How commitment affects trust in communication: coordination, confidence and evidence

Given the risks of defection and misinformation, humans have evolved mechanisms of strategic vigilance to evaluate speakers’ disposition to be good partners (Heintz et al., 2016) and epistemic vigilance to evaluate speakers’ reliability as sources of information (Sperber et al., 2010), and calibrate their trust accordingly. This thesis focuses on trust calibration as a result of tracking speakers’ commitments. It investigates different types of commitment that speakers are required to fulfill in communication and it experimentally tests the impact of violating such commitments on speakers’ perceived trustworthiness. The thesis thus brings about an interdisciplinary perspective on the study of how commitment affects trust in communication by integrating methods borrowed from experimental psychology and pragmatics, enriched with philosophical and linguistic analyses of speaker commitment and trust. First, individuals have practical commitments to act in a certain way. One practical commitment that speakers are required to meet in conversation is to respect the “conceptual pacts” they have established with their interlocutors, i.e., their implicit agreements to reuse lexical expressions on which they converged to refer to their environment (Brennan & Clark, 1996; Metzing & Brennan, 2003). Based on previous literature suggesting that coordination elicits a sense of commitment to act in the expected way (Michael, 2022; Michael et al., 2016a), the first study of this thesis, presented in Article 1 Does lexical coordination affect epistemic and practical trust? The role of conceptual pacts, experimentally investigates the effect of lexical coordination on trust. Two online experiments test the effect of establishing, maintaining and breaking conceptual pacts on speakers’ perceived trustworthiness. The results indicate that people are more likely to trust a partner who maintains conceptual pacts than a partner who breaks previously established conceptual pacts when confronted with both practical decisions (who to trust as a partner for future interactions) and epistemic ones (who to trust as a source of information). Second, speakers have epistemic commitments that they are required to respect in communication. Notably, speakers are expected to express confidence when the information they convey is accurate (as opposed to inaccurate) and supported by strong evidence (Tenney et al., 2007, 2008, 2019; Vullioud et al., 2017). The second study of this thesis, presented in Article 2 Speaker trustworthiness: Shall confidence match evidence?, investigates the effects of overconfidence and bad confidence-evidence calibration on speakers’ perceived trustworthiness. Two online experiments show that overconfidence (expressing confidence when the information is inaccurate) is not (or less) detrimental to speakers’ trustworthiness if speakers are justified by strong evidence, whereas expressing confidence in the presence of weak evidence is detrimental to speakers’ trustworthiness, even when the information shared turns out to be accurate. This suggests that trust calibration depends more on the evaluation of speakers’ confidence-evidence calibration than on the assessment of the accuracy of the information shared. Speakers can also modulate their epistemic commitments by directly communicating their evidential basis via evidential claims. For instance, they can take responsibility for the truth of the information they convey by using claims to first-hand evidence such as “I saw it”, or they can defer responsibility by providing claims to second-hand evidence such as “somebody told me”. Previous empirical research indicates that speakers providing claims to first-hand evidence are rated as more believable compared to speakers providing claims to second-hand evidence, but are also more likely than the latter to be punished if their message turns out to be inaccurate (Mahr & Csibra, 2021). As humans may provide (voluntarily or not) inaccurate evidential claims, the third study of this thesis, presented in Article 3 Believe me, I saw it! Speaker’s responsibility for evidential claims, investigates whether addressees track the accuracy of speakers’ evidential claims and use it as a cue to calibrate their trust, over and beyond the accuracy of the information communicated. An online experiment reveals that a speaker sharing accurate information with inaccurate evidential claims suffers a higher reputational loss when using claims to first-hand evidence compared to second-hand evidence. Finally, the fourth study of this thesis, presented in Article 4 What’s your evidence? The psychological foundations of the evaluation of testimony, investigates how the empirical findings of this thesis can shed light on the psychological foundations of legal principles regulating the evaluation of testimony in court, including the so-called “law against hearsay” found in common law systems, which bans hearsay from testimony (Tuzet, 2021a). Overall, this thesis shows that communication is governed by numerous (implicit) expectations, which speakers are committed to fulfilling to maintain their reputation as trustworthy partners for social interactions and sources of information. The experimental studies of this thesis shed light on how commitments are modulated by speakers to manage their trustworthiness as well as tracked by addressees to calibrate their trust and penalize speakers who violate them. Ultimately, the thesis highlights the crucial role of commitment in managing trust in communication.