Dr. J. K. Burns

Word counts: Abstract: short - 97 long - 249

Main Text: 9296

References: 4398

Entire Text: 14297

AN EVOLUTIONARY THEORY OF SCHIZOPHRENIA: CORTICAL CONNECTIVITY, METAREPRESENTATION AND THE SOCIAL BRAIN.

Dr. Jonathan Kenneth Burns

Clinical Lecturer

Department of Psychiatry

University of Edinburgh

Morningside Park

Edinburgh

EH10 5HF

U.K.

jkburns@holyrood.ed.ac.uk

SHORT ABSTRACT

Schizophrenia is a universal disorder with largely genetic aetiology. A theory is proposed that schizophrenia is a disorder of cortical and specifically fronto-temporal connectivity that evolved in association with emerging complex neural circuitry in human ancestors. These circuits evolved under selective pressures involving group living, and regulate aspects of social cognition such as metarepresentation and affective responsiveness. Evidence from various scientific fields suggests that the evolutionary advantages conferred by these changes rendered the hominid brain vulnerable to insults. I argue that schizophrenia exists as a costly trade-off in the evolution of social cognition and the creative mind.

LONG ABSTRACT

Schizophrenia is a worldwide prevalent disorder with multifactorial but highly genetic aetiology. Prevalence rates are approximately 1% in most societies surveyed, despite lowered fertility in affected individuals. Thus it is argued that an evolutionary advantage exists in genetically related unaffected relatives. Various theories of ultimate causation of the schizophrenic genotype and Tim Crow’s hypothesis regarding cerebral asymmetry and language are reviewed and found wanting. In keeping with available biological and psychological evidence, an alternative theory of the origins of this disorder is proposed. Schizophrenia exists as a costly trade off at two stages in the evolution of metacognition and the social brain. Paleoanthropological and comparative primate research suggests that hominids evolved complex cortical interconnectivity (in particular fronto-temporal circuits) in order to regulate social cognition and the intellectual demands of group living. Ontogenetic mechanisms underlying this cerebral adaptation rendered the hominid brain vulnerable to genetic and environmental insults. This is the first trade off experienced by hominid ancestors. I argue that genetic events occurring prior to the migration of H. sapiens out of Africa 150 -100 000 years ago gave rise to a genetic spectrum that, in it’s homozygous form resulted in the schizophrenic phenotype, while heterozygous ‘schizotypal’ individuals possessed cognitive advantages that enhanced their relative fitness. Thus schizophrenia evolved as a trade off firstly in the emergence of complex social cognition and secondly in the emergence of a phenotype that exhibited unusual creativity and iconoclasm and may be associated with the great cultural and scientific advances of human history.

KEYWORDS: adaptation, cognition, connectivity, creativity, evolution, evolutionary psychiatry, metarepresentation, primates, schizophrenia.

1. Introduction

“Reasonable people adapt themselves to the world.

Unreasonable people attempt to adapt the world to themselves.

All progress, therefore, depends on unreasonable people.”

[George Bernard Shaw]

Schizophrenia is a complex and widespread disorder, giving rise to a great burden of suffering and impairment in both patients and their families. It is human nature to seek meaning behind experiences and indeed patients with schizophrenia seek meaning in the bizarre and perplexing phenomena of their psychoses. So too, in our scientific endeavour to understand this complex disorder, it is important to undertake a search for the meaning of the existence of schizophrenia in the human species.

In recent years a number of researchers have sought to adopt an evolutionary perspective to explain the continued persistence of clearly maladaptive psychiatric disorders. In the field of schizophrenia research, Tim Crow, John Price, Anthony Stevens and others have pioneered an evolutionary approach to understanding the meaning of this distressing illness. They have asked questions about the evolutionary origins of schizophrenia and about why it is maintained in the human genome.

This paper critically assesses these models drawing on evidence from various fields including psychiatry, psychology, paleoanthropology and primatology, finds them wanting, and proposes an alternative and more integrated theory of the origins of schizophrenia that is in keeping with all the available evidence. The central argument proposed here is that schizophrenia reflects the severe end of a spectrum of disordered fronto-temporal (FT) connectivity, which is related to the evolution of metarepresentation and the ‘social brain’ in the human line. It is argued that schizophrenia exists as a costly trade-off at two stages of cognitive evolution.

I suggest that the first trade-off occurred between 16 and 2 million years ago (mya) when human ancestors evolved complex cerebral interconnectivity and specialised neural circuits in order to regulate social cognition and the intellectual demands of group living. Owing to anatomical constraints on fetal brain size, these changes necessitated the ‘infantalisation’ of the child and the prolongation of brain maturation well into adolescence. This meant that the human brain, with its complex and recently evolved circuitry, became increasingly vulnerable to both genetic and environmental developmental insults. This vulnerability was the trade-off for the advantages gained in social cognition.

The second trade-off occurred approximately 150-100 000 years ago when a genetic mutation resulted in aberrant connectivity in these FT circuits. The exact mechanism responsible at the phenotypic level is debateable and will be discussed in this paper. It is argued that the expression of this genetic change is variable in different individuals and that ‘milder’ expression proved to be adaptive in the environment of evolutionary adaptedness (EEA). Therefore the schizophrenic genotype was retained in the human genome. Hence the disorder we call schizophrenia emerged as a trade-off for the advantages gained by some individuals in the schizotypal spectrum (ie. individuals who share some of the clinical features of schizophrenia such as magical and referential thinking, but never have overt psychosis.)

