Antisaccade performance in schizophrenia: a neural model of decision making in the superior colliculus

Antisaccade performance deficits in schizophrenia are generally interpreted as an impaired top–down inhibitory signal failing to suppress the erroneous response. We recorded the antisaccade performance (error rates and latencies) of healthy and schizophrenia subjects performing the mirror antisaccade task. A neural rise-to-threshold model of antisaccade performance was developed to uncover the biophysical mechanisms giving rise to the observed deficits in schizophrenia. Schizophrenia patients displayed greater variability in the antisaccade and corrected antisaccade latency distributions, increased error rates and decreased corrected errors, relative to healthy participants. Our model showed that (1) increased variability is due to a more noisy accumulation of information by schizophrenia patients, but their confidence level required before making a decision is unaffected, and (2) competition between the correct and erroneous decision processes, and not a third top-down inhibitory signal suppressing the erroneous response, accounts for the antisaccade performance of healthy and schizophrenia subjects. Local competition further ensured that a correct antisaccade is never followed by an error prosaccade.