4.1. Special case of the Q-learning model (F-Q model)
We began the simulations by confirming the analytical relation
between the Q-learning model with the special case αF = αL (F-Q
model) and the logistic regression model.
We set the inverse tem-perature parameter β = 3.0. Fig. 1(A) shows the regression coef-ficients of the logistic regression that were analytically predicted
from Eq. (17) (squares) and those obtained by statistical model fit-ting to the simulated data (solid lines). We can confirm that these
two results almost perfectly agree, which supports the validity of
the analytical calculation.