Error Patterns in Progressive Matrices in Schizophrenia and Bipolar Disorders

Conventional evaluations of matrix reasoning have focused predominantly on total accuracy scores, thus overlooking qualitative error patterns that may reflect distinct cognitive mechanisms in psychiatric populations. This study examined performance on MatriKS, a digital matrix reasoning task, in a sample of 28 individuals diagnosed with Schizophrenia, 31 individuals diagnosed with Bipolar disorder, and 371 healthy controls. The present study analysed four conceptual error types: repetition (R), wrong principle (WP), distractor (D), and incomplete correlate (IC). The results showed that both clinical groups exhibited a lower level of accuracy and a higher total number of errors when compared to the control group. However, these differences were not statistically significant. The qualitative error analysis revealed informative dissociations: D errors, which reflect the selection of visually salient but relationally irrelevant options, were found to be more prevalent in both clinical groups, with a tendency observed in individuals with Schizophrenia. Multinomial logistic regression with bootstrap resampling identified selective visual attention, measured by the Attentive Matrices Test, as the only significant cognitive predictor of reasoning errors. It was also found that higher visuospatial attentional capacity significantly reduced WP errors and showed a marginal protective effect against IC errors. Furthermore, the study revealed that there were no significant interactions between cognitive measures and diagnostic group. Symptom severity, as measured by the Brief Psychiatric Rating Scale, did not demonstrated correlations with total errors. The overall findings of this study underscore the clinical value of qualitative error profiling as a complement to total score metrics, and highlight selective visual attention as a key cognitive mechanism in matrix reasoning.