Fig. 1: Overview of value-based decision-making tasks (acquired data).
From: Action repetition biases choice in context-dependent decision-making
Top row: Probabilistic Tasks. A Reward probabilities for all four options (A–D), encountered in both the low contrast (LC) and high contrast (HC) choice contexts in tasks p1.1 to p3, and in the low gain (LG) and high gain (HG) choice contexts in tasks p4.1 and p4.2. B Learning phase for probabilistic tasks. Participants are presented with pairs of options (abstract fractal stimuli) within both LC/LG and HC/HG contexts and must select one option per trial. Across multiple trials, they learn to choose the better option through trial-and-error. In tasks p1.1–p1.3 and p3–p4.2, full feedback was provided for both the chosen and unchosen options. In task p2, only partial feedback was provided for the chosen option. Trials are presented in a pseudorandomized order, ensuring that the actual reward probabilities are maintained across every 10–12 trials (for exact sequences, see Supplementary Table 4). For analysis purposes, tasks p1.1 to p1.3 were pooled into a single dataset (task p1) due to their shared features and the absence of significant differences in results when analyzed individually. C In the transfer phase, options from both LC/LG and HC/HG choice contexts are tested against each other to evaluate whether participants show systematic biases, such as a preference for the CHC(0.70) option over the ALC(0.70) option. For all probabilistic tasks introduced in this study, we tested two cross-context comparisons during transfer (ALC vs CHC and BLC vs DHC) D Reward magnitudes for Gaussian tasks. In these tasks, trial-wise reward magnitudes for each option are sampled from a Gaussian distribution with a standard deviation of six. Each choice context consists of three options. In the LC context, reward distributions for different options do overlap, making it more challenging to identify the best option. E Learning phase for Gaussian tasks, equivalent to B. Participants make repeated choices between options within the LC and HC choice contexts. To help distinguish between contexts, visual cues are provided: distinct background colors (blue or green) and different tree types (evergreen or broadleaf trees) displayed in the four corners of the screen. F In the transfer phase, equivalent to C, participants’ preferences across contexts, options from both the LC and HC choice contexts are tested against each other (4 trials for each of 9 cross-context comparisons). This cross-context presentation is indicated by combined visual cues, including a mix of background colors and tree types. In all tasks, participants’ cumulative monetary reward was calculated by summing their trial outcomes: €0.10 for each correct choice in probabilistic tasks, or an amount proportional to the points earned in Gaussian tasks (0 to 100 points multiplied by €0.01). Participants received their cumulative reward as a bonus payment upon completing each task.
