Appendix 1—table 4. Comparison of prediction accuracy (± SE) across four models in four game contexts for the two monkeys.

	Strategy
Context	Dynamic	Static	LARL	Perceptron
Overall	0.904 ± 0.006	0.816 ± 0.010	0.669 ± 0.011	0.624 ± 0.010
Early game	0.886 ± 0.0016	0.804 ± 0.025	0.775 ± 0.021	0.582 ± 0.026
Late game	0.898 ± 0.011	0.805 ± 0.019	0.621 ± 0.018	0.599 ± 0.019
Scared ghosts	0.958 ± 0.010	0.728 ± 0.031	0.672 ± 0.025	0.455 ± 0.030

LARL: linear approximate reinforcement learning.