Operationalising “Theory of Mind”: The Psychometrics of Mentalising in Schizophrenia

Abstract

Schizophrenia is a serious mental illness often accompanied by disabling impairments in neurocognition and social cognition, the latter including deficits in “theory of mind” (ToM). Since Frith (1992) first proposed the presence of this deficit, a proliferation of tasks have been used to assess ToM in people with schizophrenia, despite little being known about their psychometric qualities. This thesis aimed to provide the first focused investigation of the psychometric properties of such measures incorporating a schizophrenia sample. Six tasks (containing 11 indices of ToM) were selected for evaluation based on frequency of use, and coverage of the different dimensions of ToM proposed by Bell, Langdon, Siegert, and Ellis (2010): Frith and Corcoran’s (1996) False Belief and Deception task (FBDT); Langdon and Coltheart’s (1999) False Belief Picture Sequencing Task (FBPST); Brüne’s (2003) Picture Sequencing Task (BPST); Corcoran Cahill, and Frith’s (1995) Hinting task, administered alongside Marjoram, Gardner et al.’s (2005) alternate version to produce an “Extended” Hinting task; Sarfati, Hardy-Bayle, Nadel, Chevalier, & Widlocher’s (1997) Character Intention task (SCIT); and Baron-Cohen, Wheelwright, Hill, Raste, and Plumb’s (2001) Reading the Mind in the Eyes task (RMET). All 102 participants (42 patients, 60 healthy controls) were administered the ToM tasks, alongside measures of psychiatric symptoms, neurocognition, and social/interpersonal functioning. Sixteen participants from each group were reassessed on the ToM and symptom measures following one (patients) and four (controls) weeks. Psychometric evaluation criteria were developed from those used in investigations of measures in other fields (Terwee et al., 2007). Face validity and utility were convincing for all tasks. Content validity was consistently strong across measures, though a high reading level for the RMET suggested this might also tap comprehension of mental state language; and two tasks (the FBDT and BPST) aggregated different facets of social cognitive mentalising. Floor effects were not observed, though ceiling effects were noted on three ToM indices for patients. Ceiling effects were marked for controls across all ToM indices except the RMET, BPST ToM Total score, and Extended Hinting task. Consequently, correlational analyses for controls were less meaningful than for patients. Controls scored at significantly higher levels on all ToM tasks except the SCIT, and all indices (except the FBDT first-order score) showed discriminative validity in ROC analyses, with the Hinting tasks most discriminating. Construct validity was displayed in a hierarchical pattern of significant correlations where strongest relationships were observed between the ToM tasks, followed by those between ToM and neurocognitive measures. Anticipated correlations between ToM tasks and indices of social/interpersonal functioning were not observed, likely due to the lack of informant-based information regarding those variables. While reliability was generally acceptable across ToM tasks (though compromised by ceiling effects and short scale lengths for some measures), stability was largely indeterminate due to the small retested sample. Overall, the results provide encouraging preliminary evidence for the psychometric robustness of those ToM tasks reviewed, and a reference point to inform the refinement of evaluation criteria and study design of future psychometric studies. While limitations of the project are acknowledged, recommendations are made around future applications of the tasks.