Background. Several studies have examined the ability of schizophrenic patients to represent mental states (‘theory of mind'; ToM). There is consensus that some patients have impaired ToM, but there is disagreement about the relation between ToM and symptomatology, and about the severity and specificity of the deficit.

Methods. Two first-order and one second-order false belief tests of ToM were given to groups of schizophrenic patients and psychiatric and normal controls. The relation between ToM and symptomatology was explored using regression and symptom subgroup analyses. Severity was investigated by using the same task methodology as in autism research, to enable direct comparison with that disorder. Specificity was investigated using matched control tasks which were as difficult as the ToM tasks, but did not require ToM.

Results. Symptom subgroup analysis showed that schizophrenic patients with behavioural signs were impaired relative to controls on ToM, and that remitted patients and a single case with passivity symptoms performed as well as controls. Regression analysis showed that ratings of behavioural signs predicted impaired ToM in schizophrenia. There was weak evidence that a subgroup with paranoid symptoms had ToM impairments, although these were associated with low IQ. Schizophrenic patients only showed ToM deficits on the second-order task. No impairments appeared on the matched control tasks which did not require ToM.

Conclusions. There is a clear association between ToM impairment and behavioural signs in schizophrenia. Deficits in paranoid patients are harder to detect with current tasks and may be compensated for by IQ-dependent problem-solving skills. ToM impairments in schizophrenia are less severe than in autism, but are specific and not a reflection of general cognitive deficits.