Almost human: Anthropomorphism increases trust resilience in cognitive agents.

Abstract

We interact daily with computers that appear and behave like humans. Some researchers propose that people apply the same social norms to computers as they do to humans, suggesting that social psychological knowledge can be applied to our interactions with computers. In contrast, theories of human–automation interaction postulate that humans respond to machines in unique and specific ways. We believe that anthropomorphism—the degree to which an agent exhibits human characteristics—is the critical variable that may resolve this apparent contradiction across the formation, violation, and repair stages of trust. Three experiments were designed to examine these opposing viewpoints by varying the appearance and behavior of automated agents. Participants received advice that deteriorated gradually in reliability from a computer, avatar, or human agent. Our results showed (a) that anthropomorphic agents were associated with greater , a higher resistance to breakdowns in trust; (b) that these effects were magnified by greater uncertainty; and c) that incorporating human-like trust repair behavior largely erased differences between the agents. Automation anthropomorphism is therefore a critical variable that should be carefully incorporated into any general theory of human–agent trust as well as novel automation design.

A Social Cognition Perspective on Human–Computer Trust: The Effect of Perceived Warmth and Competence on Trust in Decision-Making With Computers

“…Our results also speak to the ongoing debate of humanhuman vs. human-computer trust (Madhavan and Wiegmann, 2007;de Visser et al, 2016). While some argue that both forms share the same underlying mechanisms (Reeves and Nass, 1996), others maintain that trust in computers is different from trust in people (Lee and See, 2004).…”

“…We extend work on the development of trust in computer agents by emphasizing the relation between behavioral or performance factors, respectively, and warmth and competence. Previous research focused on trusting and cooperative decisions based on artificial emotion expressions (Antos et al, 2011;de Melo et al, 2014), non-verbal behavior (DeSteno et al, 2012), human-likeness (Kiesler et al, 1996;Parise et al, 1999;de Visser et al, 2016), reciprocity (Sandoval et al, 2016), and agency . Our framework demonstrates that the behavioral preconditions of trust in computer agents such as selfishness and performance are translated by humans into warmth and competence attributions which, in turn, determine trust.…”

“…In contrast, inappropriate trust facilitates the disuse and misuse of computers (Lee and See, 2004), whereas trust violations due to errors decrease trust (Muir and Moray, 1996) and make it hard to reestablish trust (Hoffman et al, 2013). However, finetuned trust regulation mechanisms are useful in overcoming such trust violations (de Visser et al, 2016), allowing for feasible adjustments like trust calibration in contrast to costly solutions like technology disregard. Another novel challenge for trust evaluations comes from computers often being endowed with human-like capabilities that have been shown to bring forth natural human responses (Sidner et al, 2005;Walter et al, 2014;Krämer et al, 2015).…”

A Little Anthropomorphism Goes a Long Way

“…Recent research has expanded on this paradigm, finding that more salient human characteristics in automation (e.g., appearance, background story, observable behavior, etiquette) are associated with a tendency to treat that automation like another human (Hayes & Miller, 2011; Parasuraman & Miller, 2004). For example, human-like automation seems to resist breakdowns in trust, having greater trust resilience than computer-like automation (de Visser et al, 2012, 2016; Madhavan, Wiegmann, & Lacson, 2006). A theoretical explanation for this result is that people activate different schemas for humans compared to automation (Madhavan & Wiegmann, 2007).…”

“…Kosfeld et al’s comparison between social and non-social agents—between a human and a computer—is a binary representation of context that has been adopted by the majority of oxytocin researchers to date (Bartz, Zaki, Bolger, & Ochsner, 2011; Bethlehem, Baron-Cohen, van Honk, Auyeung, & Bos, 2014; Carter, 2014; Kanat, Heinrichs, & Domes, 2014). Rather than depict context dichotomously, a more nuanced approach is adopted in the current study consistent with recent research on automation anthropomorphism that manipulates social context with greater granularity (de Visser et al, 2016; Pak et al, 2012, 2014; Wiese et al, 2012). Under this design, oxytocin can be used as a drug probe to help determine the number and type of anthropomorphic features that are required to elicit the known biological effect of the peptide.…”