['Animal Thought' © Stephen Walker 1983]
115
gif pdf file of this chapter

4   The phylogenetic scale, brain size and brain cells

Even if one is convinced that all mental processes, whether in man or in animals, depend on the activities of the brain, it does not follow that all psychological investigations can be replaced by the examination of brain tissue. However, if the assumption that mental states reflect brain-states is taken at all seriously, then it is clear that the physical structure of animal brains must be taken into account in any theory about the existence of mental activity in animal species. In this chapter I will therefore examine the evidence concerning the anatomical and physical characteristics of vertebrate brains. Inevitably, some of this evidence is rather technical and detailed. But first we should consider the broad and general question of the physical evolution of the brain. A concept which has the immediate appeal of simplicity is that vertebrate brains can be placed on an evolutionary scale of excellence, with the human brain coming at the top of the scale as the end result of a long series of brain improvements. There are many dangers and difficulties in this idea of a phylogenetic scale, but clearly we should expect to place physical brain characteristics within an evolutionary context, and I shall start with a brief account of vertebrate evolution. This will, I hope, clear the way for an examination of two specific issues: the nature of the variations in the quantity and the quality of brain tissue across vertebrate species. The underlying question is of course the extent to which animal brains are similar to the human brain. If the human brain is quite radically different from the brains of all other animals in its physical characteristics, then one might conclude that human mental processes could not possibly have counterparts in other species. But if, as I think the evidence will suggest, the physical superiorities of the human brain are a matter of degree, then one may argue backwards, and say shared

116

brain processes imply shared mental processes. This argument does not depend on any particular course of vertebrate phylogenesis, or even on evolutionary theory, but of course we would expect species differences in brain anatomy and physiology to reflect evolutionary relationships.

The phylogenetic scale

Evolutionary theory is often vague and indefinite, but it can hardly be ignored if we are to make systematic comparisons of any kind between species. The most controversial and dubious aspects of the Darwinian theory of natural selection have to do with gradual progression and its causes, and many of the assumptions about these aspects made in discussions of psychological differences between species can be severely criticised (Hodos and Campbell, 1969). Part of the explanatory value of Darwinian theory is the concept of a gradual step-by- step sequence of design improvements, with superior designs supplanting the inferior (Yarczower and Hazlett, 1977). But we cannot use this idea to rank species on a single ladder of ascending complexity. Hodos and Campbell argue that there can never be any valid classification of species based on a linear grading, and say that the use of terms such as 'the phylogenetic scale', 'the Great Chain of Being' or 'Scala Naturae' has been grossly misleading. They would prefer to abandon even such familiar dichotomies as that between higher and lower vertebrates. This would be an uncomfortable abstinence in comparative psychology since the framework of progressive increases in intelligence, brain size, or learning ability is a very common assumption. It is very hard to avoid dealing with the proposition that animals in general have cognitive abilities that are inferior to those of the human species, and that some animals have less cognition than others. In itself this implies a scale of psychological complexity, but it is difficult, if not impossible, to give such a scale any biological meaning. The most obvious attempt to give evolutionary backing to intuitive ideas of progress in psychological complexity is in terms of an historical account of vertebrate ancestry. The three main objections to this are: (1) accounts of phylogenetic sequence in vertebrate evolution are highly speculative; (2) convergent and parallel evolution resulting from similar ecological specialisations in unrelated groups mean that ancestry is not always important; and (3) what is known about the evolutionary tree of vertebrate development cannot possibly be mapped on to a single linear scale.

117
gif

118

Given these and other criticisms of the phylogenetic scale we should be wary of all general theories of vertebrate brain evolution. But the extent of the disparity between species genealogy and the assumption of an ordered scale, as well as the give and take between ancestry and ecological specialisation, deserves closer analysis.

Sequence and relations in vertebrate ancestry

It would be wonderfully convenient for the purpose of comparative experiments if vertebrate phylogeny had proceeded in a strict linear succession. The story has sometimes been presented almost as if it did—as if a swimming chordate with no jaw developed a minimal brain and became a lamprey, which survives as the oldest vertebrate but which was superseded by animals with jaws and a bony skeleton in a series of improvements by which fish changed into frogs, frogs turned into lizards, lizards turned into birds, birds turned into rats, rats turned into monkeys, and monkeys turned into man. If things had happened like this, then we would have some reason to expect that we should be able to plot increasing levels of psychological complexity by making comparisons between living representatives of each of the stages. But in fact, the family relationships between living species are rather involved and complicated. What can be done, with a considerable amount of speculation, is to chart roughly the evolutionary tree of the vertebrates, to show relationships between taxonomic divisions above the species level such as classes (for instance birds and mammals) and orders (for instance rodents and primates). A diagram of this kind is shown in Figure 1. Rather than providing a linear scale of vertebrate species it can be seen that comparisons between living animals have to be judged against a variegated pattern of phylogenetic relatedness, recency of origin, and species specialisation. In other words the vertical axis of Figure 1 does not represent a single biological dimension, although it is certainly true that species towards the top are more closely related to man than species near the bottom.

To go back to the beginning of the story, the earliest vertebrate fossils are indeed of jawless fishes, and there are surviving jawless fishes in the lampreys and hagfishes (cyclostomes). Lampreys and hagfishes are therefore conveniently called primitive. But their lifestyle is highly specialised: they are parasites and scavengers depending on much

119

more recently evolved fishes. The habits and instincts of the living parasitic cyclostomes cannot therefore be representative of the original, and now extinct, 'primitive vertebrates'. (There might have been a short and drastic period of mutual cannibalism, but things could hardly have started that way.) We must do the best we can with whatever evolutionary left-overs are available, and therefore we assume that the brains of lampreys and hagfishes are more like primitive vertebrate brains than are those of other species of fish. To support this assumption, we may appeal to the necessarily vague 'principle of conservation', which says that essential features of organisation and structure, once evolved in a group of animals, will be retained in all the direct descendants of the group (Stebbins, 1969; Jerison, 1976). But there is always the danger that modern lampreys and hagfishes have developed their own specialisations, and that this makes them differ significantly from the original jawless fish that they are presumed to represent.

The earliest jawed fish provide the next most ancient branch of the vertebrate tree, and this branch divided, about 500 million years ago, into fish whose descendants now have skeletons composed of cartilage, and those which gave rise to the numerous living tribes of bony fish. Although many changes may have taken place in 500 million years, modern cartilaginous fish (sharks and rays) are usually considered to represent very old vertebrate stock. The vast majority of modern species of fish evolved in several stages from primitive bony fish through a line in which isolated examples, such as the garpike and sturgeon, remain from otherwise extinct radiations. The modern bony majority, the teleosts (e.g. goldfish, cod, perch) come from a line of ray-finned fish (actinopterygians) which diverged more that 400 million years ago from the fleshy-finned orders (sarcopterygians) whose more substantial appendages provided the basis for the limbs of amphibians, and thus for the limbs of all other land-going vertebrates. It is fair, then, to regard the teleosts as phylogenetically very remote from birds and mammals, even though the teleosts did not become common until the Cretaceous period (about 100 million years ago), and in this sense are almost as recent as birds and mammals.

