The current model is able to reproduce such results (see Figure 5
C, D). By running a simulation for each group of rats, using
different parameters (mainly varying the v parameter) the model
reproduces the different tendencies to engage with the lever
(v~0:499), with the magazine (v~0:048) or to fluctuate between
the two (v~0:276). A high v strengthens the influence of the
Feature-Model-Free system, which learns to associate a high
motivational value to the lever CS, and a sign-tracking CR
dominates. A low v increases the influence of the Model-Based
system, which infers the optimal behaviour to maximize reward,