UCSF

Auditory hallucinations, a hallmark symptom of psychosis, are experienced by most people with a diagnosis of schizophrenia at some point in their illness. Auditory hallucinations can be understood as a failure in predictive coding, whereby abnormalities in sensory/perceptual processing combine with biased cognitive processes to result in a dampening of normal prediction error signaling. In this paper, we used a roving mismatch negativity (MMN) paradigm to optimize evaluation of prediction error signaling and short-term neuroplasticity in 30 people with schizophrenia (n = 16 with and n = 14 without recent auditory hallucinations) and 20 healthy comparison participants. The recent hallucinations group exhibited an abnormal roving MMN profile [F(2,27) = 3.98, p = 0.03], significantly reduced prediction error signaling [t(28) = -2.25, p = 0.03], and a trend for diminished short-term neuroplasticity [t(28) = 1.80, p = 0.08]. There were no statistically significant differences between the healthy comparison group and the combined schizophrenia group on any of the roving MMN indices. These findings are consistent with a predictive coding account of hallucinations in schizophrenia, which posits reduced prediction error signaling in those who are prone to hallucinations. These results also suggest that plasticity-mediated formation and online updating of predictive coding models may also be disrupted in individuals with recent hallucinations.