechanism of association, in a certain group of vertebrates. The 'migration of the dominant controlling centre' (Bastian, 1880) appears to involve the stripping of the correlative and associative assets of the fish and amphibian midbrain by the reptilian and higher vertebrate forebrain. The phylogenetic take-over is not thought of as complete until the rise of cerebral cortex, and therefore the midbrain is considered to be a going concern in reptiles and birds, finally going into liquidation only in mammals. As a consequence of the take-over by cortex the formerly dominant centre may be expected to become almost vestigial and in mammals the midbrain is often given credit for little more than a few minor reflexes. On the other hand when the midbrain remains the seat of sensory integration and action the forebrain as a whole ought to be no more than up and coming. It can hardly be said to be vestigial in fish and amphibians, but in these classes the forebrain is sometimes viewed as merely a holding company for olfaction.

The take-over idea is supported by obvious differences in the physical proportions of the brains of the various classes, but it is open to various objections. Weiskrantz (1961) doubts the evolutionary logic of competition for business within reproducing units. If non-mammalian vertebrates were already equipped with efficient midbrain functions, what was the point of shifting these functions to a new location? One obvious answer to this sort of question is that the evolution of brain organisation took the form of adding extra or different capacities,


rather than the migration of executive centres from one location to another. This gives another version of the encephalisation principle—the hypothesis of addition of functions.

Addition of functions

It is reasonable to assume that the elaboration of higher brain levels adds something new in terms of behavioural capacities. This is the root idea of most psychological discussion of phylogenetic comparisons, and is given detailed support by Bass (1968a), Riss et al. (1972), and Diamond and Hall (1969) among others. The cortex in particular ought to allow for better or more associations, correlations, integrations and so on. If there are significant differences in brain organisation between the vertebrate classes then, provided that we test numerous animals to encompass idiosyncratic species specialisations, systematic differences in the psychological abilities of the classes should emerge. This was originally one of the goals of comparative psychology: the problem is that after about a century of experimentation, there is little agreement even on whether such psychological dimensions exist (Hodos and Campbell, 1969; Warren, 1974; Bitterman and Woodward, 1976). Part of the difficulty undoubtedly lies in the naivety of the phylogenetic expectations as exposed by Hodos and Campbell (1969). However, I refuse to abandon the expectation that chimpanzees may be demonstrated to possess psychological capacities which are superior to those of the goldfish. The current failure to distinguish between the abilities of the vertebrate classes rests largely on the adoption of standardised conditioning techniques which may indeed produce much the same outcome when applied to a wide variety of species, and alternative strategies of assessment may be more successful in revealing cross-species differences in brain power. But it is necessary to acknowledge that behavioural evidence suggests that widely disparate species have a good deal of associative, correlative and integrative abilities in common, and this would imply that the doctrine of encephalisation must come to terms with constancies as well as changes in brain function.

Conservation of functions

If one reliable universal principle can be distilled from the study of evolutionary changes in brain and behaviour, it is that there are no


universal principles. However, allowing for numerous exceptions, the sweeping generalisation I shall attempt to defend for the rest of this chapter is that vertebrate phylogenesis reflects conservation of functions, rather than take-overs or additions. The essence of this hypothesis is that all vertebrate brains are built on more or less the same plan, give or take a few special adaptations for particular species and orders. The hypothesis was formulated as the Conservation Principle by Stebbins (1969) and endorsed by Jerison (1973, 1976). It states that the functions of brain divisions and brain pathways are preserved through phylogenesis, rather than altered.

It should be emphasised that the Conservation Principle does not entail the assumption that the brain remained utterly unchanged throughout vertebrate evolution: even if the general role played by a certain brain division is always the same, the importance of that role may change, and it may be performed with a greater or lesser degree of expertise. The importance and effectiveness of particular brain structures should, according to Jerison, vary to conform with two other principles: the Principle of Adaptation—species ought to have evolved brains appropriate to their particular needs (an animal which depends for its survival on the sense of smell should have efficient olfactory apparatus); and the Principle of Proper Mass—the amount of use given to a brain structure should bear some relation to its size (large olfactory bulbs should be better than small ones).