Amphibians as a class, arriving about 350 million years ago, are clearly ancestral to later land-based vertebrates; but the common modern amphibians, the frogs and toads, are highly specialised as Jumpers and insect catchers and cannot be taken as early kinds of reptile. The early reptiles themselves, having come into the possession

120

of a mode of reproduction which allowed them to exploit purely land dwelling lifestyles (as their eggs could be laid on land), diverged into both mammals and birds, and surviving forms such as lizards and snakes. In a typical evolutionary quirk reptile groups which remained loyal to sub- aquatic environments, the turtles and crocodiles, are among the survivors.

The fossil record for birds is poor, and as a class they are sometimes treated as 'glorified reptiles', rather unjustly in view of what I shall say later about their psychological capacities. Bipedal, highly visual reptiles developed feathers some 50 million years after quadrupedal, nocturnal, smell and hearing specialists grew hair, and thus birds could be regarded as more recent than mammals. Intermediary species which can be classified as 'mammal-like reptiles', then 'reptile like mammals', provide fossil evidence for the connection between reptiles and mammals, and living species which are different again from these seem to have part- mammal and part-reptile characteristics. The monotremes—the duck-billed platypus and two species of spiny anteater—lay eggs but suckle their young, and one of the spiny anteaters carries its eggs about in an external pouch (Tachyglossus: Grifflths, 1978). The more numerous marsupials (e.g. the opossum or the koala bear) suckle their young, and do not lay eggs, but, lacking a placenta for the internal nourishment of their offspring, give birth very early in foetal development. One reason for regarding this as an intermediate level of reproductive strategy is that marsupials were once very much more widespread, but were superseded by placental mammals.

The ancestral placental mammals were rather unimpressive, small, shy, nocturnal insectivores, according to the usual interpretation of fossil evidence. If we are to believe this, the hedgehog is closer to the typical primitive placental mammal than any living species is to a primitive amphibian, reptile or bird. Unflatteringly, it is the primitive insectivores, represented by the hedgehog, which are the ancestors of the modern primates (which include monkeys, apes and man). The early branching off within mammals was fan-like, rather than step-by-step, and there is therefore no question of a rat or a cat representing an intermediate stage along a line stretching from hedgehogs to monkeys. The rapid fan-like branching out of mammalian orders, 50—60 million years ago, meant that rodents, carnivores, insectivores, primates and so on followed separate lines of development (see Figure 1, p. 117). Although mammalian species fall within families, and the families fall

121

within orders, descent alone cannot be used to put them into a natural hierarchy. We may put primates at the top of the mammalian scale because we are primates, but no one has discovered any other reason for singling out this particular branch of the phylogenetic tree. In particular, recency of origin of a species or group, in so far as this can be guessed at, has very little relevance. For instance, rodents first appear in the fossil record rather later than primates, and may be derived from primates, and Old World rats and mice (murids) are a particularly recent family, first appearing only 10 million years ago. Recency in the fossil record does not allow one to put Old World monkeys above Old World rats (Radinsky, c1976). Because of this lack of phylogenetic distinction between the various orders of mammals, and because of the uncertainty and irregularity of vertebrate evolution which led to mammals, we must accept the caution that a continuum of living species with man at the top is an arbitrary and unfounded invention (Hodos and Campbell, 1969). But, although vertebrate evolution cannot provide a neat and linear scale of superiority, we are not left with only confusion and anarchy in phylogenesis. It is illegitimate to regard animal species as small steps made towards the human species, but there is no reason why we should not work backwards, and construct an explicitly anthropomorphic scale in terms of phylogenetic relatedness to man (or even in terms of superficial similarity to man). If we simply wish to classify other species according to how close they are to our own species in the phylogenetic tree, there is no great problem in identifying other primates as more like Homo sapiens than other mammals, or in saying that mammals in general should share more human characteristics than other classes of vertebrate. By being explicitly anthropomorphic we can provide some justification for the use of a natural scale.

A second and quite different reason for not completely discarding the notion of a natural scale is that the theory of progression in evolution, or anagenesis, still carries a certain amount of weight (Rensch, 1959; Jerison, 1973; Gould, 1976; Yarczower and Hazlett, 1977). It is not unreasonable to suppose that, other things being equal, evolution via natural selection should lead to increases in the complexity, variety and efficiency of life-forms. The question is whether such general theoretical changes can be related to specific issues such as the increase in the size of the brain during vertebrate evolution without the commission of the vitalist and teleological errors discussed by Monod (1972) and Hodos and Campbell (1969). I shall

122

come back to the issue of progressive increases in brain size shortly, and theories of evolutionary changes in the complexity of brain organisation come up in the next chapter: in both cases there are obvious connections with arguments about animal cognition.

For the moment, if we simply assume that thought and cognition are special characteristics of the human species, Figure I provides a rough guide for estimating genealogical relationships with other vertebrates. These may or may not give a useful indication of which species are most likely to exhibit approximations to human psychological qualities. But in asking questions about thought, one is inevitably being anthropomorphic, and the first hypotheses would be that chimpanzee cognition is more similar to human cognition than is that of other mammals, mammalian brain function is more like human brain function than is reptilian brain function, and so on. Purely on the basis of ancestry, human characteristics should be traceable through primates and insectivores, to primitive mammals, and then back to reptiles, thus bypassing all other mammalian orders and all birds. One problem with this is the enforced absence of the most interesting members of the sequence: primitive hominids, early anthropoids, primitive mammals, and early, non-specialised reptiles. The strategy usually recommended for minimising this problem is to study species which come as close as possible to the supposed lines of descent, that is to compare man with chimpanzees, then monkeys, then hedgehogs, then opossums and then turtles (turtles being optimistically considered to represent primitive reptiles —Riss, 1968a; see also Diamond and Hall, 1969; Hall and Ebner, 1970; Ebbesson and Northcutt, 1976). However, this strategy has rarely been followed by psychologists and, instead, vast amounts of the available evidence from behavioural experiments come from domesticated species of teleost (goldfish), bird (pigeon), rodent (rat) and mammalian carnivore (cats and dogs).

There are compensations in adopting a resolutely non- phylogenetic strategy of comparisons, since even the most sensible assumptions about ancestry are still subject to the difficulties arising from convergence, and ecological specialisation. The general pattern of vertebrate phylogeny necessarily points to relations between broad groups or classes of species, there being important common attributes within each group. Thus one emphasises the breathing of amphibians, the terrestrial egg of the reptiles, the feathers of birds, and so on. But an important aspect of evolutionary theory not represented in diagrams

123

of phylogeny such as Figure 1 (p. 117) is the adaptive radiation of particular groups into different habitats. A factor which arises from this is convergence, which means that species with very different ancestries may independently evolve similar characteristics: the usual example is the streamlined shape which appears in fast-swimming sharks and teleosts, and in marine reptiles and marine mammals. On the other hand, species with a good deal of common ancestry may adopt radical specialisations, so that within the same group there may be many different forms. Within mammals, for instance, there are aquatic forms, amphibious forms, ground feeders, arboreal feeders, insectivores and carnivores, among others, all with specialised anatomical features. Even within orders, similar species may share some features because of inheritance from 'common ancestral stock', and other features because of convergence, while being separated by idiosyncratic species specialisations (Martin, 1973). Shared features due to common ancestry are said to be homologous, but shared characteristics resulting from convergence are distinguished from these as analogous (Simpson, 1953). The decision as to whether a particular feature is shared by two species because of analogy or because of homology is not always clear cut, but if species behaviour as well as anatomical structure may be determined by specialisations which override phylogenetic factors, then great care must be taken with the psychological implications of any purely genealogical comparisons. Hodos and Campbell (1969) make the case that psychology may reflect the current lifestyle of a species, rather than its pedigree.

