Computationally modeling interpersonal trust

Abstract

We present a computational model capable of predicting—above human accuracy—the degree of trust a person has toward their novel partner by observing the trust-related nonverbal cues expressed in their social interaction. We summarize our prior work, in which we identify nonverbal cues that signal untrustworthy behavior and also demonstrate the human mind's readiness to interpret those cues to assess the trustworthiness of a social robot. We demonstrate that domain knowledge gained from our prior work using human-subjects experiments, when incorporated into the feature engineering process, permits a computational model to outperform both human predictions and a baseline model built in naiveté of this domain knowledge. We then present the construction of hidden Markov models to investigate temporal relationships among the trust-related nonverbal cues. By interpreting the resulting learned structure, we observe that models built to emulate different levels of trust exhibit different sequences of nonverbal cues. From this observation, we derived sequence-based temporal features that further improve the accuracy of our computational model. Our multi-step research process presented in this paper combines the strength of experimental manipulation and machine learning to not only design a computational trust model but also to further our understanding of the dynamics of interpersonal trust.

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

“…As part of this, they often mimic human appearance and behavior to create trustworthiness in accordance with how humans develop trust (DeSteno et al, 2012;Lee et al, 2013). The present experiment suggests that people's willingness to trust computer systems depends on fundamental attributions of warmth and competence.…”

“…It was explained that the computer was in the same position and faces the same decision. The game provides an incremental measure of behavioral trust, operationalized as the number of tokens being exchanged, and instead of measuring purely economic decision-making, choices in the give-some dilemma reflect social perceptions of the counterpart and are positively correlated with subjective trust assessments (Lee et al, 2013). Participants were told that although both players decide simultaneously, the computer's decision would only be revealed at the end of the experiment to avoid confounding the following measures.…”

Improving Team's Consistency of Understanding in Meetings

“…Smart interfaces have been developed to suggest review topics in online education [24] and to generate feedback in personal tutoring systems [25]. Robots have been integrated into meetingsfor example, to serve as moderators in balancing engagement and dominance levels [26], and to influence conflict dynamics during team problem-solving tasks [27], and to predict levels of interpersonal trust [28]. Our work addresses the novel task of predicting the consistency of understanding during team meetings.…”

“…We modeled our problem using hidden Markov models (HMMs) [41] because of their applicability to modeling systems with temporal sequences, as well as for their prior success within the human communication and social interaction domains [42], [28]. An HMM is defined as a 5-tuple {S, O, A, B, π}, where:…”

Automatic prediction of consistency among team members' understanding of group decisions in meetings

“…Given a sequence of features per discussion point, HMMs are used to predict either strong or weak group consistency (a form of binary classification). We incorporated HMMs because of their applicability to modeling systems with temporal sequences, as well as for their prior success in modeling human communication and social interactions [22], [23], [24]. An HMM can be described as a tuple {S, O, A, B, π}, where S is the set of hidden states, O is the set of observations, A is the state transition matrix, B is the observation probability matrix and π is the initial state distribution.…”