One could choose to believe in the Conservation Principle without knowing what any of the supposedly constant functions of particular brain structures were, but clearly the principle would be more plausible if it was possible to say what it is that is conserved. Claims of this kind are either tentative or trivial, but it is worth listing some presumed constancies. It is not being unduly speculative to say the spinal cord always subserves some local reflexes and transmits information to and from the brain. The brainstem is always the main site of the reticular activating system, and contains nuclei associated with the sensory and motor cranial nerves which go to the head and the face and the viscera. In species which are not blind and deaf, there will be some visual and auditory processing in the midbrain. The midbrain certainly plays a larger part in the processing of visual information in fish and amphibians than in other classes, but the type of visual processing may perhaps be constant across classes—the midbrain appears to be specialised for reacting to movements in the visual field and detecting location in the field, and is said to ask 'where is it?' visual


questions in mammals. In all vertebrates the thalamus receives inputs direct from the eye and contains nuclei which can be characterised as sensory projections in vision and other modalities, from the brainstem and midbrain. It is likely that post-thalamic projections of sensory information are universal: they are certainly not confined to mammals, having been very reliably demonstrated in birds, reptiles and sharks (Nauta and Karten, 1970; Ebbesson, 1980). Certainly, the elaboration and extent of post-thalamic projections varies from class to class, but their functions may show some constancies: visual postthalamic projections, for instance, may contribute to object recognition (answering 'what is it?' visual questions) in all classes. The internal striatum of the telencephalon is generally regarded as furnishing some basic control over movement and action, perhaps being concerned with relatively slow and sustained directions. The cerebellum is always firmly identified with motor co-ordination, and may be especially adapted for the sequencing and timing of movements which need to be put together quickly (Kornhuber, 1974). The limbic system as a whole, or some of its constituent parts, is specifically involved with olfaction, but also appears to exert a regulating effect on instinctive and motivated behaviours, which leads to its identification as the source of instinctual drives and emotions in all vertebrates (Isaacson, 1974).

Given that this list establishes a certain minimal plausibility for the slogan of phylogenetic conservation of function, let us now test the virtues of the conservation idea against the take-over and additive versions of the encephalisation principle by considering more fully two particular issues. First, is there sound evidence of the atrophy of lower levels of brain function as a consequence of the elaboration of higher circuits, as the take-over version implies? And second, does the enlargement of the forebrain, which undoubtedly takes place in higher vertebrates, bring about completely new psychological capabilities, as would be predicted by the additive theory, or does it expand and develop functions which can already be seen to be present in the lower vertebrate forebrain?

Evolutionary development in non-forebrain structures

Encephalisation in its purest form would have it that, with the rise of the forebrain and the formation of its cortical crust, hindbrain and


midbrain foundations should become vestigial evolutionary relicts, or vanish altogether. It is apparent that this did not happen—the question is only whether the hindbrain and midbrain suffer any predations at all from the imperialism of the forebrain, or whether their functions remain fixed, or whether they undergo parallel improvements. We may briefly say that the spinal cord shows considerable change from class to class with the cord of the cyclostomes being particularly rudimentary (Kappers et al., 1936). Elasmobranch fish can be said to possess the prototype spinal cord for other vertebrates with the separation of dorsal afferent and ventral efferent roots, but many fairly obvious new adaptations, such as local enlargements for the control of limbs and tails, occur in amphibians and reptiles. Some of the most peripheral elements of the nervous system, such as the somato-sensory nerve endings which feed into the spinal cord, show a pronounced elaboration in mammals (although birds have relatively little tactile sense, presumably because of the sensory limitations of feathers as opposed to hairs, and have noticeably small dorsal roots in the spinal cord). However, there is some tendency for the spinal cord of higher vertebrates to become simplified, in the sense of having specialisations for single-synapse reflexes, which do not occur in the spinal cords of lower vertebrates.