In the extreme, this implies that general questions based on phylogeny, such as whether the learning abilities of fish differ from those of birds, may be completely misleading. A goldfish may be representative only of goldfish and not of teleost fish as a group, and the pigeon may have evolved behavioural characteristics which serve as a poor guide to the traits of other birds. However, there is a theoretical alternative to the uninviting prospect of treating each species as a law unto itself. This is to follow for ecological classifications as well as, rather than instead of, phylogenetic taxonomy. (In Julian Huxley's terms, we can classify species into clades of ecological specialisation as well as into phylogenetic grades: Gould, i 976.) There may certainly be striking analogies between the psychological capacities of species, based not on common ancestry but on common behavioural functions. An example is the vision of owls and cats. Phylogenetically these families are from different classes, but both are composed of

124

carnivorous predators, and include species that can be said to have specialised as nocturnal mouse catchers. Such owls and such cats have binocular vision which is remarkably effective at low levels of illumination, and which seems to involve similar brain mechanisms (Pettigrew and Konishi, 1976). One may speculate that the anatomical analogies in perceptual apparatus should be accompanied by psychological analogies in the ways in which the apparatus is used, and that stealth, patience and rapid attack should be required of mouse catchers, whether they are owls, or cats. In general, just as species in different classes may arrive at similar anatomical solutions to problems associated with particular ecological niches, so there may be some degree of equivalence in the psychological capacities of such species. But an owl is still a bird, and a cat is still a mammal: ecological specialisations do not necessarily abolish all phylogenetic differences.

Thus, although we might circumvent all the difficulties of genealogy by adopting the principle 'ecology overrides phylogeny', this would be going too far. Phylogenetic position clearly limits the available ecological options (only warm-blooded vertebrates could hunt mice on cold winter nights) and the fact that there may be analogies of behavioural function does not mean that ancestry is always irrelevant. In fact, the analysis of the diversification of species into ecological specialisations can be used to support the idea that a particular ancestry brings with it a flexible potential, rather than only a fixed and static species-specific inheritance. The classic example of a natural experiment which illustrates the potential for adaptive radiation into varied ecological niches is the case of the Galapagos finches. The Galapagos are volcanic islands, only about 20 or 30 million years old, which have been colonised by chance arrivals from South America, which is several hundred miles to the east. Several species of the birds observed on the Galapagos are clearly recognisable as South American, and some may be very recent immigrants, but there are fourteen unique species of finches. Although the presumed ancestor of all these fourteen species was a seed-eating ground finch, there is now a wide divergence in their eating habits (Lack, 1947; Young, 1962). Five species remain ground feeders, although two in this category feed predominantly on cactus. But several species have become insectivorous, and this dramatically violates normal finch behaviour patterns. The convenient anatomical measure of divergence in the Galapagos finches is the shape of the bill, but presumably behavioural adaptations to suit warbler-like and tit-like habits have become

125

necessary in the species with warbler-like and tit- like bills. And although one would have thought that variation in bill growth should be the simplest and quickest mechanism of adaptive radiation, the insectivorous Galapagos finch that mimics woodpeckers (Camarhynchus pallidus) is a major star of behavioural evolution, since instead of having a long woodpecker-like bill, it uses its short finch's bill to hold cactus spines, with which it picks out insects from crevices in trees.

From the point of view of Darwinian theory, the Galapagos finches are important because the existence of fourteen new species in geographic isolation suggests that their origin should be attributed to the natural selection exerted when a seed-eating ground finch found itself in novel conditions of life. The example could also be taken as an illustration of ecology overriding phylogeny, since the ancestral seed eating has been displaced by cactus-eating or insect-eating in some of the descendants. But the Galapagos finches can be used equally well to emphasise that phylogeny is more than a reflection of ecology, in the sense that the original seed-eating species was not eternally bound to its seed-eating habits, but possessed the potential for adaptive radiation into the fourteen new ecological niches. Critics of phylogenetic comparisons, such as Hodos and Campbell (1969), imply that each species is bound to a particular ecology and that a single species should thus never be taken as a representative of its family or class. But the possibility of adaptive radiation means that each species must represent something more than its own ecological specialisation. It was because the ancestral Galapagos finch was a bird that it could be the progenitor of other species of bird, and it was because it had the brain of a bird that its descendants could develop new ranges of bird behaviours.

There is thus a case for saying that every species must to some extent be representative of its class and grade, and that to this extent comparisons between individual species in different vertebrate groupings should not always be dismissed. More generally, phylogenetic classifications of some kind or other are essential in any discussion of brain evolution.

The physical characteristics of vertebrate brains

Although animal behaviour and its relation to evolution may be studied without reference to brain mechanisms, as it typically is by

126

ethologists (Eibl-Eibesfeldt, 1970; this methodology is explicitly advocated by Dawkins, 1976a), there is a long tradition in comparative brain anatomy and physiology which deserves to be recognised and which currently seems to be undergoing a minor revolution (Cajal, 1911; Kappers, Huber and Crosby, 1936; Elliot Smith, 1910; Herrick, 1948; Hodos and Karten, 1970; Ebbesson, 1980). Part of the reason for the changing of views lies in technical advances, and it is instructive to consider the sorts of evidence accumulated by these brain sciences.

Techniques of brain study

Evidence about the brain itself includes how much it, and its constituent parts, weigh; what the individual cells in various brain regions look like; where the axons of these cells go to; and the extent to which brain cells respond electrically when the sense organs are experimentally activated or when the animal is doing things of its own accord. Conversely brain cells themselves can be electrically stimulated, to see what behaviour this produces, or groups of cells can be damaged and subsequent changes in psychological capacities assessed. Slightly more detail concerning these techniques is given below.

 

Gross anatomy

The more directly observable physical features of the brain, such as overall size and shape, have been responsible for some of the most far-reaching speculations about brain function and importance. Total weight or volume of the brain provides an objective and unambiguous scale and indeed cranial volume in cubic centimetres continues to be used as the major index in the study of human evolution based on fossil skulls (Pilbeam, 1972). Considerable emphasis has also been given to the relative mass of different parts of the brain, especially in comparisons between 'higher' and 'primitive' structures, such as the ratio of cerebral hemisphere size to brainstem size (Portmann and Singling, 1961, in birds; Passingham, 1975, in primates: see below and Figure 4, p. 175, for descriptions of brain parts). The general shape and arrangement of vertebrate brains has been assessed in terms 'fissurisation' (foldings of the surface), and 'flexure' —the bending and realignment of parts—which appears to change the longitudinal layout of the fish brain into the more compressed and spherical shape of the brain of birds and mammals (see Figure 3, p. 150). The form and

127

size of fossil brains is sometimes inferred from 'endocasts': a fossil skull may be filled with a suitable material such as latex rubber in order to. produce a rough copy of the original brain surface, and sometimes naturally occurring endocasts are found (Jerison, 1973).

