Figure 1. Alternation of learning and forgetting in multi-task performance.

Percentage of networks learning two (left) or three (right) rules as a function of the number of times the rules are applied for 500 configurations with N = 1000 neurons (k_d = 3, α = 0.001, p_in = 0.3). Left: Performance in parallel learning the OR and AND rules. Right: Performance in learning three rules in parallel (OR, AND and RAN). Inset: Performance for the RAN rule at a finer temporal scale.