2. Evolutionary origins of the schizophrenic genotype

The WHO has undertaken a number of important cross-cultural studies of the epidemiology of schizophrenia. The International Pilot Study of Schizophrenia, conducted in nine countries (World Health Organization 1973) and a later larger study (Sartorius et al 1986) demonstrated remarkable consistency in the prevalence and the core symptoms of the disorder. One of their ‘first-rank’ findings was that the evidence points to a significant genetic component in the transmission of schizophrenia (Jablensky 1988). Other evidence suggests that this is a polygenetic disorder (Kendler et al 2000).

The fact that this disorder is found universally implies that the schizophrenic genotype dates to at least 150-100 000 years ago when the migration of Homo erectus out of Africa occurred (Crow 1995a). A less parsimonious theory would have to argue that there was parallel evolution in geographically separate populations. This is unlikely. There are at least some historical references to schizophrenia-like illnesses in early Islamic, Ayurvedic and Classic societies (Gottesman 1991;Youssef & Youssef 1996) but these obviously post-date the development of written records.

It is a well-accepted fact that schizophrenics have lower fertility (Larson & Nyman 1973) and increased early mortality (Brown 1997). Therefore the disorder is maladaptive and, according to the laws of natural selection (Darwin 1859), should have disappeared from the gene pool long ago. The constant prevalence of core features therefore implies that there is some adaptive advantage conferred by one of two genetic mechanisms: 1) the ‘pleiotropy hypothesis’ - that the gene or genes for the trait/disorder is/are linked to other genes which confer an advantage; or 2) the same genes that cause disorder in some individuals have advantages in others (Cartwright 2001). This might involve either a different combination of genes or different gene expression. The example of sickle cell anaemia is commonly cited where homozygous expression of the genetic defect results in the disease and heterozygous expression confers resistance to malaria. The latter mechanism is preferred again for parsimonious reasons for the purposes of this discussion of schizophrenia.

Tim Crow has proposed that psychotic illness should be viewed as lying on a “continuum of variation” (Crow 1998a;Crow 1995a). It seems that there may be at least two dimensions to this spectrum – one stretching from schizophrenia to the affective psychoses and the other from disorder to trait to ‘normality’. Evidence from neuroimaging and neuropsychology studies suggest that schizotypal people have milder but similar deficits to schizophrenic patients (Buchsbaum et al 1997b;Buchsbaum et al 1997a;Cadenhead et al 1999;Dickey et al 2002). That these deficits are also found in relatives of patients with schizophrenia suggests a genetic cause (Lawrie et al 2001;Byrne et al 1999). This supports the idea of a genetic continuum. Thus those individuals in the spectrum, representing perhaps different expressions or combinations of genes - those individuals we might describe as schizotypal or schizoid for example - could have mild and adaptive traits that were selected for during the EEA. Support for this notion of enhanced fitness in ‘carriers’ comes from recent research by Avila and colleagues that demonstrates an increased number of children in first degree relatives of schizophrenic individuals (Avila et al 2001).

This leads one to considering the ultimate causation of the schizotypal genotype. The focus of most psychiatric research is on proximal causation (e.g. neurotransmitter activity). A number of theories have been proposed over the last 30 years regarding the possible adaptive advantages conferred by the schizophrenic genotype. These include: that the genotype confers some physiological advantage on the individual such as resistance to infection (Carter & Watts 1971) – but there is no evidence for this; that the advantage lies in some aspect of social interaction (Kuttner et al 1967), specifically that the genotype mediates ‘territoriality’ and therefore the survival of the family unit (Kellett 1973); that there is variability in multiple genes that control social skills and personality traits (Farley 1992;Farley 1976); and/or that the genotype is linked with ‘sociality versus asociality’ in the preservation of the individual (Allen & Sarich 1988).

Stevens and Price have argued in their two books, Evolutionary Psychiatry and Prophets, Cults and Madness, for a “group-splitting hypothesis” of schizophrenia (Stevens & Price 1996;Stevens & Price 2000). In the ancestral environment a group would reach a critical size at which it began to outgrow its resources. At this point a schizotypal individual, having undergone a “mazeway resynthesis” and spurred on by his or her iconoclastic ideas and possible ‘voices of the gods’ (Jaynes 2000), would offer a vision of a new and better ‘promised land’ to those who would follow. The shaman of traditional cultures such as the Inuit may be a more recent example of this type of character. The power of the schizotype was such that followers would enter his/her ‘delusional’ world (or at least go along with it) and the group would split. The converted and their leader would either stay and enter into a genocidal conflict for resources with the ‘outgroup’ or set off on a migration, dispersing our ancestors across the planet. Thus the particular personality and behaviour of the schizotype proved adaptive for the group, according to these authors. They cite a number of clearly schizotypal but also paranoid and psychopathic cult leaders such as David Koresh, Jim Jones and Adolf Hitler as examples of this phenomenon in recent times (Storr 1997), and interestingly note that both Koresh and Jones fathered many children – suggesting increased fitness as a result of their schizotypal personalities. The eventual failure of these modern gurus may reflect greater social intolerance and censure relative to the EEA. Emmanuelle Peters has studied members of a number of religious cults in Britain and discovered a high level of near-psychotic delusional beliefs (Peters et al 1999). This suggests that schizotypal traits flourish in cults or that schizotypes flock to them and it may also support the notion that in the ancestral environment, where ‘cults’ may have had a greater impact on society, these traits may have played a very significant role in the splitting and dispersal of people.

Finally, many writers and thinkers, both in remote and recent times, have linked ‘madness’ to creativity, genius and religiosity. John Dryden wrote:

“Great wits are sure to madness near alli’d

And thin partitions do their bounds divide.”