The brainstem, while retaining its primitive and reflexive characteristics, undergoes progressive evolutionary changes, as may be illustrated by the importance of the facial nerves and their corresponding cranial nuclei in mammals. Chewing, tasting, facial expression and hearing all require the operation of cranial nuclei, and one would presume that the refinement of muscular control in the cranial regions necessary for song in birds and speech in man should be represented in some way in relays in the brainstem. The modality of taste is strong in many teleost fish (which may have taste buds outside the mouth, or covering the whole body) and in mammals, while reptiles and birds apparently have very few taste buds (sometimes only 40-60 in birds, although the gustatory sense is nevertheless a powerful influence on their behaviour—Garcia et al., 1977). The change from a 'lateral line' external pressure sense, and other methods of detecting vibration in fish, to hearing techniques more appropriate for airborne sounds, must have been paralleled by changes in the brainstem, where auditory input enters the brain. Reptiles have no external auditory canal, and lack the complex sound transducing mechanism of levers in the inner ear, but the more sensitive hearing of birds and mammals had more


effect on the growth and specialisation of lower brainstem nuclei than it did on the pattern of auditory connections to higher centres, since the latter show a 'remarkable resemblance' in reptiles to those observed in birds and mammals (Foster and Hall, 1978).

The cerebellum attains healthy, highly fissurised proportions in the larger sharks, and the same cerebellar components, such as Purkinje cells with climbing- fibre and mossy-fibre input to their elaborate dendrites, are thought to be involved in cerebellar activity in all vertebrates (Llinás, 1970). Although the direct connection from the motor cortex of the cerebral hemispheres to the spinal cord is a mammalian speciality and is therefore given much significance, there has never been any suggestion that the cortex has taken over cerebellar functions: if anything, the size and differentiation of the cerebellum increases with the size and differentiation of the cerebral hemispheres, and the cerebellum is thought to show improvements in internal organisation for the purpose of skilled rapid locomotion in the higher vertebrates.

One would be very hard-put, therefore, to support a claim that the hindbrain — the brainstem and cerebellum —degenerates in structural or functional importance as the midbrain and forebrain which surmount it expand. As far as the midbrain goes, no one denies that it retains well- developed functions in birds, and, as will be emphasised when I discuss perception (Chapter 7), midbrain pathways are present in mammals, and even in primates have much more than the marginal role in visual perception that was accorded to them twenty years ago (Humphrey, 1974; Weiskrantz, 1977). The Conservation Principle seems to apply reasonably well to non- forebrain structures — they do not wither away or substantially change their function as a consequence of phylogenetic progression. One could incorporate the stability of hindbrain and midbrain functions into the additive version of encephalisation if the major point of addition was narrowed down to forebrain expansion. It is all very well to have the Conservation Principle apply to the supposedly more ancient brain divisions: but how can it be relevant to later modifications?

Evolutionary conservation and development in the forebrain

In the forebrain the evolution of the thalamus has usually been supposed to have proceeded hand-in- hand with corticalisation, so that


the mammalian 'neo-thalamus' appears only in primates (e.g. Diamond and Hall, 1969). In contrast, the status of the corpus striatum is thought to be very stable. Differences in the complexity and size of the forebrain limbic structures (for instance the hippocampus) are extremely obvious, and are certainly in the direction of progressive development of the limbic system along with the rest of the cerebral hemispheres, but there is still a tendency to regard the limbic system as the 'regulator of the reptilian core brain' (Isaacson, 1974) and therefore primitive.

I put the case in the previous section that, whatever new capacities might have accrued as a result of the development of the cerebral hemispheres, these capacities are not accompanied by underemployment of other brain regions. The higher cerebral attainments occur in conjunction with, and not instead of, other functions of the brain. Given this pointer, how can we characterise cerebral, or cortical activity? The obvious, usual, and possibly correct rule of thumb is that the cerebrum is the seat of mentation—of contemplation, inference, reason, volition, anticipation, foresight and cognition generally. There are many variations on this theme. The crux of any treatment of cognition in animals is of course the apparently qualitative difference between human faculties and those of all other beasts. This is frequently dealt with by adopting some form or other of the human cortex hypothesis: the assumption that the human cerebral cortex, or some portion of it, is so superior in its quality or quantity to all other mammalian cortices, that virtually no mentation or thought is conceivable in any brain except our own. A sophisticated Cartesian version of this hypothesis is the one put forward by Eccles, who proposes that a small and as-yet-unidentified population of neurons in the cortex of the human left cerebral hemisphere is in communication with a disembodied and reasoning soul. Many less explicitly Cartesian theories attribute unique properties to one or both of the human hemispheres. Less radical proposals identify either the cortex of primates, or the association cortex of all mammals, with cognitive functions. Anatomically the strongest case is that the mammalian neocortex in general differs so much from any other kind of vertebrate brain tissue that mammals should be set apart from all other classes. The difficulty in this case is that psychological tests of cognitive capacity do not clearly separate mammals from vertebrates lacking a neocortex, if the latter happen to be birds. Evidence from behavioural experiments, reviewed in later chapters, supports a rather more


