Quantifying the Latent-Cause Inference Process and Its Relationship With Schizotypy

Abstract

Latent-cause inference is a process through which humans determine the hidden (latent) causes of events. Aberrant latent-cause inference could lead to suboptimal interpretation of causality, characteristic of certain psychopathologies like schizotypy or schizophrenia-spectrum disorders. In this study, data from N=565 subjects were analyzed. Subjects responded to a subset of questions from the Personality Inventory for DSM-5 that assessed schizotypal traits, and completed the “Microbes Task” – a clustering task developed to quantify latent-cause inference in humans. Four individual-level parameters were derived by fitting a Bayesian model of latent-cause inference to the Microbes Task data. It was hypothesized that higher schizotypy scores would be positively associated with a higher tendency to create new clusters, as well as a higher propensity to create wider clusters, indicative of a fractured and incoherent interpretation of the causal relationships between events. This hypothesis was borne out when schizotypy scores were treated as a categorical variable. However, when the schizotypy scores were analyzed as a continuous variable, only the propensity to create wider clusters was found to correlate significantly with schizotypy scores. Regression analysis was performed on subscale scores, yielding significant correlations between at least one latent-cause inference parameter and the subscales “Suspiciousness,” “Perceptual dysregulation,” and “Unusual beliefs & experiences.” These are the three subscales that assess “positive” schizotypal traits, which correspond to the positive symptoms of schizophrenia. These findings suggest that schizotypy and other conditions with a proneness to psychosis may involve carving the world into too many clusters or latent causes, offering a new theoretical understanding of cognitive deficits in schizophrenia-spectrum disorders.