A win-win situation: Does familiarity with a social robot modulate feedback monitoring and learning?

Abstract

Social species rely on the ability to modulate feedback-monitoring in social contexts to adjust one’s actions and obtain desired outcomes. When being awarded positive outcomes during a gambling task, feedback-monitoring is attenuated when strangers are rewarded, as less value is assigned to the awarded outcome. This difference in feedback-monitoring can be indexed by an event-related potential (ERP) component known as the Reward Positivity (RewP), whose amplitude is enhanced when receiving positive feedback. While the degree of familiarity influences the RewP, little is known about how the RewP and reinforcement learning are affected when gambling on behalf of familiar versus nonfamiliar agents, such as robots. This question becomes increasingly important given that robots may be used as teachers and/or social companions in the near future, with whom children and adults will interact with for short or long periods of time. In the present study, we examined whether feedback-monitoring when gambling on behalf of oneself compared with a robot is impacted by whether participants have familiarized themselves with the robot before the task. We expected enhanced RewP amplitude for self versus other for those who did not familiarize with the robot and that self–other differences in the RewP would be attenuated for those who familiarized with the robot. Instead, we observed that the RewP was larger when familiarization with the robot occurred, which corresponded to overall worse learning outcomes. We additionally observed an enhanced P3 effect for the high-familiarity condition, which suggests an increased motivation to reward. These findings suggest that familiarization with robots may cause a positive motivational effect, which positively affects RewP amplitudes, but interferes with learning.

Examining Social Cognition with Embodied Robots: Does Prior Experience with a Robot Impact Feedback-associated Learning in a Gambling Task?

“…Support for this assumption comes from a recent neurophysiological study using the gambling task to examine whether familiarization with the robot Cozmo 1 modulates feedback processing after a one-time interaction: one group of participants engaged in a 20-minutes interaction with Cozmo (playing interactive games with the robot via the associated mobile app) and the other group played the Simon Says game (an interactive game that does not involve Cozmo) before gambling for themselves ("Self") or Cozmo ("Other"). The data shows larger RewP amplitudes and slower learning of the contingencies between color and chance of winning after one-time familiarization with Cozmo (Abubshait et al, 2021). Although this effect was not specific to the "Other" condition, but affected the "Self" condition as well, the results indicate that familiarization with a social robot does indeed modulate feedback-based learning in short-term interactions.…”

“…The paradigm presented here shows that a "hybrid" solution, that is: the combination of an unscripted interaction component and a scripted measurement component, is worth considering as it does not place significant constraints on the social interaction, but allows behavioral and neurophysiological correlates of social cognition to be obtained with high internal validity. A few early studies pioneering this approach have shown its feasibility for behavioral (current study), neurophysiological (e.g., Abubshait et al, 2021; "hybrid" with an ERP component) and neural (e.g., Cross et al, 2019; "hybrid" with fMRI component) outcome measures.…”

“…The current study also does not allow us to completely control for differences in reward valuation of winning a gift card for one's self vs. upgrades for the robot. However, since prior work has failed to find differences in RewP amplitudes for "Self" versus "Other" in the one-time interaction condition (Abubshait et al, 2021), this suggests that the reward values for these two conditions were comparable after a short interaction. This interpretation is also supported by other work that has shown that RewP amplitudes are sensitive to the magnitude of the reward in the gambling task (Hassall et al, 2016).…”

“…The goal of this paper is to present a proof-of-concept that hybrid paradigms that separate a dynamic, relatively unscripted interaction component (here: interaction with Cozmo via its associated app) from a highly controlled, scripted data collection component are suitable to investigate mechanisms of social cognition in human-robot interaction. Specifically, we explore how different degrees of familiarity with a robot due to prior interactions of different durations impact behavioral correlates of reward processing in form of feedback-associated learning in a gambling task (see Abubshait et al, 2021). For that purpose, we compare feedback-associated learning (i.e., speed with which participants learn to pick the square associated with the higher chance of winning) after a one-time interaction in the lab (about 20 minutes) to feedbackassociated learning after repeated interactions with Cozmo in participants' homes (over five consecutive days); both conditions are compared to a no interaction condition, in which participants do not familiarize themselves with Cozmo before performing the gambling task but instead play an interactive game that does not involve Cozmo (i.e., Simon says).…”

“…Please note that in addition to the behavioral data, we collected EEG data in all three conditions. The EEG data from the no (NI) and the one-time (OTI) interaction condition are reported in Abubshait et al (2021). The EEG specific procedures are described there.…”