general anatomical suggestion, the post-thalamic hypothesis: postthalamic forebrain circuits, whether cortex or not, mediate the more cognitive of psychological capacities. Since post-thalamic circuits are more pronounced in higher vertebrates, both birds and mammals, and take up an increasing proportion of tissue in large mammalian brains, the usual expectation of greater cognition in primates and other large-brained mammals is thus retained. But anatomical and physiological evidence now suggests that post-thalamic sensory projections are a small but significant reality in the forebrains of lower vertebrates. The post-thalamic hypothesis therefore implies, as an aspect of the Conservation Principle, that some degree of cognition is a vertebrate universal. At the moment this has an extremist and unlikely ring to it, but an examination of some of the relevant data may make it seem somewhat less outlandish.


Forebrain and cognition in non-mammals

It is appropriate to begin with Flourens (1794-1861), who may be regarded as one of the earliest supporters of the post-thalamic hypothesis. In a series of clear and simple experiments, Flourens surgically removed the cerebral hemispheres of frogs, pigeons and chickens, and made careful observations of their behaviour after they had recovered from the operation. In some cases he cut successive slices from the hemispheres, in separate operations. His conclusion was that in these animals the cerebral hemispheres are the organ of 'feeling, willing and perceiving'. A frog with no forebrain he described as having 'lost all volition', and a chicken with a similar deficit could be kept alive, and could stand up and move about, but had to be hand fed (Flourens, 1960). By experiments of this kind, Flourens established the traditional functional divisions in the vertebrate brain, identifying the brainstem as the centre for vital and visceral reflexes, the cerebellum as the organ of balance and movement, and the cerebrum as the seat of intelligence and will.

However, Flourens's contention that the cerebral hemispheres in amphibians and birds control volition and cognition has often been rejected. There is more than a suspicion that the rejection derived from phylogenetic dogmas, rather than contradictory evidence, but two empirical generalisations have seemed to go against Flourens. First, many have claimed that the removal of all or part of the forebrain has a negligible effect on the behaviour of non- mammals; and second, it is argued that such behavioural effects as do occur can be interpreted in


terms of the role of the forebrain in olfaction (Aronson, 1970; Ebner, 1976). Underlying these points is the assumption that non-mammals do not possess anything approximating to volition or intelligence, and therefore have no need of an organ which subserves these faculties.

It is not too much of a caricature to present the following as the source of considerable confusion in theories of forebrain function:

The forebrain is of little importance in this species. Animals with extensive forebrain damage can be kept alive indefinitely and show locomotion in response to tactile stimulation. Several workers have shown that classically conditioned reflexes can be established in forebrain-ablated subjects, and therefore the forebrain is not necessary for associative learning in this species. The domination of social behaviour by olfaction in this species is illustrated by the fact that forebrain-ablated subjects show no signs of aggression, reproduction, or other instinctive social behaviours.

The disparity in this invented paragraph is between the retention of limited locomotor and sensory abilities and the capacity to perform at a satisfactory level on standard laboratory tests of conditioning, and the complete disruption of behaviour as observed from the standpoint of clinical or naturalistic assessments. The varying inferences which may be drawn from performance in laboratory tests of conditioning were discussed in Chapter 3: the main point is that evidence of 'associative learning' is obtainable at very low levels of nervous system organisation, if a simple enough experimental procedure is used. With more demanding tasks, deficits in the learning of both teleost fish (Farr and Savage, 1978) and sharks (Graebner et al., 1978) follow forebrain damage. There is sufficient, though limited evidence to suggest that some forms of learning are impaired by lesions of the cerebral hemispheres in amphibians and reptiles (Aronson, 1970), and quite small localised lesions of the hemispheres of birds produce reliable deficits in certain kinds of acquired discriminations (Macphail, 1975; Stettner, 1974).

