Taking the perspectives of many people: Humanization matters

Abstract

In a busy space, people encounter many other people with different viewpoints, but classic studies of perspective-taking examine only one agent at a time. This paper explores the issue of selectivity in visual perspective-taking (VPT) when different people are available to interact with. We consider the hypothesis that humanization impacts on VPT in four studies using virtual reality methods. Experiments 1 and 2 use the director task to show that for more humanized agents (an in-group member or a virtual human agent), participants were more likely to use VPT to achieve lower error rate. Experiments 3 and 4 used a two-agent social mental rotation task to show that participants are faster and more accurate to recognize items which are oriented towards a more humanized agent (an in-group member or a naturally moving agent). All results support the claim that humanization alters the propensity to engage in VPT in rich social contexts.

Human but not robotic gaze facilitates action prediction

“…It has been proposed that mentalizing abilities can be influenced by the ability to detect others’ gaze direction ( Stephenson et al., 2021 ), and that we may have an advantage in taking the perspective of agents we perceive similar to ourselves (e.g., ingroup vs outgroup; Ye et al., 2021 ).…”

Manipulating avatar age and gender in level-2 visual perspective taking

“…The absolute differences in response times between avatars were minimal and the effects of social cues, compared to spatial cues like angle, were small. Previous research that has found larger group-related effects embedded VSPT in social contexts that explicitly relate to the target agents (Todd et al, 2011;Ye et al, 2020). For example, communicative tasks (Savitsky et al, 2011) indicate a "mind to be known," evoking theory-of-mind, and tasks pre-testing to allocate groups (Ye et al, 2020) suggest avatar similarity and differences are important considerations.…”

“…Previous research that has found larger group-related effects embedded VSPT in social contexts that explicitly relate to the target agents (Todd et al, 2011;Ye et al, 2020). For example, communicative tasks (Savitsky et al, 2011) indicate a "mind to be known," evoking theory-of-mind, and tasks pre-testing to allocate groups (Ye et al, 2020) suggest avatar similarity and differences are important considerations. Furthermore, Tarampi et al (2016) found that framing spatial orientation tasks as measuring empathy instead of spatial skills significantly improved female performance.…”

Human-like interactions prompt people to take a robot’s perspective

“…Humanisation can also impact VPT through top-down processes, where prior information, context or task instructions make a participant believe the agent is similar to themselves. Recent findings revealed that compared with an outgroup member, we have a stronger propensity to take an ingroup member’s perspective ( Vaes et al, 2004 ; Simpson and Todd, 2017 ; Ye et al, 2021 ). The ingroup effect is not restricted to groups formed by physical features, but also applies to arbitrarily formed groups.…”

“…Visual perspective taking (VPT) refers to the ability to consider what another person can see and how they see it ( Flavell, 1977 ; Samson et al, 2010 ; Elekes et al, 2016 ). Previous studies on VPT suggest that people are more likely to take the perspective of an agent that they consider to be human, including agents that move like a human, look like a human, and belong to an in-group ( Vaes et al, 2004 ; Mazzarella et al, 2012 ; Zhao et al, 2016 ; Ye et al, 2021 ; Zhao and Malle, 2022 ). Based on these findings, our previous research proposed the ‘humanisation account’ of VPT, suggesting that the psychological process of humanisation might serve as a gateway towards spontaneous VPT.…”