Figure 7.
Choice probability is largely unaffected by choice history. a, b, We computed the residuals of the spike counts by subtracting from each measured spike count that predicted by a regression that included the current choice (Cn; a) or that included choice history (Hn-1; b) as covariates. In a second step, we then computed the “semipartial CPs” from these residual spike counts and the monkeys' choices on the current trial. a, To verify the validity of this approach we first removed the component predicted by choice. As expected, this removes CPs deviating from 0.5. b, Semipartial CPs for the residual spike counts after the contribution predicted by choice history was removed and the animals' choices, are plotted against the raw CPs (circles: Monkey 1, n = 41; squares: Monkey 2, n = 34). The values changed little and are highly correlated, supporting the view that choice probability are largely unaffected by choice history.