Proceedings of the Annual Meeting of the Cognitive Science Society

Classic tasks of pragmatic reasoning involve inferring a possible referent given a speaker's utterance (Frank & Goodman, 2012). Given prior work demonstrating self-favoring biases in perception and causal attribution, we investigated whether listeners show biased pragmatic reasoning when the referent could be themselves. Participants in a fake multi-player reference game were assigned an avatar with two features that were shared or not shared with two other avatars. After completing a drawing task, they were asked to guess who performed the best (prebet), self-rated their performance, and then given a single feature of the player who did best before guessing the winner again (postbet). Preliminary results suggest participants were more likely to choose themselves as best in the prebet, while the postbet reflects a complex relationship between participants' own avatar, their self-rated performance, and the described feature. Planned work will investigate the possible effects of outcome valence and stimulus set.