Background
Recent theoretical models of schizophrenia posit that dysfunction of the neural mechanisms subserving predictive coding contributes to symptoms and cognitive deficits, and this dysfunction is further posited to result from N-Methyl D-aspartate glutamate receptor (NMDAR) hypofunction. Previously, by examining auditory cortical responses to self-generated speech sounds, we demonstrated that predictive coding during vocalization is disrupted in schizophrenia. In order to test the hypothesized contribution of NMDAR hypofunction to this disruption, we examined the effects of the NMDAR antagonist, ketamine, on predictive coding during vocalization in healthy volunteers and compared them to the effects of schizophrenia.
Methods
In two separate studies, the N1 component of the event-related potential (ERP) elicited by speech sounds during vocalization (Talk) and passive playback (Listen) were compared to assess the degree of N1 suppression during vocalization, a putative measure of auditory predictive coding. In the cross-over study, 31 healthy volunteers completed two randomly ordered test days, a saline day and a ketamine day. ERPs during the Talk/Listen task were obtained pre-infusion and during infusion on both days, and N1 amplitudes were compared across days. In the case-control study, N1 amplitudes from 34 schizophrenia patients and 33 healthy controls were compared.
Results
N1 suppression to self-produced vocalizations was significantly and similarly diminished by ketamine (Cohen's d=1.14) and schizophrenia (Cohen's d=.85).
Conclusions
Disruption of NMDARs causes dysfunction in predictive coding during vocalization in a manner similar to the dysfunction observed in schizophrenia patients, consistent with the theorized contribution of NMDAR hypofunction to predictive coding deficits in schizophrenia.