Social Transmission of Experience of Agency: An Experimental Study

Abstract

The sense of controlling one’s own actions is fundamental to normal human mental function, and also underlies concepts of social responsibility for action. However, it remains unclear how the wider social context of human action influences sense of agency. Using a simple experimental design, we investigated, for the first time, how observing the action of another person or a robot could potentially influence one’s own sense of agency. We assessed how observing another’s action might change the perceived temporal relationship between one’s own voluntary actions and their outcomes, which has been proposed as an implicit measure of sense of agency. Working in pairs, participants chose between two action alternatives, one rewarded more frequently than the other, while watching a rotating clock hand. They judged, in separate blocks, either the time of their own action, or the time of a tone that followed the action. These were compared to baseline judgements of actions alone, or tones alone, to calculate the perceptual shift of action toward outcome and vice versa. Our design focused on how these two dependent variables, which jointly provide an implicit measure of sense of agency, might be influenced by observing another’s action. In the observational group, each participant could see the other’s actions. Multivariate analysis showed that the perceived time of action and tone shifted progressively toward the actual time of outcome with repeated experience of this social situation. No such progressive change occurred in other groups for whom a barrier hid participants’ actions from each other. However, a similar effect was observed in the group that viewed movements of a human-like robotic hand, rather than actions of another person. This finding suggests that observing the actions of others increases the salience of the external outcomes of action and this effect is not unique to observing human agents. Social contexts in which we see others controlling external events may play an important role in mentally representing the impact of our own actions on the external world.

When robots tell you what to do: Sense of agency in human- and robot-guided actions

“…Notably, indistinguishable SoA between human and robotinstructed actions was found in both experiments regardless of the value or relevance of the outcomes produced by the instructed actions. These findings are partly in line with a previous finding that no difference in intentional binding was exhibited in one's actions when observing human and robot produced actions (Khalighinejad, Bahrami, et al, 2016). Additionally, the current findings support the previous studies that reported similar Social Simon effect when cooperating with another person and a humanoid robot (Stenzel et al, 2012(Stenzel et al, , 2014, although their finding hinged on perceived intentionality (Stenzel et al, 2012) and more strongly on perceived agency of the robot (Stenzel et al, 2014).…”

“…However, other studies demonstrated that actions of a humanoid robot could be co-represented depending on the perceived intentionality (Stenzel et al, 2012) and more strongly on the perceived agency (Stenzel et al, 2014). Additionally, observing actions of both a human and an anthropomorphic robotic hand could yield similar effects on intentional binding for one's own actions (Khalighinejad, Bahrami, et al, 2016). These results, overall, suggest that human-like features of artificial agents as well as higher level beliefs about their agentic capacity could play a critical role in how people experience agency in joint actions with these agents.…”

“…Taken together, previous research examining SoA in interactions with different artificial systems suggests that performing actions in collaboration with these agents might impair one's SoA as compared to individual and human-human joint actions (Berberian et al, 2012;Obhi & Hall, 2011b;Sahaï et al, 2019), while human-like features of the artificial partner could also play a critical role (Khalighinejad, Bahrami, et al, 2016). It should be noted that majority of previous studies involved joint action scenarios in which participants and artificial agents either jointly or alternately performed the experimental tasks.…”

“…That is, an implicit level of SoA indexed by intentional binding could only be exhibited when the interaction partner was another human in contrast to an artificial agent such as a computer. Conversely, another study showed that repeated experience with observing another person and a human-like robotic hand performing the same action yielded similar progression of temporal binding between one's actions and the corresponding outcomes (Khalighinejad, Bahrami, Caspar, & Haggard, 2016).…”

Action co-representation and the sense of agency during a joint Simon task: Comparing human and machine co-agents

“…Conversely, it has been shown that observing an action performed by a human-like automaton (e.g. an anthropomorphic hand with servo-actuated tendons rotating each finger joint) induced a temporal binding effect in a similar way as during the observation of another person performing the same action (Khalighinejad, Bahrami, Caspar & Haggard, 2016).…”