Table 1.

Fit quality per subject

	Trials	Coefficient of determination, R²			AIC	AICconst	KS
	Trials	Chron	Psych	Avg	AIC	AICconst	KS
Palmer et al. (2005)
AH	554	0.948	0.974	0.961	−167.85 (±4.16)	33.90	1 (p < 0.01)
EH	560	0.952	0.944	0.948	245.11 (±3.71)	347.41	1 (p < 0.05)
JD	567	0.897	0.974	0.936	−558.39 (±57.68)	516.36	3 (p < 0.01)
JP	555	0.976	0.998	0.987	−330.01 (±5.28)	2754.84	2 (p < 0.05)
MK	573	0.973	0.993	0.983	−504.24 (±4.60)	2041.61	2 (p < 0.01)
MM	564	0.928	0.978	0.953	−262.25 (±3.65)	107.83	0
Avg	562.2 (±3.3)	0.946 (±0.013)	0.977 (±0.008)	0.961 (±0.009)
Roitman and Shadlen (2002)
B	2615	0.985	0.989	0.987	−809.16 (±18.99)	26,436.32	0
N	3534	0.983	0.991	0.987	620.34 (±33.69)	48,005.64	3 (p < 0.05)

The table shows, for each subject, the number of trials that were fitted; the coefficient of determination (R²) for the chronometric function (Chron), the psychometric function (Psych), and averaged over both (Avg); the goodness of fit of our model according to the AIC (smaller is better), and the comparison goodness of fit of a diffusion model with constant bound (AICconst); the number of conditions (out of 12) for which the Kolmogorov–Smirnov test revealed a statistical significance between the reaction time distribution featured by the subject and that predicted by the model, together with the level of significance (KS). For the human dataset, we also provide the mean across subjects (±1 SEM) for the number of trials and the coefficient of determination. The AIC measure for our model is computed separately for 500 posterior samples, and here we provide mean ± 1 SD.