Fig 1.
Study design.
Fig 2.
(A) The auditory MMN task consists of a train of alternating tones of different frequencies, and the probability of the tones is systematically varied with a given probabilistic schedule including alternating stable and volatile phases. (B) Three-level hierarchical Gaussian filter (HGF) binary perceptual model, a Hierarchical Bayesian Model, is used to model the effect of ketamine on MMN.
Fig 3.
Predicting antidepressant ketamine effects in TRD using a conductance-based neural mass model.
This flowchart provides a step-by-step approach to modeling ketamine effects of perceptual inference using the auditory MMN paradigm and EEG recordings. 1) The auditory MMN, which occurs when a sequence of low tones (predictable stimulus or standard) is unexpectedly interrupted by a high tone (unpredictable stimulus or deviant). 2) The second step includes specifying the generative model, which embodies a probabilistic forward mapping from hidden brain states to observed auditory MMN. 3) The model can be inverted and applied to data for inferring underlying individual pathophysiology. (3-population neural model, with inhibitory interneuron, spiny stellate cell, and pyramidal cell populations in layer IV cortical column drawn using Adobe Illustrator). For illustration purposes, intrinsic glutamatergic connections mediated by the AMPA and NMDA receptors are grouped. 4) The utility of these models for detecting physiologically defined subgroups will be tested in TRD patients.