Table 3.

Recovered parameters for the Q-learning network model including both state-cues (when appropriate) and the magnitude of the reward signal on the last trial as input. The first number in each cell is the mean value of the parameter across all participants assigned to the respective condition. The second number reports the median. Finally, standard deviations are shown in parentheses.

Condition	α	τ	γ
Experiment 1
cont.	.3,.2 (.28)	.02,.01 (.02)	.73,.88 (.38)
prob.	.05,.04 (.4)	.09,.08 (.03)	.1, 0.0 (.32)
Experiment 2
no-cue	.41,.36 (.31)	.03,.001 (.13)	.48,.44 (.42)
shuffled-cue	.16,.13 (.12)	.002,.001 (.002)	.77,.81 (.26)
consistent-cue	.12,.09 (.12)	.002,.001 (.002)	.69,.86 (.34)