 

Microscopic anatomy and histology

Brains are solidified by in situ perfusion or by later immersion in a solution such as formalin and may also be frozen or embedded in wax before a set of slices a few microns thick are removed by a microtome and mounted on slides for microscopic analysis. There is a vast array of photographic and staining procedures for bringing out different features of 'cellular architectonics'. The silver impregnation methods developed in the late nineteenth century by Golgi and Cajal provided the tool for much of standard comparative anatomy, but there have been relatively recent improvements in this technique. The silver methods of Nauta and Gygax (1954) and Fink and Heimer (1967) are of particular importance for the tracing of degenerating nerve pathways, while the study of cell- body architecture has benefited from electron-microscopy. Many of the new ideas about brain circuitry in animals have resulted from the more powerful silver methods. If cell bodies are destroyed, their axons and axon terminals (which make up nerve tracts or pathways in the brain) atrophy within a matter of days, and if the animal is then sacrificed and the degenerating pathways are located by these histological techniques, functional relationships between parts of the brain may be discovered.

Of course even the modern methods are not infallible, and it is always possible that the pattern of connections manifested by degeneration is subtly different from that of an undamaged brain. But confirmatory results may be obtained from another technique, which involves the transport of radioactively labelled protein tracers down functioning axons ('autoradiography': see, for instance, Cowan et al., 1972). For example, radioactively labelled proline or leucine may be injected into the eye, and then some days later the animal may be sacrificed and its brain prepared for sectioning in order to map out the paths followed in the brain by the optic nerve. The cross sections taken from the brain are simply coated with photographic emulsion, and left until residual radioactivity has drawn visible silver grains into the places reached by the proteins which travelled down the optic nerve. A refinement of autoradiography allows cells in the brain to be marked according to their degree of activity: if a monkey is given an intravenous injection of labelled 2- deoxyglucose and then exposed for

128

about an hour to vertical stripes, immediate autoradiography picks out cells in visual projection area on the surface of the monkey's brain. It is deduced that these cells were activated by the vertical stripes, and therefore absorbed glucose, to fuel their activity, to a greater extent than other cells; and the organised layout of the cells conforms to the results obtained with other physiological methods (Sokoloff, 1975; Hubel, Wiesel and Stryker, 1978).

The degeneration method and the use of radioactively labelled proteins give an idea of the neural pathways out from a particular point to other parts of the brain. It is also helpful to examine which tracts feed into a particular brain structure. Fortunately some substances—as it happens, horseradish peroxidase and egg albumen are convenient ones—are moved backwards along nerve axons ('retrograde transport') and are microscopically visible (Lavail and Lavail, 1972; Kristensson and Olsson, 1974). This has allowed the filling in of some of the gaps in the pathways discernible with the anterograde techniques (Benowitz and Karten, 1976) as well as the plotting out of the whole range of inputs into particular structures (Brown et al., 1977). By combining retrograde and anterograde procedures, reciprocal connections between interacting brain areas can be observed (Jacobson and Trotanowski, 1975).

 

Brain chemistry

The study of brain chemistry is now an extremely active research area, the main questions revolving around the nature of synaptic transmission, and the distribution and function of the various different neurotransmitter substances (Warburton, 1975). In some instances the distribution of brain chemicals may provide evidence of homologies between brains in separate vertebrate classes (Juorio and Vogt, 1967), but on the whole biochemical data have played a relatively small part in comparative investigations (Pearson and Pearson, 1976) since in the main the emphasis is on the biochemistry shared by phylogenetically disparate animals, as in the use of rats to study the brain chemistry of human schizophrenia (Carlsson, 1978) or in the use of goldfish to provide a model for the biochemistry of memory (Agranoff, 1972).

 

Electrophysiological measurement of brain activity

General electrical activity can be detected with electrodes placed on the surface of the skull, and gives the familiar electroencephalogram

129

(or EEG) of brain rhythms. More detailed knowledge can be obtained by recording from individual brain cells or groups of cells—advances in electronics and in the production of fine electrodes make this tactic readily available. The most successful application of this methodology is probably in the study of the visual system: spots or bars of light, of various sizes and with different directions and speeds of movement, are projected on to given points on the retina of the eye and the resulting electrical activity is detected in the retina itself, in intermediary stages of the optic pathways, and in the visual areas of the cortex of mammals. Individual cells may then be classified according to their 'receptive fields'. (A receptive field is the area of the retina which, when appropriately stimulated, causes a particular cell to respond electrically: Kuffler, 1953; Hubel and Wiesel, 1962.) Other sensory modalities are open to the same kind of analysis: the cortical neurons which respond to the touching of points on the body surface can be mapped (e.g. Woolsey, 1965) and the destiny of the information received by an individual whisker on the nose of a mouse can be discovered (Welker, 1976).

 

Electrical stimulation of the brain

The position of brain structures in terms of three- dimensional spatial co-ordinates has been ascertained in many species by histological methods and compiled into stereotaxic atlases. Thus stereotaxic instruments can be used to insert electrodes or surgical tools into predetermined places. Apart from recording, this means that small electrical currents can be instigated via implanted electrodes wherever the experimenter chooses. Reportable sensations in man and an array of natural acts in animals (eating, attack, vocalisation, etc.) can be reliably elicited in this way (Penfield and Roberts, 1959; Valenstein et al., 1970). Not surprisingly, such stimulation also produces emotional and motivational effects, even in the absence of stereotyped behaviours, but often related to them. The examination of 'pleasure and pain areas' in various species has been a source of immensely important data over the past three decades (Olds and Milner 1954; Olds and Olds, 1963; Valenstein et al., 1970; Benninger et al., 1977).

 

Experimental interference with brain function

One of the oldest forms of evidence about brain function comes from human clinical neurology—information concerning the behavioural effects of brain damage caused by disease, tumours and accidents or by

130

remedial surgery connected with these factors. In animals, experimental damage to the brain has been used for more than 1 50 years in the search for explanations of brain function. Other kinds of experimental interference include the administration of drugs and the application of electrical currents to suppress local brain action. There are logical difficulties in interpreting the effects of localised brain damage, which cannot always be overcome, but the combination of data from this sort of study with knowledge assembled from other kinds of investigation often provides the only way of making inferences about brain function. The difficulties arise mainly from the complexity and flexibility of brain circuitry which make it hazardous to assign particular functions to physically identifiable structures. It has often been pointed out that even with standard man-made electronic circuits such as those involved in a television set it is unwarranted to infer component function simply from component damage or removal. If poking a screwdriver into a section of a television circuit produces picture fuzziness, it does not follow that the section damaged is normally a fuzziness inhibitor—which would be the natural conclusion of a physiological psychologist. However, it is hard to see how else we are to proceed except by building up a collection of pieces of evidence about correlations between structure and function, and experimental alterations of structure are frequently useful as tests of hypotheses which may be derived in the first instance from anatomical and physiological descriptions of brain circuits.

