Ketamine Affects Prediction Errors about Statistical Regularities: A Computational Single-Trial Analysis of the...

Area of Science:

• Neuroscience
• Computational Psychiatry
• Cognitive Neuroscience

Background:

• Auditory mismatch negativity (MMN) reduction is observed in schizophrenia and with NMDA receptor antagonist (NMDAR) use.
• These reductions are linked to impaired predictive coding and hierarchical Bayesian inference in the brain.
• NMDARs are implicated in minimizing prediction errors (PEs) within hierarchical models.

Purpose of the Study:

• To provide empirical evidence for hierarchical Bayesian theories of schizophrenia.
• To investigate the role of NMDARs in processing hierarchical prediction errors (PEs).
• To examine the effects of S-ketamine on neural responses related to PEs.

Main Methods:

• Utilized a roving auditory mismatch negativity (MMN) paradigm.
• Applied a hierarchical Bayesian model to single-trial EEG data.
• Administered S-ketamine or placebo in a double-blind, within-subject design to healthy volunteers.

Main Results:

• Identified distinct neural expressions of low-level (stimulus transitions) and high-level (transition probability) PEs.
• Low-level PEs were observed early (102-207 ms), while high-level PEs appeared later (152-199 ms and 215-277 ms).
• S-ketamine significantly reduced the neural expression of high-level PEs, disrupting inference on abstract regularities.

Conclusions:

• NMDAR antagonism impairs hierarchical Bayesian inference of the world's statistical structure.
• Findings support the role of NMDAR dysfunction in schizophrenia's predictive coding deficits.
• Computational single-trial EEG analysis offers a method to assess pathophysiological mechanisms.