Karlsson demonstrated an increased incidence of psychotic illnesses including schizophrenia in a cohort of particularly gifted artists, philosophers and politicians (Karlsson 1973;Karlsson 1984;Karlsson 2001). The Nobel Laureate, John Nash springs to mind as a contemporary example, while an attempted reading of Ullyses by James Joyce must raise some questions as to the mental state of this great author. Jamison has pointed out the high incidence of mood disorders in creative individuals such as Schumann, Shelley, Byron and Van Gogh (Jamison 1993;Jamison 1995), while Baron-Cohen and colleagues, in a fascinating study conducted recently at Cambridge University (UK) have demonstrated higher than expected scores for Asperger syndrome/ high-functioning autism among scientists and mathematicians (Baron-Cohen et al 2001). Others who have written about the association between mental illness and creativity/genius include Parfitt (Parfitt 1956), Maudsley (Maudsley 1908), Post (Post 1994), O’Reilly (O'Reilly et al 2001), Nettle (Nettle 2001) and Horrobin (Horrobin 2001;Horrobin 1998).

In attempting to summarise all of these theories about ultimate causation and the schizophrenic genotype, two central themes emerge. The first is that of the social role of the schizotype. It seems that in one way or another the schizotype, by virtue of his/her particular personality structure, is thought to perform the social function of ‘spacing’ individuals – that is, facilitating and maintaining distance between individuals and between groups. The second theme is that of unusual and perhaps iconoclastic giftedness, without which culture may never have evolved and science and art may never have existed.

Obviously all of this is speculative and limited by its untestable nature but I would contend that there is worth in such speculation, for it challenges our current thinking and may expand our areas of scientific enquiry. Indeed, the creativity in science is in developing ingenious methods of testing what superficially appears unamenable to empirical investigation.

In summary then: those on the continuum with a partial or schizotypal genotype may be adaptively advantaged and responsible for maintaining the genes in the human gene pool; while those with a fuller expression or combination of genes and/or environmental triggers (drugs, life events, head trauma, etc.) may develop schizophrenia. This hypothesis would explain the apparent variability in the clinical presentation of schizophrenia and ‘schizophrenia-spectrum disorders’ (SSD, i.e. a spectrum of clinical presentations that share some features of schizophrenia but insufficient for a diagnosis, and may or may not manifest other symptoms e.g. affective or autistic). Predictions one might make therefore are: 1) that schizotypy/SSD is more common than schizophrenia; 2) that it is especially evident in geniuses; and 3) that it is especially evident in eccentric ‘outcasts’ of society. It is also important to consider that schizotypy may well have been particularly adaptive during the EEA and have lost some of its impact in this very different modern world.

3. What is the schizophrenic phenotype?

Several authors have attempted to address the question of what evolutionary mechanisms might account for the anatomical and cognitive findings in schizophrenia. Tim Crow has pioneered this quest, arguing that schizophrenia results from the evolution of brain asymmetry and specialisation of the language area on the left side (Crow 1997a;Crow 1995a;Crow 1991). The schizophrenia gene is XY-linked and is propagated by sexual selection (Crow 1993). He identifies the corpus callosum as the primary site of pathology (Crow 1997b;Crow 1998b).

While agreeing with Crow that an evolutionary perspective is essential in trying to understand this most human of disorders, I argue that FT dysconnectivity is better supported, relating it to the evolution of metarepresentation and social cognition and subject to the selection pressures of group living.

3.1. A disorder of metarepresentation

The concept of a ‘Theory of Mind’ (ToM) has become increasingly popular in attempting to describe the cognitive deficits in schizophrenia. The term was first used in association with autism, where the inability of these individuals to attribute mental states to others and to “mind-read” was described as an abnormal or absent ToM (Frith & Happe 1994;Kleinman et al 2001;Baron-Cohen 1988).

In recent years, ToM deficits have also been described in people with schizophrenia – the hypothesis being that in these individuals a ToM developed but was then lost with the onset of illness, as opposed to autistic individuals in whom a ToM never developed (Corcoran et al 1995;Frith 1994;Frith & Corcoran 1996;Pickup & Frith 2001;Sarfati et al 1999). The term ‘metarepresentation’, used synonymously with ToM, is probably a better term as it implies the ability to ‘represent mental representations’. When ToM tests are conducted in schizophrenics in conjunction with functional imaging, the areas that correlate with impaired performance are the PFC, the temporal lobes and their intervening connections (Russell et al 2000).

This suggests that FT connections play a critical role in metarepresentation and ToM ability and that it is these connections that are structurally and/or functionally impaired in schizophrenia.

3.2. A disorder of fronto-temporal connectivity

Indeed there is a host of evidence from structural and functional imaging studies, supporting a theory of abnormal fronto-temporal connectivity in schizophrenia. Early work had suggested that the primary defect lay in abnormal functioning of the frontal lobes, the so-called ‘hypofrontality’ hypothesis, but it is now clear that this is task-dependent (Weinberger et al 1986;Andreasen et al 1997;Fletcher et al 1998a). Recent studies demonstrate abnormal fronto-temporal (FT) activations on verbal fluency and verbal memory tasks, especially in the presence of auditory hallucinations, lending support to the hypothesis that the core feature of schizophrenia is a disruption of normal FT integration (McGuire & Frith 1996;Friston et al 1996;McGlashan & Hoffman 2000;Hoffman & McGlashan 1998;Friston & Frith 1995;McGuire et al 1995;Fletcher et al 1998b;Yurgelun-Todd et al 1996b;Lawrie et al 2002).

