Table 2. Results of the logistic regression analysis of stage 1 and stage 2 choices in the probe session.

For the stage 1 choices, the analysis focused on staying on the same stage 1 action on the next trial, based on whether the previous trial was rewarded and whether it was common or rare (trans). ‘reward:transition’ is the interaction between reward, and transition type. For stage 2 choices, the analysis focused on staying on the same stage 2 action, based on staying on the same stage 1 action (stay) and earning a reward on the previous trial. ‘reward:stay’ is the interaction between ‘reward’, and ‘stay’.

stage 1 actions
	intercept	reward	transition	reward:transition
p-value	0.373	0.003	0.164	< 10−5
β (SE)	-0.16 (0.18)	0.61 (0.20)	0.37 (0.26)	0.92 (0.20)
stage 2 actions
	intercept	reward	stay	reward:stay
p-value	0.001	<0.001	0.019	0.027
β (SE)	-0.95 (0.29)	0.84 (0.23)	0.60 (0.25)	0.49 (0.22)