If volition is measured as the occurrence of spontaneous actions, then there has been widespread confirmation of Flourens's finding that damage to the forebrain reduces volition, since spontaneous action is lost after removal of the cerebral hemispheres in amphibians, reptiles and birds (Aronson, 1970; Ebner, 1976). A forebrain-ablated bird


may 'fly' if thrown into the air, but unless some similarly drastic goad is imposed on it from without, it will just sit still with ruffled feathers. Forebrainless frogs may snap at passing insects, but without eating the target if it is caught, even though the swallowing reflexes are still intact. Coupled with the almost complete absence of social and reproductive behaviour, these findings ought to be sufficient to illustrate the biological necessity of a functioning forebrain for these species.

The alternative suggestion that the forebrain in lower vertebrates is used exclusively for olfaction is now no longer very plausible. Anatomical and physiological data show that in teleost fish (Scalia and Ebbesson, 1971), sharks (Ebbesson and Heimer, 1970), amphibians (Vesselkin et al., 1971) and reptiles (Ebner, 1976) olfactory projections are more limited than had previously been assumed, and on the other hand visual and auditory projections to the telencephalon appear to be available. Even if the cerebral hemispheres were used only for olfaction, there would not be much point in it if there were not outputs to other brain divisions. But since there are specific groups of motor cells in the hemispheres that put out efferent fibres which reach midbrain and hindbrain motor systems, and which may go as far as the spinal cord in the case of sharks (Ebbesson and Northcutt, 1976) and also in birds (Adamo, 1967), there is support for the interpretation that the hemispheres of lower vertebrates are involved directly in the initiation of action.

We can say fairly firmly then that the forebrain in lower vertebrates ought not to be dismissed as merely an extension of the olfactory bulbs but may supply, albeit on a very limited scale, the same sort of facilities for behavioural control as are provided by the cerebral hemispheres of mammals. But what are these facilities? In a very general way, it may do no harm to retain the old characterisations of perception, volition and intelligence, not because these are entirely satisfactory, but because attempts to go beyond these intuitive categories have not yet added very much to the conclusion reached by Flourens:

Animals without cerebral lobes have therefore no perception, no judgement, no memory, no volition, because there is no Volition when there is no judgement, no judgement when there is no memory, and no memory when there is no perception. (Flourens, 1824; see Boring, 1950, p. 77 ff.)

This is in accord with Flourens' belief in the unitary action of the


hemispheres, but various distinctions between the functions of forebrain components are possible. For some purposes, it is useful to draw lines between perceiving and knowing, localised in the sensory thalamus and the post- thalamic projections, feeling and wanting, supposedly the main business of the limbic system, and acting, mediated by both the internal striatum and the higher motor centres of the external striatum or cortex. Each of these could be analysed in greater detail into several substages, even on the basis of current knowledge, but if forebrain function can be given any overall defining features it is probably in terms implying values, perception, memory and action. One way of incorporating these things into a single package is to speak of goals. Nauta and Karten (1970), for instance, see vertebrate forebrain function as 'the perception of goals and goal priorities, as well as the patterning of behavioural strategies serving the pursuit of these goals'. To narrow down some of the ways in which the vertebrate forebrain may be a very sophisticated kind of goal-seeking device, we may consider intermodal abstraction, selective information pick- up memory, anticipation and choice of action. I will discuss perception and memory in some detail in later chapters but give now some briefer comments.


Crossmodal abstraction

Especially in higher vertebrates, the independence of an animal's actions from specific individual cues is often apparent—a goal can be identified by either taste, or smell, or visual pattern, or spatial position, or sound, or touch. Experiments with rats performed with the intention of finding the single crucial cue that controls behaviours have shown that, for instance, the identification of a female by a male can occur in any sense modality — males deprived successively of each sense continue to attempt to mate a female as long as any sense modality at all remains (Beach, 1942, 1947). Similarly, rats deprived of vision, smell, or muscular feeling will negotiate a maze successfully with whatever information is left to them (Honzik, 1936; Hunter, 1940). When a rat encounters a novel object it will be seen, sniffed, whiskered and listened to—subsequent identification may be based on any or some of these modalities.