Furthermore MR Spectroscopy studies show reduced N-acetyl-aspartate (NAA) in the frontal and temporal lobes, suggesting abberant connectivity, but this could be either a functional or structural lack of integrity (Bertolino & Weinberger 1999;Steel et al 2001). With the recent development of Diffusion Tensor Imaging (DTI), which yields information about the structural integrity of white matter connections in the brain, it may be possible to resolve this issue (Buchsbaum et al 1998;Foong et al 2000;Lim et al 1999;Agartz et al 2001). DTI studies to date have demonstrated reduced diffusion anisotropy (a marker of the integrity of white matter connections) in the prefrontal cortex, the corpus callosum and parieto-occipital association regions, although resolution and reliability are as yet poor.

The anatomical structures most commonly implicated in studies of connectivity include the dorso-lateral prefrontal cortex (DLPFC) (Perlstein et al 2001;Frith 1997), the anterior cingulate cortex (ACC) (Fletcher et al 1999;Carter et al 2001;Tamminga et al 2000;Szeszko et al 2000;Nordahl et al 2001;Yucel et al 2002), the fornix (Davies et al 2001), the parahippocampal gyrus (PHG) and the superior temporal gyrus (STG) (Yurgelun-Todd et al 1996a;Frith et al 1995).

3.3. Connectivity and gyrification

Of great interest to this discussion of normal, enhanced and abnormal cognitive development and it’s relation to connectivity is the issue of gyrification. Importantly, gyrification is thought to correlate with connectivity. The degree of gyrification is expressed in terms of the Gyrification Index (GI) which is the ratio of the inner and outer surface contours. In normal ontogeny, the formation of gyri commences early in fetal life, originating from a single subplate lamina, and the GI correlates with brain size (Armstrong et al 1995). The GI shows maximal values over the PFC and the parieto-temporo-occipital association cortex (Zilles et al 1988).

A post-mortem study of Einstein’s brain may provide clues as to the cerebral processes underlying genius (Witelson et al 1999). There was significant enlargement of the gyri comprising the parietal association cortices, suggesting variation at some early stage of cerebral ontogeny. The authors conclude that this may reflect an extraordinarily large expanse of highly integrated cortex within a functional network – a notion consistent with Cajal’s speculation that variation in axonal connectivity may be a neuronal correlate of intelligence (Cajal 1989).

In schizophrenia, there is evidence that gyral patterns are abnormal, where increased GI has been demonstrated in the right PFC (Vogeley et al 2000) and reduced GI demonstrated in the left PFC (Kulynych et al 1997). Interestingly right PFC hypergyria has also been shown in unaffected siblings of schizophrenics (Vogeley et al 2001b), supporting the notion of a genetic continuum.

3.4. Schizophrenia and ‘the social brain’

Leslie Brothers described “the social brain” as the higher cognitive and affective systems in the brain that evolved as a result of increasingly complex social selective pressures (Brothers et al 1990;Brothers 1990). These systems underlie our ability to function as highly social animals and provide the substrate for intact ToM, sociability and affective responsiveness. He suggests that these systems are located in structures including the amygdala, the orbital-frontal cortex (OFC) and the medial temporal lobe. Thus these sites and the connections between them are implicated in disorders of social cognition such as autism and schizophrenia. Brune includes the DLPFC and the anterior cingulate cortex (ACC) in addition to the above areas as candidate areas for social cognition (Brune 2001). Is there support from imaging studies for these areas being identified as constituting ‘the social brain?’

A number of functional imaging studies of normal individuals have demonstrated increased activity in the medial PFC, the ACC, the superior temporal sulcus and the anterior temporal cortex during ToM tests (Vogeley et al 2001a;Fletcher et al 1995;Frith & Frith 1999;Castelli et al 2000;Gusnard et al 2001;McCabe et al 2001). Another fMRI study identified the ACC as playing a central role in the monitoring of performance and conflict monitoring. (In other words, the evaluating and selecting of choices and reactions to stimuli) (MacDonald, III et al 2000). The anterior cingulum bundle (AC) runs within the anterior cingulate gyrus and connects the DLPFC with the PHG (Morris et al 1999;Pandya et al 1981;Petrides & Pandya 1999;Petrides & Pandya 1988;Seltzer & Pandya 1989). Several authors have argued that the anterior cingulum, as an important connecting tract between the frontal and temporal lobes, is centrally involved in working memory (Morris, Pandya, and Petrides 1999), error recognition, adaptive response to changing conditions (Allman et al 2001) and consciousness (Dehaene & Naccache 2001). Interestingly, a neuroethological study using fMRI linked parental and infant separation to the ACC (Lorberbaum et al 1999). Mothers were scanned while listening to recorded infant cries, and the activity demonstrated in the ACC suggests that this structure plays a role in attachment and bonding, arguably the ontogenetic precursors to human sociability.

