Simulation of experimental data on rodent latent inhibition.

<p>(<b>A</b>) Fear expression in rats in a latent inhibition paradigm in which animals were either pre-exposed (black bars) or not pre-exposed (white bars) to the conditioned stimulus, and treated with either saline, a GABA-A antagonist during conditioning, or a GABA-A antagonist during testing (reproduced by hand from [<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1006315#pcbi.1006315.ref043" target="_blank">43</a>]). (<b>B</b>) Data from the model simulation of the same latent inhibition paradigm. Bars show median activity across 30 model runs (errorbars are 90% CI generated by bootstrapping 5-sample medians), of the average Amygdala layer activity during the final (test) stimulus presentations. (<b>C</b>) Cortex excitatory unit activity in response to stimuli across trials in an example run of the model. The downward curve during the first 30 presentations shows that the network learned to ignore the CS in all simulations with CS pre-exposures. The activity during conditioning and test periods shows the combined impact of relevance learning and impaired inhibition. (<b>D</b>) Amygdala activity levels in an example run of the model over trials with Conditioning and Testing epochs (as in the right-side panel of part C). Test period activity shows a pre-exposure effect in the control condition (solid versus dashed gray lines). This is amplified when inhibition is disrupted during conditioning (solid versus dashed red lines) but was lost when inhibition was disrupted during test (solid versus dashed blue lines).