Neurodevelopmental models of schizophrenia
Over the last decade the concept of schizophrenia as a neurodevelopmental disorder has taken a strong hold,  although ‘neurodevelopmental’ is often used with a wide range of meanings. It is possible to distinguish ‘early’ and ‘late’ neurodevelopmental models, with a third ‘risk’ model incorporating ideas from developmental psychopathology (Hollis & Taylor 1997).

The ‘early’  neurodevelopmental model emerged from the ideas of Fish (1957) who proposed that the neuropathology in schizophrenia was of perinatal origin. The ‘early’ neurodevelopmental model views the primary cause of schizophrenia as a static ‘lesion’ occurring during fetal brain development (Murray & Lewis 1987; Weinberger 1987). The putative ‘lesion’ could be of either neurogenetic or environmental origin (e.g. virus infection or fetal hypoxia). Two main lines of evidence support the ‘early’ neurodevelopmental model. First, there is an absence of gliosis in the postmortem brains of schizophrenic patients, which suggests a neurodevelopmental rather than a neurodegenerative pathology. Secondly, a more indirect line of evidence includes the association of schizophrenia with premorbid social and cognitive impairments (Foerster et al. 1991; Done et al. 1994; Jones et al. 1994), pregnancy and birth complications (Lewis &  Murray 1987;  McNeil 1995)  and minor physical anomalies (Gualtieri et al. 1982; Guy et al. 1983). According to this ‘early’  model,  during childhood the ‘lesion’  is relatively silent giving rise only to subtle behavioural symptoms (premorbid social and cognitive impairments). However, in adolescence, or early adult life, the ‘lesion’ interacts with the process of normal brain maturation (e.g. myelination of corticolimbic circuits and/or synaptic pruning and remodelling) to manifest itself in the form of psychotic symptoms.

There are several weaknesses in the ‘early’ neurodevelopmental model. First, it fails to provide a satisfactory account of the long latency between the putative perinatal damage/lesion and the typical onset of symptoms in late adolescence or early adult life. Secondly, an early neurodevelopmental insult on its own cannot account for the finding of increased extracerebral (sulcal) cerebrospinal fluid (CSF) space in schizophrenia. Diffuse loss of brain tissue limited to the pre- or perinatal periods would result in enlargement of the lateral ventricles but not increased extracerebral CSF space (Woods 1998).

The ‘late’  neurodevelopmental model,  first proposed by Feinberg (1983, 1997), argues that the key neuropathological events in schizophrenia occur as a result of abnormal brain development during adolescence. The current formulation of the ‘late’ neurodevelopmental model proposes that excessive synaptic and/or dentritic elimination occurs during adolescence producing aberrant neural connectivity and psychotic symptoms (Woods 1998; McGlashen & Hoffman 2000). This ‘late’ model characterizes schizophrenia as a progressive late onset neurodevelopmental disorder in contrast to the ‘early’ model that proposes a static lesion during the perinatal period. The ‘late’ model predicts that progressive structural brain changes and cognitive decline will be seen in adolescence around the onset of psychosis.

The excessive synaptic pruning during adolescence proposed in the ‘late’  model is simply an amplification of the normal process of neuronal remodelling with progressive pruning and elimination of synapses that begins in early childhood and extends through late adolescence (Huttenlocher 1979; Purves &  Lichtmen 1980). These major regressive changes in adolescence with remodelling of neural connections are likely to be under genetic control with synaptic elimination in schizophrenia representing an extreme of normal variation (Feinberg 1983). In the ‘late’ model, premorbid abnormalities in early childhood are viewed as non-specific risk factors rather than early manifestations of an underlying schizophrenic neuropathology.

Both the ‘early’  and ‘late’  models suppose that there is a direct and specific expression of the eventual brain pathology as schizophrenic disorder.  A third viewpoint,  the ‘risk’  model,  proposes that early and/or late brain pathology acts as a risk factor rather than a sufficient cause so that its effects can only be understood in the light of an individual’s exposure to other risk and protective factors (Hollis & Taylor 1997). This latter formulation provides a probabilistic model of the onset of schizophrenia in which aberrant brain development is expressed as neurocognitive impairments that interact with the environment to produce psychotic symptoms.  The following sections examine how well current neurobiological research evidence supports these competing neurodevelopmental models of schizophrenia.