While it is known that patients with schizophrenia recognize facial emotions, specifically negative emotions, less accurately, little is known about how they misattribute these emotions to other emotions and whether such misattribution biases are associated with symptoms, course of the disorder, or certain cognitive functions.

Outpatients with schizophrenia or schizoaffective disorder (n = 73) and healthy controls (n = 30) performed a computerised Facial Emotion Attribution Test and Wisconsin Card Sorting Test (WCST). Patients were also rated on the Positive and Negative Syndrome Scale (PANSS).

Patients were poor at recognizing fearful and angry emotions and attributed fear to angry and angry to neutral expressions. Fear-as-anger misattributions were predicted independently by a longer duration of illness and WCST perseverative errors.

The findings show a bias towards misattributing fearful and angry facial emotions. The propensity for fear-as-anger misattribution biases increases as the length of time that the disorder is experienced increases and a more rigid style of information processing is used. This, at least in part, may be perpetuated by subtle fearfulness expressed by others while interacting with people with schizophrenia.

Schizophrenia has been associated with limited abilities to interact effectively in social situations. Face perception and ability to recognise familiar faces are critical for social interaction. Patients with chronic schizophrenia are known to show impaired face recognition. Studying first-episode (FE) patients allows the exclusion of confounding effects of chronicity, medication and institutionalisation in this deficit.

To determine brain (dys)functions during a face encoding and recognition paradigm in FE schizophrenia.

Thirteen antipsychotic-naïve FE schizophrenia patients and 13 age- and sex-matched healthy controls underwent functional magnetic resonance imaging during a face encoding and recognition paradigm. Behavioural responses were recorded on line.

Patients recognised significantly fewer of previously presented faces than the controls (p = 0.008). At the neural level, both groups activated a network of regions including the fusiform area, occipital, temporal and frontal regions. In brain activity, the two groups did not differ in any region during encoding or recognition conditions (p > 0.05, corrected or uncorrected).

Our findings show impaired face recognition without a significant alteration of related brain activity in FE schizophrenia patients. It is possible that neural changes become more strongly evident with progression of the illness, and manifest themselves as behavioural impairments during the early course.