The OFC has robust connections with the amygdala including the uncinate fasciculus (UF) which constitutes much of the white matter of the anterior temporal stem. The UF runs from the orbital-frontal cortex (OFC) via the anterior temporal stem to the amygdala (Morris, Pandya, and Petrides 1999;Pandya & Yeterian 1996). There is a wealth of evidence for the key involvement of the amygdala and its anterior connections in emotion as well as affective and social responsiveness (Davis 1992;Le Doux 1994;Barbas 2000). The OFC is implicated in Damasio’s ‘somatic marker hypothesis’, an adaptive mechanism by which we acquire, represent and retrieve the values of our actions (Damasio 1994). The OFC generates representations of emotional or somatic states that correspond to the anticipated future outcome of decisions, thus steering the decision making process towards those social outcomes that are advantageous for the individual (Adolphs 1999). Farrow used fMRI to demonstrate the role of the OFC and anterior temporal cortex in the social phenomena of empathy and forgiveness (Farrow et al 2001). Emery argues that eye gaze plays an important signalling role in conveying emotional and mental states between individuals (Emery 2000). In ‘higher primates’ the following and interpretation of gaze is an essential part of ToM ability and social cognition. Several studies have identified neurones in the OFC, the amygdala and the superior temporal sulcus that respond selectively to facial expression and eye gaze (Perrett et al 1985;Perrett et al 1992), while others describe disorders of facial recognition and affective expression following bilateral amygdalotomy (Jacobson 1986). Finally, evidence from autism research implicates the amygdala and its connections in the deficits of affective responsiveness and social cognition that characterise this disorder (Baron-Cohen et al 2000;Baron-Cohen et al 1999).

Conclusions from this discussion of the evolution of the social brain are: firstly that two FT white matter circuits, namely the anterior cingulum bundle (AC) and the uncinate fasciculus (UF), are prominently involved in social cognition and in particular ToM ability and affective responsiveness; and secondly that these circuits underlie the very cognitive and affective abilities that are so disrupted in schizophrenia and autism – thus supporting the hypothesis of specifically impaired fronto-temporal connectivity in schizophrenia.

3.4. A cognitive model of schizophrenia

The symptoms of schizophrenia too are best understood within a cognitive framework resulting from abnormal connectivity in select FT circuits. Positive symptoms, paranoia and referential delusions reflect impaired metarepresentation, while negative symptoms, including avolition, autism, ambivalence and affective blunting reflect deficits of social cognition, motivation and affective modulation.

Thus this model proposes that both these FT circuits are implicated in the genesis of symptoms. Abnormal connectivity in the AC results in ToM deficits, attentional and working memory deficits, ambivalence, minor motor abnormalities, and abnormal social cognition; while abnormal connectivity in the UF results in affective blunting, avolition and autism.

4. Neuropathology of schizophrenia and schizotypy

What pathological process underlies the findings of abnormal connectivity in schizophrenia? If we are arguing that schizophrenia lies on a genetic continuum with schizotypy and schizotaxia (an historical term describing the ‘pre-schizophrenic’ phenotype which may include so-called ‘schizophrenic-spectrum disorders’) and entails a variation in normal white matter connectivity, how could this schizophrenic genetic spectrum translate into altered anatomy?

First we need to briefly consider the normal ontogeny of the brain. A fuller discussion will follow later in this paper. There is a mitotic process in utero (neurogenesis) with migration of neurons to their final sites. Myelination continues in highly connected regions well into late adolescence while arborisation occurs in some areas. Simultaneously there is a normal process of pruning, or apoptosis, that results in fine-tuning of connections. This fine-tuning is necessary for specialisation of cognitive skills and is under genetic and environmental control (Bock & Braun 1999;Bock & Braun 1998;Chechik et al 1998;Casey 1999;Changeux & Danchin 1976).

Regarding schizophrenia one may speculate at which phase/s of ontogeny the genetic defect becomes manifest. Does the schizophrenic genotype disturb normal mitotic division, cell migration, synaptogenesis, myelination (Randall 1998;Randall 1983), arborisation or pruning (Feinberg 1983) or a combination of several phases? It is also possible that the disturbance involves either an increase or decrease in one or more of these processes. Thus abnormal connectivity may for example reflect either reduced pruning of abnormal connections or increased pruning of healthy connections or reduced dendritic arborisation in the ‘right’ places or increased arborisation in the ‘wrong’ places, etc. There are numerous possibilities, and different processes may be implicated to varying extents in particular sub-populations of schizophrenics.

Adolescent onset implies later processes, although Benes describes reduced GABA interneurons and excessive ingrowth of dopaminergic neurons, with augmentation by serotonergic neurons, in specifically the PFC, the hippocampus and the ACC in schizophrenia (Benes 2000). She however identifies abnormal myelination and pruning as the pathological mechanisms that satisfactorily explain the timing or triggering of psychosis.

Horrobin has postulated that a genetic defect in lipid metabolism (possibly of the phospholipase A2 cycle) underlies the schizophrenic/ schizotypal/ creative phenotype, disturbing the normal growth and myelination of neurons (Horrobin 2001;Horrobin 1999;Horrobin 1998). He suggests that a deficiency of essential fatty acids relative to saturated fatty acids, due to both the enzyme defect and to dietary changes occurring during our evolutionary history, underlie the abnormal connectivity in these individuals. However a recent trial of fatty acid therapy, failed to support his hypothesis (Fenton et al 2001)

Finally, McGlashan has used computer-simulated pruning to demonstrate enhanced cognition by means of pruning and refinement (Hoffman & McGlashan 1997;McGlashan and Hoffman 2000). He hypothesises: that with further pruning, certain cognitive skills might be further refined, giving rise to creative genius in, for example, the schizotypal individual, and to possible select genius in the autistic savant; and that schizophrenia may represent an overshoot of the pruning process, resulting in severely abnormal connectivity. This seems a plausible hypothesis and one which is supported in this discussion of schizophrenia.

