Fig. 5
From: Cortical recruitment determines learning dynamics and strategy
The effects of neuronal recruitment in the model are explained by the differential adjustment of synaptic weights during learning. a (top) Sketch of the initial synaptic weights and simulated model performance for S+ (red), S− (light blue), and both stimuli (black). (middle) Values of the connections to the excitatory (G) and inhibitory decision populations as indicated by the schematics on the left-hand-side. Yellow: connections from the “common” Ĉ population. Red: connections from the Ŝ+ population. Blue: connections from the Ŝ− population. (bottom) Sketch of the connectivity pattern after and response probability after learning for the S+ (co-activation of Ŝ+& Ĉ) and S− (co-activation of Ŝ− and Ĉ) reinforced stimuli as well as for the common stimulus component alone (Ĉ, yellow). Simulation parameters: X = [1; 1; 2], α = 0.01, σ = 0.6195, v = 6, wCi = 1, wCe = 2, and wS = 0.01. b. Same as a, but with wCi = 0.1, wCe = 0.2. c Same as a, but with X = [1; 2; 1]. d Same as c, but with wCi = 0.1, wCe = 0.2
