Fig. 6

Computational model of hippocampus-prefrontal cortex for memory acquisition, consolidation and recall. (A) Scheme of the computational model. The model consisted of 2 layers with connections within and between them. New patterns were learned through an acquisition phase and a subsequent consolidation phase, in which the strength of the connections was modified. Recall was tested by presenting the 50% of a learned pattern in the PFC layer and comparing the retrieved pattern with the one in memory. (B) Left panel, Score recall for the first pattern learned by the model as new patterns were learned before and after manipulating the connections from HC to PFC by randomly eliminating connections (from 0–100% in steps of 20%). Horizontal dashed gray line indicates the chance level for recall. Middle panel, Score recall for the last pattern learned before manipulation of the HC-PFC connections (pattern 10). Same conventions as in Left panel. Right panel, Score recall for the first pattern learned after manipulation of the HC-PFC connections (pattern 11). Same conventions as in Left and Middle panels. Inset panel, Proportion of new patterns (patterns from 11 to 20) that were consolidated after HC-PFC impairment and for different values of it. Error bars indicate SEM. (C) Score recall for all patterns learned by the model, after all of them had been learned. Same conventions as in (B) (D) Proportion of recall of new patterns vs. old patterns for different values of “forgetting” learning rate and HC-PFC impairment when the model learned patterns that shared a 50% of overlap. Shaded areas indicate SEM.