In summary then, as a result of genetic influence, as well as possible fetal and perinatal insults, there may be a variety of pathological processes occurring. There may well be abnormal migration of neurons, particularly in the PFC, the AC, the UF, the PHG and the amygdala-hippocampal complex (AHC). Exuberant synaptogenesis follows (Innocenti 1995). In late adolescence, delayed myelination and excessive pruning of abnormal connections (under both genetic and environmental control) may take place in these sites. This, in its severest form, gives rise to cognitive deficits that underlie the symptomatology of schizophrenia. A lesser genetic loading in genetically related individuals, results in a downscaled version of the above pathological process. Thus, slightly excess pruning may fine-tune certain circuits giving rise to enhanced creativity and intellect in some and isolated areas of genius in others, with or without the milder array of cognitive deficits and deficits of metarepresentation that characterise the schizotypal individual.

Having described a theory of the genotype (genetics) and anatomical phenotype (pathology) of the schizotypal/ schizophrenic continuum, it is instructive to consider the evolution of the human brain. By examining both phylogenetic and ontogenetic aspects of brain evolution, the author proposes to provide further important evidence for the central argument of this paper – that schizophrenia is a disorder of primarily FT connectivity, a part of the brain that evolved late in human history in relation to increasing pressure for complex social cognition.

5. Evolution of Metarepresentation

Does current theory about how the brain evolved in Homo sapiens and his ancestors lend any support to the hypothesis that schizophrenia is a disorder of predominantly FT connectivity? We have evidence from neuropathology, neuroimaging and neuropsychology. But does evidence from paleoanthropology, paleoanatomy and primatology add any supporting evidence? And does this evidence support Crow’s hypothesis of cerebral asymmetry and language?

Evidence from all these disciplines suggest that cerebral connectivity evolved late in hominid descent.

Humans are genetically very close to the African apes (98,5% of the genome is identical (Allen and Sarich 1988)) and especially close to the two species of chimpanzees (Waddell & Penny 1996). Therefore human cognitive advances are only 5-6 million years old, the date of our last common ancestor, which is very quick by evolutionary standards. Mesulam argues that increased brain size alone could not

accommodate and explain adequately the enormous cognitive advances that occurred during this short period (Mesulam 2000). He argues that it was the evolution of cerebral connectivity that allowed for the huge leaps forward.

 An increase in brain size commenced in the common ancestor of Old and New World monkeys and hominoids approximately 35- 40 million years ago (mya) (Fig.1.) Cladistic analysis suggests that this volume increase occurred under the selective pressures of increased social complexity as a result of group living (Chance & Mead 1953;Byrne & Whiten 1988). Jerison used the encephalization quotient (EQ) as a measure of progressive encephalization, but this does not correlate with group size or increased intelligence (Jerison 1973). Dunbar showed a constant relationship between neocortex ratio (NR) and group size - the former representing the selective increase in neocortex size that marked the clade (i.e. neocortex : rest of the brain), and the latter being a crude measure of social complexity (i.e. the number of relationships the animal has to keep track of) (Dunbar 2001). Monkeys share social skills and increased memory (especially visuo-spatial) for finding fruit and remembering allies (Mackinnon 1978). Thus the first stage in the evolution of intelligence, commenced approx. 35 to 40 mya with an increase in brain size (Byrne 1999).

However it is not just brain size that determines cognitive ability. Byrne has pointed out that apes do not have an increase in NR relative to monkeys, yet apes show increased cognitive abilities (Byrne 2001). He also notes the smaller group size in some apes (e.g. single or mother-child pair in orang-utans) and argues against the extrapolation of Dunbar’s hypothesis to the hominoid super family.

Various authors have argued for the existence of a Theory of Mind (ToM) or elements of ToM in the great apes (Russon 1999;Premack & Woodruff 1978;van Schaik et al 1996). This has been termed Machiavellian Intelligence by de Waal (de Waal 1982) and metarepresentation by Byrne& Whiten (Byrne & Whiten 1991). Byrne argues that apes demonstrate the ability to represent “thoughts” in mind in the absence of a direct stimulus, an ability not found in other primates (Byrne 1999). He terms this ‘representational intelligence’ and cites complex tool use by chimps (McGrew 1992) and orang-utans (van Schaik, Fox, and Sitompul 1996), ability to perform false belief tasks (Byrne & Whiten 1991), complex political dynamics (de Waal 2000;de Waal 1982) and the ability to attribute causality (Limongelli et al 1995) as examples of representational intelligence. He estimates that the origin of this further step in the evolution of mind dates from approximately 16-13 mya when orang-utan ancestors (such as Sivapithecus) split off from the African apes, and was complete by 5 mya when human ancestors split from those of the chimpanzees. He suggests that the biological basis for this cognitive step was an organizational change in the brain, allowing for increased flexibility. Also that the selective pressure for such change in the apes was the need to evolve complex new strategies for food acquisition in order to compete with monkeys who were better adapted for tree climbing. These new strategies include tool-use to extract embedded foods (Parker & Gibson 1977;Parker 1996) and novel ways of manipulating nutritious plant foods, for example nut cracking (Byrne & Russon 1998).

Simon Baron-Cohen argues that our last common ancestor with chimps, 5 mya, could only have possessed immature elements of a ToM (Baron-Cohen 1999). His evidence comes from ToM tests in apes that showed only a limited ability to attribute mental states and intentionality to others (Premack 1988;Povinelli & Eddy 1996). He suggests that our common ancestor may have possessed an “intentionality detector module” and an “eye detector module”, both of which are apparent in chimps. He puts the time frame for the evolution of a full ToM at approx. 150 – 40 000 years ago, supported by archeological records which show the earliest fictional art and symbolic adornments dating from that period (Mithen 1996).