The overall size of vertebrate brains

One of the most straightforward kinds of comparison between animal brains is simply the measure of their overall size. We should expect gross size to bear some relation to total number of neuronal elements available in a given brain, and hence overall size ought to provide a rough estimate of its computational power (Von Neumann, 1951). Certainly one of the most frequently heard explanations for the superior cognitive capacities of modern man is the relatively large volume of his brain. Despite the apparent directness of this argument, however, there are numerous subtleties and pitfalls which require extremely careful logical navigation.

The first consideration is that overall changes in size may involve complex changes in physical proportions. Some of these follow from the elementary mathematical differences between the effects of size changes on surface areas and on volumes. Surface areas are critical: in

131

loss of heat from the body; in the space available for chewing and for absorption in the intestine; in the cross- section and therefore the strength of the bones; and, in more intellectual functions, surface areas determine the space available for the retina and for the mammalian cerebral cortex. Volume (or weight) is most crucially correlated with total number of cells. The mathematical law is that area increases as the square, and volume as the cube, of linear dimensions. Therefore small animals have a relatively high ratio of body surface to internal volume, and this ratio must systematically decrease as linear dimensions increase. Another deduction, which goes back to Galileo (1564—1642) is that large animals must need bones which are relatively much thicker than those of smaller animals. The deduction which concerns us here is that, other things being equal, the ratio of surface area to internal volume of a large brain must be very much less than the ratio of surface area to internal volume of a small brain.

gif

Figure 2. Brain weight as a function of body weight (from Jerison, 1973; see p. 132)  

 

There is another point, however, that we must deal with first, and this is that although as animals become larger we might expect them to need large bones to support themselves and large digestive systems for their nutritional requirements, there is less reason to suppose that they should need particularly large eyes and brains, since what has to be

132

seen and understood by a large animal does not necessarily differ very much from what has to be seen and understood by much smaller creatures. This point was put forward long ago by von Haller (1762) —more recent reviews of various biological corollaries of size increases have been provided by Rensch (1959) and Jerison (1973). Notwithstanding the antiquity of the groundwork provided by von Haller, there is still a good deal of confusion about the relationship of brain size to body size. This arises because there are two essential propositions which seem to point in opposite directions. The first proposition is that larger animals have brains which are absolutely larger than the brains of more diminutive creatures. Second, the brain of a large animal is smaller, relative to its overall body weight, than would be expected by strict repetition of the proportions seen in a little animal. It is important to bear in mind both these opposing ideas—the brain tends to become absolutely larger, but relatively smaller, as body size increases.

Most systematic studies of brain and body weights across different species suffer from arbitrary selection of species and even more arbitrary selection of individual animals to represent a particular species (Sholl, 1948). However, allowing for a generous margin of error, the data used by Jerison (1973) provide a basis for discussion, and one of Jerison's plots of brain weight against body weight is reproduced in Figure 2. The data are shown as points within two statistically separable polygons. It is vital to stress that the logarithmic co-ordinates used for the plot mean that a straight line sloping upwards does not indicate that brain weight is a constant proportion of body weight. Such a relationship would produce a line with unit slope, with the intercept on the vertical axis indicating the constant of proportionality. The gradient of less than 45 degrees represented by the continuous line in Figure 2 means that brain weight tends to increase from species to species with a fractional power of the body weight. For instance, the line fitted for higher vertebrates makes brain weight proportional to the two-thirds power of body weight. (E = .07P0.6667, where E is brain weight and P body weight). The extent to which this gives brain weights for large animals which are dramatically smaller than would be expected by extrapolation of strict proportions can be illustrated by numerical examples.

Using the formula E = .07P0.6667, an animal weighing in at one kilogram would be expected to have a brain weight of 7 grams-seven-thousandths of its mass. A species ten times larger would not have a ten times bigger brain according to the equation: a 10kg body weight

133

gives an expected brain weight of only 32.5 gm, not the 70 gm which would come from strict proportion. The heavier the animal, the greater the disparity: the formula gives a 100 kg animal a brain of 150 gm rather than 700 gm, and a beast of 10,000 kg (a large elephant, say) gets a brain of only 3 kg, not 70 kg. On the other hand animals weighing less than the I kg example we started with get relatively large brains by the application of the formula. A rat or a pigeon with a body weight of 250 gm gets a 3 gm brain—not much it seems to us in our 1,500gm wisdom, but quite a bit more than seven-thousandths of its body weight. Such is the make-up for brains in even smaller animals that a sparrow's brain (1 gm), as a proportion of its body weight (20 gm), is larger than a man's.

Clearly body weight must be taken into account in deciding whether a particular species' brain is 'large' or 'small', but since a sparrow devotes a twentieth of its mass to its brain, while the human brain is only a fiftieth of the human body weight, an adjustment must be made for absolute size, and assuming that the brain is proportional to the two-thirds power of body weight is the best adjustment available. Given this, what can be inferred from the data presented in Figure 2? First, it is gratifying to see that modern man does have a large brain for his size, since he is further above the line than any other species. Certain other animals, in particular the crow family and the dolphins and porpoises, also show up with larger brains than would be called for by the two-thirds power of their body weight, but to a lesser extent than ourselves. Statistically speaking, the most obvious separation between groups of animals occurs between higher vertebrates and the rest. The difference between mammals and birds within higher vertebrates does not reach statistical significance, and neither does the difference between reptiles and fish (within lower vertebrates a reptile of a given size tends to have a smaller brain than a fish of the same weight: Jerison, 1973). However, it is apparent from Figure 2 that, although higher vertebrates have larger brains than lower vertebrates of equivalent sizes, the way in which brain weight increases with body weight is just the same in lower and higher vertebrates.

We now have two theoretical questions about the empirical facts distilled from the data: why do higher vertebrates devote more of their weight to the brain than lower vertebrates—and why is vertebrate brain weight proportional to the two-thirds power of body weight? It might be hoped that the large brains of higher vertebrates could be clearly ascribed to a high value placed on brain functions in these species. This may turn out to be the case, but it is difficult to decide

134

whether lower vertebrates, because of their meagre intelligence, make do with less brain tissue, or alternatively whether, because of their cold-blooded metabolism, they can afford to carry a greater mass of flesh and bone. It is usually taken for granted that the smaller brain indices of lower vertebrates indicates their lesser intelligence, perhaps quite rightly, but in the absence of independent assessment of psychological capacities the argument tends towards circularity. (Smaller brains occur in less intelligent animals, which we know are less intelligent because they have small brains.) An example of the problems lurking among the generalities is the case of sharks, regarded as a primitive order of fish, who seem in any case to have rather large brains, but whose intelligence quotient, if it is calculated on the basis of body weight, must be upped a bit by their use of a cartilaginous skeleton, which is lighter than one of bone. Jerison (1973) concludes in desperation that sharks are not to be considered as fish, and are therefore not lower vertebrates after all, but intermediates, thus saving his distinction between the brain size of higher and lower vertebrates, but throwing away part of the traditional phylogenetic scale.

