Fig. 4.

Simulated electrophysiological responses during exploration: this figure reports the simulated electrophysiological responses during the epistemic search of the previous figure. Upper panel: this panel shows the activity (firing rate) of units encoding the expected location—over 32 trials—in image (raster) format. There are $192=64\times 3$ units for each of the 64 locations over the three epochs between two saccades that constitute a trial. These responses are organised such that the upper rows encode the probability of alternative states in the first epoch, with subsequent epochs in lower rows. The simulated local field potentials for these units (i.e. log state prediction error) are shown in the middle panels. Second panel: this panel shows the response of the first hidden state unit (white line) after filtering at 4 Hz, superimposed upon a time–frequency decomposition of the local field potential (averaged over all units). The key observation here is that depolarisation in the 4 Hz range coincides with induced responses, including gamma activity. Third panel: these are the simulated local field potentials (i.e. depolarisation) for all (192) hidden state units (coloured lines). Note how visiting different locations evokes responses in distinct units of varying magnitude. Alternating trials (of two movements) are highlighted with grey bars. Lower panel: this panel illustrates simulated dopamine responses in terms of a mixture of precision and its rate of change (see Fig. 2). There phasic fluctuations reflect changes in precision or confidence based upon the mismatch between the free energy before and after observing outcomes (see Fig. 2) (colour figure online)