Suddendorf agrees, putting the emergence of the “metamind”, which is first evident during a child’s 4^th year, at approximately 2 million to 100 000 years ago, as evidenced by the complex Acheulian tool culture of Homo ergaster/ Homo erectus (Suddendorf 1999;Suddendorf & Corballis 1997). Unlike the Oldowan tradition of Homo habilis that predated this epoch and was within the scope of modern chimpanzee tool culture, the Acheulian tools required planning, precision and a concept of the future, and implied cultural learning.

Whiten (Whiten 1999), argues that hominids evolved a “deep social mind” as a “cognitive niche” (Tooby & deVore 1987) in order to compete for food with better-adapted monkeys in the trees and carnivores on the savannah during the Pleistocene period. This cognitive advance, which is probably synonymous with Byrne’s ToM and Suddendorf’s metamind, resulted from social interdependence and involved the refinement of cooperative behaviour, cultural and social learning and transmission, and mind-reading ability.

Species Anatomical changes Cognitive changes Evidence

100 thousand years ago	H. sapiens	IBNS + complex fronto-temporal connectivity	Full ToM	- Complex social cognition - Culture, religion, etc
2	H. erectus/ ergaster	IBNS + evolving connectivity	‘Metamind’	- Acheulian tool culture - Symbolic art
5	H. habilis Australopithecus	IBNS + evolving connectivity		- Oldowan tool culture
15	Great Apes	IBNS + evolving connectivity	‘Representational Intelligence’ and early ToM	- Complex tool use - Attribute causality - ToM tasks
30	Old and New World Monkeys	Increasing brain and neocortex size (IBNS)	Increasing memory and social skills	- Group relations - Finding fruit
40 million years ago

Figure. 2. Table illustrating the stages in the evolution of brain and cognition

In summary, all commentators argue for: a) an initial increase in brain and neocortex size under pressures of social living, roughly 40 to 16 mya; and b) an acceleration of evolving cerebral reorganization and connectivity from approximately 16 mya onwards (Fig.2.). This subserved the beginnings of the evolution of metarepresentation in hominoid ancestors between 16 mya and 5 mya, and the later further acceleration leading to a full ToM in the human line between 150 and 40 000 years ago. Two questions remain unanswered: why was there a period of virtual stagnation of brain/mind evolution, as evidenced by both the fossil record and the archeological record, between 5 mya and 200 000 years ago? ; and what gave rise to the great spurt in brain/mind evolution after this period?Is Crow right about a genetic mutation (Crow 1995a), which gave rise to cerebral asymmetry and the evolution of language, being the driving force for increased brain size, cognitive advances, and the vulnerability to psychosis? Or is Horrobin (Horrobin 1999) right that a genetic mutation altered lipid metabolism in some individuals, driving both cognitive advance and a capacity for schizophrenia?

I would argue that it was only with the recent evolution of cerebral connectivity and a capacity for a full ToM and complex social cognition, during the period 150 to 40 000 years ago, that a disorder of cortical connectivity such as schizophrenia was possible. Conversely, the lack of evidence to my knowledge of psychotic disorders in other animals including our nearest relatives the chimps, implies that it is precisely what sets us apart from them in terms of brain structure and psychology, that also underlies our capacity for this specifically human malady. That is our possession of complex cerebral connectivity that allows for full metarepresentation.

6. Evidence from comparative primate anatomy

Recent comparative studies of primate brain anatomy, add further support to the thesis of this paper – that schizophrenia represents a disorder of connectivity that evolved late in hominid ancestry. In contrast, the author would argue that these new findings conflict with Crow’s theory. Like Crow and Horrobin, I would argue that schizophrenia is indeed ‘the price we have paid for being fully human’, but for different reasons.

6.1. Brain size and surface area

Semendeferi has imaged the primate brain using MRI and has demonstrated that with increasing brain size, the frontal lobe does not increase relative to total hemispheric size in hominoids (Semendeferi 2001;Semendeferi et al 2001;Semendeferi et al 1997;Semendeferi et al 2002) (Fig.3.). Likewise, the parieto-occipital lobes enlarge consistently relative to total hemispheric size. However, the temporal lobe does increase (but without statistical significance) as one moves from the gibbon brain to the orang-utan to the African apes and finally to the human brain. Interestingly, the size of the cerebellum is progressively reduced (with significance) as one moves from the phylogenetically older apes to humans (Semendeferi & Damasio 2000;Rilling & Insel 1998). A comparison of sulcal patterns shows continuity in most sulci throughout the hominoids (Semendeferi et al 1997). Studies comparing the GI between primates supports this, with continuity in GI demonstrated throughout the anthropoids (monkeys and hominoids), but reduced gyrification for every unit reduction in brain weight and volume (particularly neopallial) evident in the prosimians (Zilles et al 1989;Zilles et al 1988). Armstrong and colleagues make the point that the sizes of subcortical axonal bundles (i.e. white matter volume) are not directly associated with the degree of cortical folding (Armstrong et al 1991).

Figure. 3. Allometric relationship between the frontal lobe and the cerebral hemispheres (reproduced from Semendeferi (2001) with kind permission.

Conclusions one might draw from this data, which enlighten the subject of this paper, are: 1) that it is not a relative increase in frontal or parieto-occipital lobe size, nor a substantive change in relative cortical area, that underlies the acceleration in cognitive function and the development of the metamind in the hominid line; and 2) that some feature of the human brain other than cortical size and volume is implicated in metacognition.