Many aspects of sensory-motor co-ordination, such as the maintenance of visual fixations during movements of the head and body, are definitely wired in to the midbrain and hindbrain; but the identification of an object independently via information from alternative modalities may require forebrain intervention. Indeed, until


recently it tended to be assumed that the transfer of sensory qualities between modalities, as when something that feels round to the touch is also expected to look round when it can be seen, was unique to man (Geschwind, 1965). Monkeys and apes have, however, also been persuaded to exhibit competence at such tasks (Weiskrantz, 1977) and specific cross-modal associations, as between the sight and sound of predators or between the smell and sight of prey, would be of great use to other animals. In mammals, the mechanism for making cross-modal abstractions or associations has been identified as cortex which is not closely tied to a particular sense— 'association cortex'—and larger brains have a greater proportion of this available (Penfield, 1966), but behavioural data to show whether and how cross-modal abilities vary with brain size are almost wholly lacking.


Selective attention

Being relatively remote from the peripheral sensory pathways, forebrain structures are in a prime position to filter out relevant details from the 'blooming, buzzing confusion' of sensory bombardment from the environment. Overall amount of reaction to stimulation is probably governed by the reticular activating system (pp. 148) which specifically does not belong exclusively to the forebrain. But valuing, searching for and picking out certain details within a given sensory modality is widely supposed to be one of the higher aspects of perception and cognition (Neisser, 1967, 1976). There is no shortage of explicit theories as to how selective attention might work in animal behaviour (Sutherland and Mackintosh, 1971; Andrew, 1976), and there is a considerable body of experimental data available for review (see Chapter 7). Although selective attention is not necessarily only a forebrain function, it seems likely that switches of attention are very much a part of the interactions between the limbic system and the thalamic and post-thalamic stages of sensory pathways.


Memory and anticipation

It should not be difficult to find agreement that memory for perceptions and actions, and anticipations of future events, are more likely to be dealt with in the cerebral hemispheres than in the spinal cord. The difficulty is in measuring memory and anticipation in animals with any confidence. Given the possibility of verbal formulations, we have very little doubt about our own memories and future plans, and very elaborate (although perhaps very irrelevant — Neisser, 1976) experimental techniques can be employed


for the analysis of human memory (Brown, 1976; Baddeley, 1976). But, although the more anthropomorphic among us may incorporate memories and expectancies into our theories of animal behaviour (Tolman, 1932, 1949), it is surprisingly difficult to produce convincing experimental proof that any animal has any memory or foresight at all. A major advance, of course, is the development of gesture language training for apes, which allows these animals to give fairly firm indications of what they want next, though they do not appear to have much interest in recounting what they did last. And some quite simple modifications to traditional laboratory tests seem to allow the measurement of a sort of short-term memory for stimuli in birds and mammals (Medin et al., 1976; see Chapter 8 below).


Choice of action

Instead of manifesting robot-like certainty, many animals give a reasonable imitation of being individuals in whom 'nothing is so habitual as indecision'. Reluctance to plump for one action over another is apparent both in natural behaviour (where it deservedly receives much attention from ethologists under the heading of 'conflict': Hinde, 1970), and in laboratory tests, where the latency of response may sometimes provide a useful behavioural index, but at other times be a considerable annoyance to the experimenter. McFarland (1973, 1976) has emphasised that the ordering and timing of numerous incompatible but necessary activities is an ever-present problem of behavioural organisation and has suggested that some positive momentum for any action already under way is essential to prevent continuous dithering between possible responses. Categories of natural and artificial decisions range from the sequencing of foraging, resting, grooming and courting to the choice between a left and right turn in a laboratory maze. A fundamental conflict is that between caution and risk, with 'fight or flight' as an example. Fear of danger or of the unfamiliar is often finely balanced against curiosity and physical needs.

Forebrain involvement in these decisions is indicated by the results of lesion experiments, especially when limbic structures are damaged. A very broad description of 'lack of inhibition' can often be applied to mammals with limbic lesions, and there is some basis for more particular speculations. For example, the amygdala appears weighted to promote both caution and aggression. Damage to this area in primates results in tameness, passivity and a marked lack of discretion


in feeding and sexual activities (Kluver and Bucy, 1937). A not dissimilar syndrome is observed after equivalent forebrain lesions in Mallard ducks (Phillips, 1964). On the other hand septal area lesions produce a touchy, sensitive and occasionally vicious animal in the normally amenable laboratory rat (Grossman, 1967). It would be foolish to draw detailed conclusions from this, but given the anatomical relationships within the limbic system one is bound to entertain the idea that the amygdala and the septal area act in tandem, and have opposing effects on mood.

