Fig. 2: The generative canonical microcircuit conductance model with NMDA channel blockade parameters and the mechanism of memantine.
a The intrinsic connectivity between cell populations within each region of the model. The NMDA switch function (Eq. 2) plotted (b) against the NMDA blockade parameter shown for increasing voltage values (from −70–0 V in 10 V steps) and (c) against voltage shown for increasing values of the exponential of the NMDA blockade parameter \({{{\boldsymbol{blk}}}}_{{{\boldsymbol{NMDA}}}}\) (−1, −0.5, 0, 0.5, 1, 2 and 4). The dashed line shows the NMDA switch function with the blockade parameter value set at the default value from the original model [82]. As the blockade parameter increases, the magnesium switch function output, which scales NMDA channel conductance, reduces. d Free energy and posterior probabilities of PEB models explaining the effect of memantine versus placebo. The PEB analysis with the NMDA channel blockade parameters had the highest posterior probability for explaining differences between neurophysiological mismatch negativity responses on placebo versus drug; memantine acts primarily on the NMDA blockade parameter. e Memantine increases the NMDA channel blockade parameter with a posterior probability >95% in the left parietal cortex. Lines are weighted by each subject’s average precision of NMDA blockade over sessions. Sup., superficial; stell., stellate; inter., interneuron; m(V), the switch function output; blkNMDA, the NMDA blockade parameter; V, voltage; AMPA-T, AMPA channel time constant; GABA-T, GABA channel time constant; NMDA-T; NMDA channel time constant; NMDA-Blk, NMDA channel blockade; All (AGN); AMPA, GABA and NMDA time constants and NMDA blockade parameters.
