Parameter values for data fits for monkeys A and T, averaged over all sessions, to the PMF (15).

<p>Upper two rows show separate fits of and for the four reward conditions. Middle two rows show fits for pooled LL and HH data, with resulting common value held fixed across unequal reward conditions. Lower two rows show results with constrained to zero for equal rewards; in columns 1 and 2 LL and HH are fitted separately, in columns 3 and 4 LL and HH data is pooled to produce , and this value is fixed across unequal reward conditions. Units of and respectively are increase in probability of a T1 choice per change in % coherence, and % coherence (see Models of stimuli and reward biasing). Values are given to 3 significant figures with residual fit errors (in mean square norm) in parentheses. In rows 5 and 6 of the HL and LH columns the first error figure refers to the LL and HH pooled data fit with .