Clearly, mood and emotion in animals should be closely related to social behaviour. Even the most rudimentary social behaviours, such as flocking and schooling, require subtle co-ordination of the individual with the group, and in teleost fish and reptiles, as well as birds and mammals, social interactions in general can be said to be notable casualties of damage to the telencephalon (Aronson, 1970; Phillips, 1964; Tarr, 1977; Rosvold et al., 1954). The dimensions of social choice engendered by communal living have been put forward by some as the driving force behind the evolution of the intellectual powers of the cerebral hemispheres of primates (Humphrey, 1976), and while there are problems in• this argument, is useful to bear in mind that the regulation of social interactions may be one of the more demanding tasks performed by the vertebrate central nervous system.

In any context, a fundamental question about the initiation of action is the extent to which the selection of an act is governed by the anticipation of its consequences. This is another case where plausible theories are difficult to pin down in terms of behavioural predictions—how are we to distinguish between an animal governed by useful reflexes and one responding to intelligent expectations? Not very easily, but there is a certain face validity to the argument that being able to anticipate the consequences of actions should have considerable survival value. While it is evident that the anticipation of consequences has proved to be an elusive form of animal cognition, it is certainly something we should look for in the cerebral hemispheres rather than elsewhere.

Evolution of the vertebrate brain—conclusions

Evolution means nothing if not changes in the complexity of biological systems over time, which account for the development of the enormous


variety of plant and animal species in terms of their genesis in earlier and simpler forms. The translation of the notion of increasing complexity into animal psychology has, however, produced many errors and misconceptions. Animal species that we may now observe, and whose brain function we wish to explain, do not fall directly into a scale of brain excellence. First, the family relationships between orders and classes of species is rather involved, with numerous branchings: deciding which species are more primitive, or more advanced, than others, is very difficult. Some kind of classification of animals cannot be avoided, of course, and the obvious starting point for theories of brain evolution is that the earliest vertebrates were fish, and were more like lampreys than they were like any other surviving species of fish. Amphibians derived from an intermediate stage of fish evolution, and reptiles derived from early amphibians. All these are 'lower vertebrates': birds and mammals, which arose from different branches of reptiles, are 'higher'. By and large, the brains of higher vertebrates are much bigger than the brains of lower vertebrates, and are therefore, perhaps, better.

Irrespective of size, all vertebrate brains can be described by reference to the same component parts, most simply in terms of the hindbrain, midbrain and forebrain divisions. There has been a tendency to say that the hindbrain is very old and primitive, the midbrain is intermediate, and the forebrain is modern and advanced. It is true that the forebrain of lower vertebrates is relatively small, and that in the larger brains of higher vertebrates, it is the forebrain which has expanded most in size. This is most obvious in man, whose brain is extremely large, and physically dominated by the cerebral hemispheres of the forebrain. It thus seems to follow that the forebrain is the most important part of the brain for intelligence, thought and cognition. However, I have supported the idea that the functions of the different divisions of the brain are more or less the same in all vertebrates, and that therefore we should expect these higher psychological functions to be present roughly in proportion to how well the forebrain seems to be developed. A fish, frog, or lizard does not have very much by way of cerebral hemispheres, and should therefore have very little cognition. But this little is a much more generous allowance than that given by other theories, which say that such forebrain as is present in non-mammals is not used for cognition, or attention, or learning, at all, but is only there to respond to smells, or to program primitive instincts. These other theories are probably wrong,


because the cerebral hemispheres of lower vertebrates share several anatomical and physiological characteristics with the cerebral hemispheres of mammals, and the hemispheres of lower vertebrates should therefore have some psychological functions in common with the forebrain of mammals as well. What these functions are exactly is difficult to say, but emotions, plans for action, goals and values, and the selective direction of perception, seem to require the operation of the higher levels of the brain. Direct evidence about these functions, in terms of how animals normally behave, and what their capabilities are in psychological tests, is not easy to produce, but I shall go on to discuss these sorts of evidence in later chapters.