# Selfhood-attribution in a social context: further evidence for a Pars-Pro-Toto account

## TL;DR

People tend to attribute selfhood to robots based on a single social cue, showing how easily we generalize complex traits in social contexts.

## Contribution

The study extends the Pars-Pro-Toto account to social selfhood cues in non-humanoid robots.

## Key findings

- A single social cue is enough to trigger overgeneralization of selfhood traits in robots.
- Participants overgeneralized more to social selfhood characteristics than non-social ones.
- Agency attributions were less consistent compared to previous non-social studies.

## Abstract

Humans show a consistent tendency to anthropomorphize or attribute aspects of selfhood to non-human agents. In a previous study, we found that people (over-)generalize from the presence of a single behavioral selfhood cue (like equifinality or efficiency) to the presence of other (actually absent) cues, suggesting that a small aspect of selfhood suffices to activate the entire selfhood concept with all its other implications (Pars-Pro-Toto).

Our previous study was exclusively manipulating non-social aspects of selfhood. However, the contribution of social interaction in developing a concept of “Self” has been stressed in the literature. Thus, in the present study we tested whether these findings can also be demonstrated for social aspects. Specifically, we manipulated the presence or absence of cues indicating social sensitivity, attention sharing, or helping behavior in small non-humanoid robots, and tested which cues would elicit attributions of various aspects of selfhood.

The results replicated our previous finding that the presence of a single cue is sufficient to (over-)generalize to other, non-manipulated cues, extended our previous observations to social conditions and provided further support for our Pars-Pro-Toto account. It is noteworthy that participants showed a stronger tendency to overgeneralize to other social selfhood-related characteristics than to the non-social characteristics. Moreover, compared to our previous study, participants no longer showed a consistently stronger attribution of agency to the robot that was exhibiting cues for one of the critical characteristics.

The missing effects of agency are discussed as reflecting how sociality might be construed vis-à-vis individual goal pursuit. The interplay between sociality and individuality might be linked to our perception of agency in other agents when these are part of a group. The results are further discussed in light of the increasing presence of robots and other artificial agents in everyday life, as they support a shift in focus from their actual capabilities toward what people's expectations of these systems are. We stress that it is important to consider technological systems in their social relation to people as they tend to attribute complex concepts such as selfhood even when only perceiving simple behavioral cues.