Incidental learning of trust from identity-contingent gaze cues: boundaries, extensions and applications.

Monitoring the trustworthiness of social interaction partners is a cornerstone of social cognition. However, the mechanics of learning about trust during online interactions as a result of a person’s behaviour can be difficult to explore. The current experiments use a gaze cueing paradigm where faces provide either valid (always shift their gaze towards the location of a subsequent target), or invalid cues (always shift their gaze to a different location). Following gaze cueing, participants rate valid faces as more trustworthy than invalid faces. We show that this incidental trust learning is sensitive to the emotional expression of the face, is specific to assessments of trust, occurs outside of conscious awareness, and is driven primarily by a decrease in trust for invalid faces (Chapter 2), perhaps reflecting a cheater detection module. Memory for incidentally learned trust is surprisingly durable, is affected by the familiarity of the cueing faces (Chapter 3), and does not affect memory for the faces’ physical features, nor does the trustworthiness of the face generalise to other stimuli (Chapter 4). Furthermore, learning is modulated by top-down knowledge of social group membership − when group identity is made experimentally salient, participants default to a group-level representation as a heuristic for social judgements (Chapter 5), while using naturally occurring group memberships (i.e. race) results in better learning for in-group members than out-group (Chapter 6). Finally, while there is evidence that trust learning is driven by learning about eye-gaze behaviour, this cannot be explained purely by disruptions to visuomotor fluency (Chapter 7), which suggests that this phenomenon is part of an active social monitoring framework that relies on physical changes or behaviours in a face to affect subsequent social judgements.