The theory which Jerison defends is that lower vertebrates have small brains because of the primitive nature of their psychology, and that early mammals also had small brains for the same reason. The crucial increases in mammalian brain size which supposedly took place during the last 50 million years or so can only be quantified by heavy reliance on estimates of brain size from endocasts of fossil skulls and guesses at body weight based on what remains of fossil skeletons. It seems plausible that there was selection pressure on brain effectiveness caused, for instance, by competition between and within mammalian herbivores and mammalian carnivores, but the evidence in support of this hypothesis remains very sparse (cf. Simpson, 1953). Even in the case of modern species which are unquestionably large-brained by comparison with other modern species, such as man or the dolphin, fossil evidence to show when the increase in brain size took place is hard to come by, although there is little doubt that hominid species prior to Homo sapiens already had large brains (Pilbeam, 1972 ;Jerison, 1973).

Jerison's conclusion, however, is clear, and worth repeating:

The evolution of intelligence occurred mainly within mammals and only in a casual way in birds, if one defines intelligence as the capacity to learn new response patterns in which sensory

135

information from various sensory modalities is integrated as information about objects in space. (Jerison, 1973, p. 433)

 

The main difficulty with this is that the brain size data do not support such a clear distinction between mammals and birds. The data as they stand say only that cold-blooded animals, with the notable exception of sharks, tend to have smaller brain weight/body weight ratios than the warm-blooded vertebrates, which include both birds and mammals. But since mammals are in general bigger than birds they also in general have bigger brains, and it is perfectly legitimate to argue, as does Rensch (1959), that absolute size of the brain may be a more valuable pointer to absolute intellectual capacity than any brain/body ratio. Accepting this would oblige us to consider whether an ostrich is more capable than a small monkey and whether an elk has greater psychological capacity than a chimpanzee.

This brings up the question of the interaction between body size and brain function which we started off with. The problem of why a particular species has the body size that it does is difficult enough in itself. There are some disadvantages in being small (for instance, excessive heat exchange via the body surface) and there are also disadvantages in being very large (mechanical awkwardness in locomotion and simple weight support—these factors, and the technicalities of food supply, are perhaps sufficient to explain why the largest land mammals tend to be herbivores). It is obvious that flying imposes a much lower limit on size than swimming—it is hardly surprising that there are no whale-sized birds. But the various limits on size leave a very' large range of possible sizes, and many groups of animals seem to have adopted varying proportions at different stages of evolution. 'Cope's Law' says that there is a progressive tendency for species to get bigger and bigger as they evolve, but there are numerous counter-examples (Simpson, 1953).

The point of psychological interest is whether animals which happen to have different gross weights need brains with different capacities. On the face of it, the behaviours to be controlled by the brain of a gorilla are not radically more demanding than the behaviours of a very small monkey. The same thing could be said of an elk and a miniature deer, or a tuna and stickleback. Other things being equal it is arguable that, due to reduced urgency for nutrition, and greater immunity from predators, large animals have a somewhat easier life than small animals, and therefore less need of intelligence.

136

But suppose we simply assume that roughly equivalent demands are placed on the brain, whatever the scale of the body whose actions need to be supervised. One hypothesis which would follow from this is that an increase in overall size should be a short-cut to greater sophistication of behavioural control, since a large animal should have available more brain capacity to devote to behavioural problems which are essentially the same as those faced by its less well-endowed brethren. The main argument against this is the trend noted above for increases in body size across species to be accompanied by what seems to be an effort to cut down on the proportion of body weight taken up by the brain. This trend suggests the alternative hypothesis—that variations in body size will take place with minimal modification of brain capacity, and hence the range of brain sizes will be very much less than would be expected on the basis of the range of overall dimensions of body size. The connection of brain weight to a fractional power of body weight, as shown in Figure 2 and discussed above, provides a measure of support for this hypothesis of conservation of brain capacity. More detailed confirmation can be obtained in the case of very closely related groups of species. Here, it is found that brain weight is proportional only to the cube root of body weight, rather than to the two-thirds root—in other words there is even less variation in brain size between closely related species of different sizes than there is between unrelated species. Illustrative examples can be drawn from families which include miniature or pigmy subspecies. The pigmy chimpanzee (Pan paniscus) has a brain which is virtually identical in size to that of the very much larger main species Pan troglodytes, and similarly within Homo sapiens a pigmy body type is accompanied by a normal size of brain. Breeds of domestic dogs and chickens show very large variation in overall dimensions with very little change in absolute brain size (Rensch, 1959; Jerison, 1973). If dolphins and porpoises are regarded as pigmy species of Odontocete whales, their very high brain weight to body-weight ratios could come under the heading of the defence of brain size against overall size variation.

It is clear that primates as an order have large brains, over a wide range of body sizes from squirrel monkeys to gorillas, but it is equally clear that an exceptional expansion of brain size took place in the hominid lines ancestral to modern man. The available fossil evidence suggests strongly that related hominid species with approximately similar body sizes possessed brains with weights varying from 500 gm (Australopithecus boisei) and 600gm (Homo habilis) to 1000 gm (Homo

137

erectus) and 1500 gm (Homo sapiens Neanderthalensis: see Jerison, 1973, p. 398). In hominid species, therefore, it looks as though there was a very special kind of selection pressure towards larger brains, but it should be emphasised that this selection pressure began to operate at the early stages of hominid evolution, long before the emergence of Homo sapiens. The brain of an orang-utan of roughly human body size weighs about 350 gm. This had been doubled by the stage of Homo habilis investigated at Olduval Gorge, who probably walked erect and made primitive tools, although without the benefits of a thumb as opposable and as useful for precision grips as that of modern man (Leakey et al., 1964). A similar degree of brain enlargement may also have been achieved as much as 2.6 million years ago by yet another hominid with bipedal habits (Leakey, R., 1973).

A brain three times as big as the orang- utan's appears in Homo erectus, who used fire and undoubtedly hunted and made tools, and left evidence of cannibalism, but who probably did not have such more advanced characteristics as care of the sick, burial of the dead and drawing on walls. The typical brain size of modern Homo sapiens, roughly 1100 to 1500 gm, or about four times bigger than the size of the orang brain, was reached, and possibly exceeded (there are examples of cranial capacities of 1700 gm) by later Neanderthals, who qualify as Homo sapiens, but are denied full Sapiens sapiens status, despite evidence of ritual burial of the dead and their 'Mousterian' stone technology which produced a great variety of scraping and cutting tools by the detailed working of stone flakes. The relation of brain size to tool-making is not at all clear. The earliest known stone implements are from the Olduvai Gorge; they consist mainly of pebble-tool choppers made by knocking off flakes from two sides of a lump of rock and using the remaining sharp core without further dressings of the edge (Oakley, K. P., 1972). However, even at this early stage, before the major expansion of the human brain, there may have been some trimming and sharpening both of the cores and of the initial bits chipped off (the primary flakes). Also, and even more remarkable at this primitive stage, there was 'machine tool' production, that is the use of special hammer stones and anvil stones as 'tools for making tools'. It is thus no simple matter to tie intelligence via brain size to the level of stone technology accomplished, although the variety, sophistication and delicacy of construction of stone tools are said to increase during the course of human evolution. In any event Neanderthal man, although technically accomplished, did not leave behind enough by

138

way of artefact or ornament to indicate that he possessed the aesthetic sense usually attributed to Homo sapiens sapiens. One view, albeit very flimsily based on inferences about Neanderthal vocal tracts, is that Neanderthals were not capable of the full range of human speech (Lieberman, 1975). In spite of (and possibly because of) this, Neanderthal brains are indistinguishable from those of modern man on the grounds of size except in so far as they are actually larger.

