To some extent, unethical behavior results from people’s limited attention to ethical considerations, which results in an ethical blind spot. Here, we focus on the role of ambiguity in shaping people’s ethical blind spots, which in turn lead to their ethical failures. We suggest that in ambiguous settings, individuals’ attention shifts toward tempting information, which determines the magnitude of their lies. Employing a novel ambiguous-dice paradigm, we asked participants to report the outcome of the die roll appearing closest to the location of a previously presented fixation cross on a computer screen; this outcome would determine their pay. We varied the value of the die second closest to the fixation cross to be either higher (i.e., tempting) or lower (i.e., not tempting) than the die closest to the fixation cross. Results of two experiments revealed that in ambiguous settings, people’s incorrect responses were self-serving. Tracking participants’ eye movements demonstrated that people’s ethical blind spots are shaped by increased attention toward tempting information.

In ambiguous settings, people are tempted to make self-serving mistakes. Here, we assess whether people make more self-serving mistakes to minimize losses compared with maximize gains. Results reveal that participants are twice as likely to make self-serving mistakes to reduce losses compared to increase gains. We further trace participants' eye movements to gain insight into the process underlying self-serving mistakes in losses and gains. We find that tempting, self-serving information does not capture more attention in loss, compared to gain framing. Rather, in loss framing, people are more likely to report the tempting, self-serving information they observed. The results imply that rather than diverting attention away from tempting information, reducing people's motivation to make self-serving mistakes, and framing goals as gains rather than losses are promising ways to decrease the occurrence of self-serving mistakes. In turn, this fosters environments with more accuracy and fewer motivated mistakes.