Fig. 5: Post-task valuation ratings and relationship between uncertainty and choice frequency (data was not collected for task p1).
From: Action repetition biases choice in context-dependent decision-making
A–F Violin plots for post-task valuation ratings (subjective beliefs) with true reward probabilities or reward magnitudes of each option (red dots), mean of participants’ beliefs about these task parameters (white diamonds), and individual participants’ ratings of each option (black dots). A Valuation ratings for task p2 (n = 49). B Valuation ratings for task p3 (n = 49), C Valuation ratings for task p4.1 (n = 29). D Valuation ratings for task p4.2 (n = 30). E Valuation ratings for task g1 (n = 50). F Valuation ratings for task g2 (n = 49). G Relationship between participants’ (group level) uncertainty about an option and normalized choice frequency in the learning phase, for all probabilistic tasks except p1. The standard deviation (SD) of participants’ individual valuation ratings as a measure of belief uncertainty is plotted on the x axis. Normalized choice frequency of each option during the learning phase is plotted on the y axis. We found substantial evidence for negative correlation (ρ = –0.56, 95% CI: [−0.83, −0.16]; BF10: 9). H Equivalent to (G) but for Gaussian tasks (g1 and g2) with anecdotal evidence for a negative correlation (ρ = −0.44, 95% CI: [−0.80, 0.09]; BF10: 1.7).