The expansion of the human brain thus began and ended before the appearance of the species we recognise as ourselves. It thus did not occur because of modern civilised habits, although these may only be feasible with a brain that is already enlarged. Since hominid brain expansion predated modern civilisation, it may perhaps be included within the scope of discussion of vertebrate brain evolution. Purely in terms of size, the extent of human brain expansion can be kept within natural proportions by comparisons with larger mammals (see Figure 2 and Quiring, 1950). Gram for gram, man has the body of an orang-utan and the brain of a large walrus. Conveniently, a stubby limbed North American rhinoceros (Teleocerus), now extinct, once had a brain listed at the same weight as man's — 1500 gm. Modern rhinoceruses and hippos are slightly bigger animals but, curiously, they have brains of only 600 or 700 gm. The African elephant is an immense animal with rather more brain than Jerison's equation would predict—5000gm, four times larger than yours or mine. The biggest brains of all are of course those of the large whales, but they are not very much bigger than the elephant's. The brains which come closest to man's in terms of relation to body size as well as in terms of absolute size are those of the small whales, the dolphins and porpoises. These brains are rounder and more encephalised than man's and have 'hyperfissurisation' of the cortex, that is the surface foldings of the cerebral cortex are more extensive than the surface foldings of the human brain. Brains like this, of human size or larger, occur in animals that would only be expected to have brains of about 250 gm, according to Jerison's equation based on body size.

Brain size and neuron density

Leaving aside man and the dolphin, the general rule seems to hold fairly true that the absolute weight of the brain increases as the two-thirds power of body weight, and therefore as the square of the length or

139

height of a species. Although this means that large animals will tend to have smaller brains in direct proportion to body weight than little animals, one would not expect this to be much comfort to the latter, faced with the problems of processing information with a smaller absolute amount of neural tissue. Irrespective of scale, the only reasonable assumption is that computing power or efficacy of behaviour-controlling function is proportional to the available number of neuronal elements in the central nervous system (Von Neumann, 1951). On the whole bigger animals are bigger because they are composed of more cells than smaller animals, and this holds roughly true of the brain, but there are several features of microscopic brain organisation which point to the neural luxuries enjoyed by large animals and the neurological thrift required of small animals.

Over and above species-to-species difference, smaller vertebrate brains are composed of smaller neurons, more densely packed together, so that a gram of brain tissue in a small brain may contain many more neural elements than a gram of tissue from a large brain. The brain as a whole obviously consists not only of neurons but includes the associated blood supply, ventricular system and covering membranes. Mixed in with the neurons themselves are large numbers of supporting or interstitial cells of various types especially glia, or neuroglia. The density of neurons can be increased either by reducing their size, or by cutting down on the amount of non-nervous brain tissue. Both these features occur in smaller brains, at least within mammals, from whom most of the quantitative data have been collected. Many types of brain neuron increase in size with the overall size of the species (see Rensch, 1959; and Tower, 1954). This is particularly visible in the length of the dendritic tree of cortical neurons (Bok, 1959). Since glial cells fill in the gaps between dendrites it is to be expected that larger brains should have a relatively high proportion of glial to nerve cells, and Tower and Young (1973) found a very orderly relationship, with glia/nerve ratio increasing with the one- third power of mammalian brain weight. The end result is that, in the cortex of mammals, neuron density varies inversely with brain weight to the one-third power. (The range is from thousands to hundreds of thousands of neurons per cubic millimetre.) Thus although large brains contain more neurons, they do not do so in strict proportion to their size. In fact the relation between total number of neurons and brain size, estimated in this way, is rather similar to the relation between brain size and body size: bigger animals have bigger

140

brains, but not in strict proportion; and bigger brains have more neurons, but not in strict proportion.

The combination of these factors means that bigger animals have more brain neurons, but extra brain neurons come only in proportion to, very roughly, the square root of extra body weight. This still leaves small animals with less powerful brains, but suggests that there is a tendency to conserve brain power in smaller mammals by allotting a generous fraction of the total of the animal's cells to brain neurons. One might put it the other way around, and say that large animals are a lot less profligate with neurons than they could be. Despite the cutback on brain neurons in large animals, one can still argue, following Rensch (1959), that because bigger animals have more brain cells greater intelligence may come as an accidental fringe benefit of any increase in scale.

Practically all of the data on variations in neuron density with brain size come from mammals. It is worthy of note that the human brain appears to have just about the cortical neuron density that would be predicted by extrapolation from other mammalian brains of different sizes: the human brain is exceptional in terms of its mass in relation to body weight, but not in terms of its internal cell density. It seems a reasonable hypothesis that smaller brains should have higher neuron densities in other vertebrate classes apart from mammals and this suggestion has been made in the case of birds by Kappers (1947). Data for birds are rather variable, and Pearson (1972) is doubtful about the correlation between brain size and cell density. It is surely no accident, however, that the highest cell densities quoted by Pearson (from Stingelin, 1958) occur in the Goldcrest (Regulus), the tiniest of all European birds, and the next highest neuron density occurs in the Goldfinch (Carduelis), which is not very much bigger. Larger birds such as the cuckoo or pigeon have lower brain-cell densities. Pearson (1972) is worried that a couple of fairly big species seem to have higher brain-cell densities than they should, but since these species turn out to be a crow and a macaw, we might be forgiven for assuming that their unexpectedly large number of brain cells is explicable in terms of the reputed high intelligence of birds in the crow and parrot families.

 

Miniaturised brains in teleost fish

The problem of cramming an adequate amount of behaviour- controlling machinery into an extremely small body is most acute in the teleost fish (which include goldfish, minnows, sticklebacks and

141

guppies). Many species of teleost fish are very small, by mammalian standards, even when adult, but it must also be considered that the teleost brain has to control free-swimming behaviour in the even more minute fry of these species. Possibly because of this need for compactness (see Aronson and Kaplan, 1968) but calling for special attention in any case, teleost fish have an unusual method for the embryological development of the forebrain. In all other vertebrate classes the forebrain grows from a U-shaped neural tube when the outside walls rise and turn inwards, to form the cerebral hemispheres. This is known as evagination, and except in the primitive cyclostomes (in which the walls never turn downwards, leaving a single internal space or ventricle) the walls curl in, down and around to make a cross-section of two separate hemispheres each with a hole in the middle (the holes being the lateral ventricles). A cross-section through the frog forebrain, for instance, looks like two misshapen doughnuts, side by side, with a lot of wasted space.

