Table 1.

Quality of model fits to participants' prediction sequences using a separate set of parameters for each participant

Model	RW	PH	PH—SD-Specific α1	Linear Adaptive PH: ν = 1	Log Adaptive PH: ν = 1	Linear Adaptive PH: ν = [0 1]
PH	dAIC: −439.67
	${\chi }_{31}^2$ = 501.67, P < 0.001
PH—SD-specific α1	dAIC: −505.78	dAIC: −66.10
	${\chi }_{93}^2$ = 691.78, P < 0.001	${\chi }_{62}^2$ = 190.00, P < 0.001
Linear adaptive PH	dAIC = −544.18	dAIC = −104.51	dAIC = −38.41
Fixed parameter adaptation: ν = 1	${\chi }_{124}^2$ = 965.18, P < 0.001	${\chi }_{93}^2$ = 463.50, P < 0.001	${\chi }_{31}^2$ = 273.40, P < 0.001
Log adaptive PH	dAIC: −621.05	dAIC: −181.38	dAIC: −115.278	dAIC: −76.87
Fixed parameter adaptation: ν = 1	${\chi }_{124}^2$ = 869.29, P < 0.001	${\chi }_{93}^2$ = 367.62, P < 0.001	${\chi }_{31}^2$ = 177.51, P < 0.001
Linear adaptive PH	dAIC: −635.86	dAIC: −196.18	dAIC: −130.08	dAIC: −91.67	dAIC: −14.80
Free adaptation parameter: ν = [0 1]	${\chi }_{155}^2$ = 945.86, P < 0.001	${\chi }_{124}^2$ = 444.18, P < 0.001	${\chi }_{62}^2$ = 254.08, P < 0.001	${\chi }_{31}^2$ = 149.79, P < 0.001	${\chi }_{31}^2$ = 76.57, P < 0.001
Log adaptive PH	dAIC: −671.02	dAIC: −231.34	dAIC: −165.24	dAIC: −35.16	dAIC: −35.16	dAIC: −35.16
Free adaptation parameter: ν = [0 1]	${\chi }_{155}^2$ = 981.02, P < 0.001	${\chi }_{124}^2$ = 479.34, P < 0.001	${\chi }_{62}^2$ = 289.24, P < 0.001	${\chi }_{31}^2$ = 184.94, p < 0.001	${\chi }_{31}^2$ = 111.73, P < 0.001

RW, Rescorla-Wagner; PH, Pearce-Hall; SD, standard deviation; α₁, initial learning rate; ν, adaptation to reward variability; dAIC, difference in Akaike information criterion value.