Cognitive predictors of violent incidents in forensic psychiatric inpatients.

This study tested the predictive value of attentional bias, emotion recognition, automatic associations, and response inhibition, in the assessment of in-clinic violent incidents. Sixty-nine male forensic patients participated and completed an Emotional Stroop to measure attentional bias for threat and aggression, a Single Target - Implicit Association Task to assess automatic associations, a Graded Emotional Recognition Task to measure emotion recognition, and an Affective Go/NoGo to measure response inhibition. Violent incidents were derived from patient files and scored on severity level. The predictive value of level of psychopathy was tested with the Psychopathy Checklist - Revised (PCL-R). Generalized linear mixed model analyses showed that increased attention towards threat and aggression, difficulty recognizing sad faces and factor 2 of the PCL-R predicted the sum of violent incidents. Specifically, verbal aggression was predicted by increased attention towards threat and aggression, difficulty to recognize sad and happy faces, and PCL-R factor 2; physical aggression by decreased response inhibition, higher PCL-R factor 2 and lower PCL-R factor 1 scores; and aggression against property by difficulty recognizing angry faces. Findings indicate that cognitive tasks could be valuable in predicting aggression, thereby extending current knowledge on dynamic factors predicting aggressive behavior in forensic patients.