In teleost fish, although there is considerable variation from species to species, forebrain development occurs when the two side walls of the neural tube expand outwards at the top, pushing the inside faces close together and swelling out and around, leaving no space for internal ventricles. This 'eversion' during the development of the teleost forebrain makes it rather difficult to find which bits go where, by comparison with other vertebrates, but gives a pleasing sense of economy of design to the forebrain section of a stickleback or goldfish, and suggests that the psychological capacities of the teleost forebrain may be slightly greater than we might expect simply on the grounds of its external dimensions.

A further special aspect of the teleost forebrain, which is extremely pertinent to lesion studies of its function, is that the embryological instructions for growth are retained in the adults of many species. This means that the forebrain can regenerate after local damage, so that the reappearance of efficient behavioural organisation after lesions to a particular part of the forebrain does not imply that that part of the forebrain is not important; an alternative possibility is that it is of such importance that arrangements have been made for effective regrowth after damage. Regeneration of central and peripheral nervous system can also occur in amphibians (especially in the larval stage), and the capacity for some adult brain growth is apparent in elasmobranch fish (Leonard et al., 1978), but effective regrowth of the damaged adult brain is most remarkable in small teleosts (see Segaar, 1965, for a

142

review). An example of highly organised regrowth in the adult teleost brain, which is scarcely credible by comparison with the very limited peripheral neural regeneration seen in higher vertebrates, occurs after the surgical removal of parts of the sending or receiving ends of the optic tract. The major connection in the teleost visual system is between the retina of the eye and the contra-lateral optic tectum of the midbrain (see Figure 4, p. 175). This connection is topographically arranged so that each point on the retina goes to a corresponding point in a systematic two-dimensional array on the tectum. Suppose that one half of the tectal array on one side is cut out. In theory, the fish becomes blind in the part of the visual field which normally projects to the portion of the tectum which has been removed. But, over a period of months, what happens is that the remaining terminals of the optic nerve move over, and the severed half of the optic nerve sticks itself on to the intact area of the tectum so that a complete, although compressed, visual picture is recovered (Sharma, 1972; Yolen and Hodos, 1976). On the other hand, if half the retina is removed, the connections from the remaining half spread out to make use of the entire optic tectum (Schmidt et al., 1978). The way in which such intelligent-seeming regrowth of adult neural circuits occurs is probably the same as that used in the initial embryological growth of all vertebrates (Miller and Lund, 1975). But in the small brains of lower vertebrates it is perhaps more important to retain the capacity for organised growth in the adult than it is in the larger brains of higher vertebrates, where recovery of function after brain damage can be more readily accomplished by reliance on the tissue that is left.

The neural components of vertebrate brains

Although the weight or volume of animal brains is easily inspected, it might be more to the point to go to a microscopic extreme, and ask whether vertebrate brains may differ in the neural components of which they are made. Is the tissue in the brain of a shark or a frog the same stuff as that which gives the human brain its remarkable intellectual powers? Surprisingly perhaps, the answer, as far as we can tell, is yes (Kuhlenbeck, 1967, 1970; Kappers et al., 1936; Sarnat and Netsky, 1974). Some brain cells, such as the giant Mauthner and Muller neurons in the medulla of lower vertebrates (Zottoli, 1978) are specific to certain classes, but on the whole the transition from one

143

vertebrate class to another is not marked by radical changes in the nature of brain neurons. Many comparisons of neurons in brains are based on what the cells look like after staining with metallic ions, which is rather remote from an assessment of what they can do, but more detailed structural descriptions via electron microscopy and electrophysiological and biochemical studies of neuron activity at present suggest similarities in the modes of operation of neurons in the various vertebrate classes, rather than differences.

It should be recognised that there are specialisations of shape of cell, and of nerve-fibre structure, for different parts of the brain (see below) and knowledge of the crucial interactions between neurons, at synapses, is expanding very rapidly. Therefore it is possible that phylogenetic changes in the sophistication of neural operation remain to be discovered—a persistent hypothesis is that there exists some as yet unknown microscopic detail in the fine structure of the human brain which results in a qualitative change in function (e.g. Popper and Eccles, 1967). But the most common assumption is that changes in brain function have evolved by changes in the amount of brain tissue and its arrangement, rather than by radical alterations in the nature of its constituent elements. As an example, one may quote the distinctive and easily recognisable cells of the cerebellum, the small 'granule' cells and the Purkinje cells, which occur in all vertebrates. Not only is the physical appearance of these cells similar in the shark, frog and cat, but analysis of their electrical responses, in the shark, shows that with these vertebrates 'already in evolutionary development the cerebellum had achieved most of the characteristic features of the neuronal machinery in the mammalian cerebellum' (Eccles et al., 1970).

If there is a typical component for the vertebrate brain, it is a 'multipolar' neuron, in which the cell body gives off several twig-like dendrites (fibrous extensions of the cell body which receive input) as well as an axon (the output line of the cell). The axon may also branch out to provide multiple terminations, providing several synapses with the dendrites of other neurons. Human axons vary in length from less than a millimetre to almost a metre (as part of a nerve or fibre tract). The basic idea is still that neurons act as logical switching elements, amplifying, inhibiting, gating and recording neuro-electrical impulses rather like the components of a computer or pocket calculator (Sommeroff, 1974). The most general categorisation of vertebrate neurons is in terms of the multipolar characteristic as opposed to the more primitive unpolarised type (no distinction between the axon and

144

the dendrites), the bipolar type (a single dendrite and an axon) and the unipolar type (dendrite and axon arising from a single stem). All types except the multipolar occur also in the nervous systems of invertebrates (Kuhlenbeck, 1967). These categories are supplemented by classifications based on the physical shape of the cell body (pyramidal, fusiform, stellate, granule and so on) or of the dendrites (Purkinje cells, for instance, can be recognised by their very extensive and distinctive dendrites). Axons may be graded according to their length and thickness and to their degree of surrounding insulation (myelination) as well as by their shape or pattern (e.g. 'mossy' fibres and 'climbing' fibres). The interaction between axonal terminations and the dendrites of other neurons takes place by the release of chemicals, and neurons can therefore be classified according to which of the numerous known 'neurotransmitter substances' is employed (e.g. noradrenaline, acetylcholine, dopamine).

Although some cell types are restricted to certain vertebrate classes, the main impression is of the repeated use of the same cell types in widely varying species. Long et al. (1968), for example, in an investigation of the fine structure of elasmobranch brains, found the same basic neuron categories in sharks as the ones which occur in mammals. This goes for the non-neural brain cells too: although more use may be made of neuroglia in the larger mammalian brains, most of the many kinds of supporting brain cells found in mammals occur not only in sharks but also in the even more venerable lampreys (Schultz et al., 1956: Pearson and Pearson, 1976, pp. 48-9). We may conclude, I think, that the major process in brain evolution has been in the deployment of brain cells rather than in the design of the cells themselves, although this does certainly not preclude cellular changes as well. Ideas as to how redeployments of brain cells may have come about, in terms of changing functions of different divisions of the vertebrate brain, are dealt with in the next chapter.