Social Inhibition of Return (sIOR) Under Partial Occlusion: A Preregistered Replication

This project preregisters a close replication of the social inhibition of return (sIOR) effect reported by Skarratt et al. (2010), in which participants’ responses are slower when their target appears at the same location as a partner’s immediately preceding target, compared to a different location. In addition to replicating the core sIOR effect, the project tests whether the amount of partner visibility, and therefore the amount of available social information during joint action, modulates the magnitude of sIOR. To test this, participants will complete a joint target-localisation task with a real partner while the partner is visible at five graded levels (0%, 25%, 50%, 75%, 100% visibility), administered in randomized blocked order from no visibility to full visibility. The primary expected outcome is an sIOR effect (slower RTs for Same vs Different trials), and a systematic change in the size of this effect as visibility increases, consistent with the idea that richer social information from the partner strengthens (or is required for) the inhibitory tagging observed in sIOR. Since there exists prior data on the sIOR effect, Bayesian analysis will be utilized as it supports evidence updating across studies. The results of the original study will be used as informative priors that will inform us of the expected effects in the replication. This approach provides a principles framework for accumulating evidence rather than relying on a single dichotomous decision based on a p-value derived from an independent experiment. Finally, this experiment is intended as the first of the initial phase of a longer-term project investigating whether sIOR generalizes from human-human joint action to human-robot interaction, where the availability and interpretation of social cues (e.g., agency, biological motion, gaze/hand information) can differ markedly from a human partner. Establishing (i) whether sIOR replicates in the current paradigm and (ii) how strongly it depends on partner visibility provides foundational evidence for designing subsequent human-robot studies that systematically manipulate the amount and type of social information conveyed by an artificial partner.