Abstract
BACKGROUND: People with schizophrenia exhibit deficits in social and non-social cognition and these deficits are associated with difficulties in everyday functioning. The best cognitive predictor of functional difficulties in schizophrenia is theory of mind (ToM), the ability to infer the mental states of others. ToM is thus an important treatment target to promote functioning in schizophrenia. To improve ToM, it is however necessary to understand its underlying processes. Associations have been reported between ToM and different non-social cognitive abilities in several studies, but the cognitive processes that are most closely linked to ToM remain to be identified. The goal of this meta-analysis were: (1) to determine and compare the magnitude of the associations between ToM and each non-social cognition domain, and (2) to test the effect of different characteristics of the ToM tasks and of characteristics of the patient samples on the magnitude of the associations between ToM and non-social cognition. METHODS: The following inclusion criteria were used: 1) participants with a diagnosis of schizophrenia or schizoaffective disorder; 2) age 18–65; 3) publication in English, French or Spanish; 4) published or Epub before January 2018; 5) reporting a correlation between the performance on a ToM task and at least one non-social cognition task. Each correlation between a ToM task and a non-social cognition task was transformed into an effect size Zr and a weighted mean was calculated for the association between ToM and each cognitive domain. These associations were compared using focused test. RESULTS: A total of 86 studies (88 samples) were included (4 508 participants, 65% male, mean age 32.8 years). When all studies were combined regardless of the type of cognitive domain, a moderate association was observed (k=88, r=0.32). The associations between ToM and each cognitive domain were as follow: attention (k=17, r=0.27), working memory (k=27, r=0.29), episodic memory (k=24, r=0.28), speed of processing (k=24, r=0.26), language (k=31, r=0.36), visuospatial/problem solving (k=23, r=0.30), executive functions (k=61, r=0.29), early processing (k=6, r=0.32) and autobiographical memory (k=2, r=0.40). The comparison between the domains revealed no significant difference (χ2(8) =12.71, p = 0.122). The heterogeneity test was however significant for the executive function domain (χ2(60)=83.62, p=0.024), and we thus further compared the different executive functions together. This revealed that abstraction was more strongly associated with ToM than the other executive functions. For the effect of different characteristics of the ToM tasks, a significant effect was observed for visuospatial/problem solving and executive functions regarding the difference between ToM tasks (χ2(20) = 9.8, p = 0.040; χ2(64) = 22.5, p = 0.007), as well as for characteristics of the required answers including modality (i.e. verbal) (χ2(29) = 14.4, p = 0.001; χ2(77) = 19.4, p = 0.001) and type (i.e. multiple choices) (χ2(30) = 16.9, p = 0.001; χ2(84) = 21.7, p = 0.001). A significant effect was also observed for attention and episodic memory regarding the type of mental state (χ2(20) = 7.3, p = 0.026; χ2(30) = 9.2, p = 0.027). As for the characteristic of the sample, duration of illness (r = 0.27, p = 0.033), age at onset of the illness (r = 0.32, p = 0.035) and the total score on the PANSS (r = 0.52, p = 0.003) were significantly correlated to the strength of the association between ToM and non-social cognition. DISCUSSION: Non-social cognition deficits, particularly abstraction, and characteristics of ToM tasks should be considered when assessing and treating ToM deficits in people with schizophrenia in order to have a comprehensive understanding of ToM performance for each patient.