6.2. Fronto-temporal connectivity

Holloway (Holloway 1966;Holloway 1974) first raised the argument that evolutionary changes in cognition reflect reorganization of systems internal to the brain, rather than increased brain size as championed by Jerison (Jerison 1973). Semendeferi has demonstrated in her MRI studies of hominoids that it is white matter that increases substantially (relative to brain size) rather than grey matter (Semendeferi 2001). Furthermore, it is specifically intrahemisheric connectivity that increases disproportionate to increasing brain size and neocortical surface area. Conversely, interhemispheric connectivity, as expressed by the cross-sectional area of the corpus callosum, decreases with increasing brain size (Rilling & Insel 1999).

This leads us to consider comparative data on the specific FT tracts we have identified, namely the AC and the UF. There is recent evidence from a comparative primate study that the ACC evolved a unique type of projection neuron in the hominoid clade (Hof et al 2001;Nimchinsky et al 1999). This large spindle-shaped cell which is characterised by immunoreactivity to the calcium-binding protein, calretin, is unique to hominoids and increases in density as one compares the ACC of the orang-utan, with that of the gorilla, with that of the chimpanzee and finally is greatest in humans. Nimchinsky and colleagues argue that this indicates that the ACC experienced strong adaptive pressure related to communication during the past 16 million years of primate evolution. They conclude that the ACC plays a significant role in recently evolved cognitive processes including self-awareness, attention, emotional control and communication.

A comparison of the macroscopic and microscopic morphology of the OFC in great apes supports the notion that this area is strongly implicated in social cognition. Semendeferi compared Brodmann areas 10 and 13 across hominoids and demonstrated that area 13 is significantly smaller in orang-utans than in gorillas and chimpanzees (and humans) (Semendeferi 1994). Area 13 lies posteriorly and medially in the OFC and is considered to be part of a circuit connecting to the limbic temporal lobe that is relevant to emotion, particularly related to social stimuli. Ablation of this area in wild monkeys results in significant reductions and losses of behaviours that are considered important for the maintenance of social bonds (Kling & Steklis 1976). Furthermore, in terms of the cytoarchitecture of area 13, Semendeferi has demonstrated a marked decrease in cortical cell density in the orang-utan relative to the African hominoids, especially in infragranular layers V and VI (which have connections with subcortical limbic structures). Thus it appears that there is decreased representation of the ‘limbic’ OFC in the orang-utan, a phylogenetically older species. She suggests that this region is important for the survival of members of complex social groups and speculates that the relative immaturity of the frontal limbic cortex in orang-utans may relate to the more solitary life style and less complex social organization of this primate compared with it’s African cousins (Semendeferi 1999;van Schaik & Van Hoof 1996). Further speculation might suggest that the OFC has, like the ACC, experienced strong adaptive pressures related to social living during the course of hominoid evolution.

The internal organization of cortical grey matter shows variability throughout the cortex between hominoids. While there seems to be continuity in terms of relative cortical area, (with the notable exception of the greatly reduced size of the visual cortex in humans (Holloway 1996)), there is a wealth of evidence for significant diversity in cortical organization (Preuss 2001;Preuss 2000). There are variations in the morphologies of neurons, in the connections between layers and in the laminar distribution of neurotransmitters and receptors within homologous areas.

Realistic conclusions from this data are: firstly that increasing intrahemispheric white matter connectivity (e.g. fronto-temporal tracts including the AC and the UF) and the related reorganization of cortical grey matter, can be correlated with the evolution of metacognition; and secondly that the corpus callosum is not specifically related to the evolution of complex cognition and a capacity for psychosis as Crow hypothesises.

6.3. Language and the brain

Another set of research findings from comparative studies of ‘language areas’ in extant primates further contradict Crow’s hypothesis. Gannon and colleagues reported in Science their discovery of marked asymmetry in the chimpanzee planum temporale (PT) – a key site in Wernicke’s posterior language area (Gannon et al 1998b) (Fig.4.). They found that the left PT was significantly larger in 94% (17 of 18) of chimpanzee brains examined post-mortem and they state:

“The evolutionary origin of human language may have been founded on this basal anatomic substrate, which was already lateralized to the left hemisphere in the common ancestor of chimpanzees and humans 8 million years ago.”

Two further studies, one post-mortem (Gannon et al 1998a) and one using MRI to image the brains of a variety of sedated primates (Hopkins et al 1998), demonstrate asymmetry of both the PT and Heschl’s gyrus in the great apes but not in the lesser apes or monkeys.

Another recent study by Cantalupo and colleagues, reported in Nature, demonstrates left-right asymmetry of Broca’s area (Brodmann’s area 44) in three great ape species, Pan troglodytes, Pan paniscus and Gorilla gorilla (Cantalupo & Hopkins 2001).

Gannon argues that these findings set the date for the origins of a lateralised ‘proto-linguistic’ area in great apes and humans, at approximately 16 to 18 mya just after the gibbon ancestor diverged from that of the other hominoids (Gannon, Kheck, & Hof 2001).

Furthermore he argues for a polymodal role for the PT in a connectionist model of ‘language’ perception. In other words, a diffuse network of lateralised neuronal connections corresponding to the left PT and related association areas, constitute a region underlying communicative skills in great apes and humans. He cites the following findings in support of his argument:

a) The complex communicative skills of great apes, including both referential and intentional gesturing and vocalisation (Corballis 1992) and their use of sign language (Savage-Rumbaugh 1990;Savage-Rumbaugh et al 1978;Shapiro & Galdikas 1999;Shapiro 1982).

b) Evidence from functional imaging of deaf-from-birth humans that signing activates classic left hemisphere language areas (Neville et al 1998).

c) Auditory hallucinations in psychotic individuals activate language areas without discernible motor or audible components (Suzuki et al 1993).

There are strong arguments from ‘neural network theory’ against Chomsky’s idea of a domain-specific, innate “human language organ”