G. Huang, Y. Xu, E. Derrington, J.‐C. Dreher, and C. Qu, “Intention–Outcome Trade‐Off in Moral Character Learning,” Annals of the New York Academy of Sciences 1557, no. 1 (2026): e70233, https://doi.org/10.1111/nyas.70233.
In the originally published article, the overall probabilities of receiving an auditory shock (p(shock)) were labeled incorrectly in Figure 1B and the figure caption. The correct figure and caption are shown below.
FIGURE 1.
Experimental procedure and setting in moral character learning task. (A) All participants first played the game in the role of the “agent” to ensure that they believed the experimental setting and understood the game. Then participants completed the receiver session. In each trail, a participant was presented with an image of one of four agents and was then asked to predict whether the agent would choose either a dark‐ or light‐colored roulette wheel. The former was associated with a high risk of the participant receiving an auditory shock, the latter with a low risk. They were presented the agent's choice as feedback after 2 s. Then participants were then asked to predict the outcome of the roulette wheel lottery on whether they would receive the shock at the end of the experiment or not. After 2 s they received the outcome of this lottery (shock or not) which was displayed for 2 s. Finally, participants made a moral evaluation of the agent on a 9‐point scale (1–9) using keyboard keys. At no point in the experiment did the participant actually receive the auditory shocks. At the end of each block, participants completed a testing phase in which they were asked to choose which of two deciders they preferred. A highly similar experimental procedure was used in Experiments 2 and 3 with any differences (e.g., whether the participant was required to predict the outcome, or not [Experiment 2] and reversal of the order of presentation of the outcome and intention [Experiment 3]). Details are specified and explained in the appropriate experimental sections. The agents’ faces were selected from the Chicago Face Database [40]. (B) Participants encountered four agents differing in their probability of selecting the dark‐colored roulette wheel (agents’ intention: 70% for negative vs. 30% for positive) and the shock probabilities associated with their roulette wheel sets (agents’ outcome: blue wheels = high shock risk [dark: 86%, light: 46%]; orange wheels = low risk [dark: 54%, light: 14%]). By varying the shock probabilities associated with the roulette wheel sets, the probability of shock outcome was manipulated independently of the probability of selecting the dark‐colored roulette wheel. As shown in Panel B, each agent's combined intention–outcome profile in Experiment 1 yielded distinct overall shock probabilities (74%, 58%, 42%, and 26%), which spanned a continuous range rather than forming distinct positive/negative categories. Experiments 2 and 3 systematically adjusted these probabilities to create two discrete valence levels (positive vs. negative) for agents’ overall shock outcome (see Tables 1 and 2 for specific configurations).
We apologize for this error.