Here, we adopt a widely accepted definition of trust put forth by Rousseau et al. (1998, p. 395): “a psychological state comprising the intention to accept vulnerability based upon positive expectations of the intentions or behaviour of another.” Trust is thus conceived as an attitude that will be weighed against perceived risk to determine risk-taking in a given situation. As such, it is conceptually distinct from (albeit an important predictor of) trusting behaviour, which we refer to as willingness to risk (see Kim et al., 2004).

Trust must also be distinguished from trustworthiness, which reflects the characteristics of the trustee. Trustworthiness includes at least three elements: the trustee’s ability (“that group of skills, competencies, and characteristics that enable a party to influence within some specific domain,” Mayer et al., 1995, p. 717), benevolence (“the extent to which the trustee is believed to want to do good to the trustor, aside from an egocentric profit motive,” ibid., p. 718), and integrity (“the perception that the trustee adheres to a set of principles that the trustor finds acceptable,” ibid., p. 719; see also Colquitt et al., 2007; Kelton et al., 2008; Schoorman et al., 2007). Trustworthiness is evaluated through third-party information and observation of the trustee’s actions, and gradually develops as more information becomes available (Hung et al., 2004; Mayer et al., 1995).

Others have distinguished between cognitive trust and affective trust (McAllister, 1995). Cognition-based trust is rooted in searching evidence or clues of trustworthiness, building upon “good reasons” to trust the person [primarily based on the ability and integrity components of trustworthiness (Tomlinson et al., 2020)]. Affect-based trust, on the other hand, is built on the emotional bonds between individuals, which grow stronger as the relationship develops (McAllister, 1995; Tomlinson et al., 2004) and corresponds mostly to perceived benevolence and shared values (Tomlinson et al., 2020).

People quickly form first impressions of others’ trustworthiness, before even meeting them. These first impressions are built based on various cues such as facial features (Todorov et al., 2008), social indicators like in-group membership (Hughes et al., 2017), and third-party information (Ferrin et al., 2006). This initial impression is then updated as people interact and gather new information about the trustee’s trustworthiness (Hung et al., 2004; Mayer et al., 1995). As the relationship consolidates, trust judgments become more resistant to new information (Cvetkovich et al., 2002; Dietz and Den Hartog, 2006). They thus evolve from a cognitive base to a more affective one (McAllister, 1995).

Different models suggest ways in which trust beliefs are updated in response to new information. First, the asymmetry principle states that negative information has a greater impact than positive information (Slovic, 1993; see also Kahneman and Tversky, 1979). Therefore, and not surprisingly, trust is more easily lost than built (Lewicki and Brinsfield, 2017). Furthermore, information is not always given equal weight. Research has highlighted a belief-congruent or confirmatory bias, wherein new information that is congruent with initial beliefs is given more consideration (Cvetkovich et al., 2002). For example, Poortinga and Pidgeon (2004) found that participants’ prior beliefs moderated the importance they assigned to new positive vs. negative information about genetically modified food. While early research mostly focused on risk assessment (e.g., Cvetkovich et al., 2002; Poortinga and Pidgeon, 2004; Slovic, 1993), other evidence suggests that judgments of individuals’ trustworthiness are similarly biased, be it as a function of the trustee’s reputation (Pilditch et al., 2020) or implicit initial trustworthiness information (Chang et al., 2010). Experimental research consistently suggests that early trust violations make it very difficult to restore trust even when the trustee later demonstrates consistently trustworthy behaviour (Lount et al., 2008). In summary, it appears that trustors do not simply take in and summarise new information, updating their trust beliefs as they go. Instead, they treat this information in a way that is biased both cognitively (e.g., the asymmetry bias) and motivationally (e.g., the belief-congruent bias).

What about trust perceptions in the workplace? Organisational trust can be apprehended at many levels and in different forms, from the inter-organisational level (i.e., between organisations) to the intra-organisational level (encompassing trust in different referents, such as the organisation itself, one’s manager and coworkers, etc., see Dietz and Den Hartog, 2006; Fulmer and Gelfand, 2012; Tyler, 2003). Here, we focus on the processes of team trust and specifically on interpersonal trust among team members.

Team trust represents the aggregated perception of trust in the team as a distinct unit or in its members collectively (Costa, 2003; Costa et al., 2018). Different theories suggest that team trust stems from interpersonal trust between team members (Costa et al., 2018; Kozlowski and Klein, 2000; Marks et al., 2001). Recent work also demonstrated that initial trust judgments emerge quickly, even in short-term teams, and that such individually held perceptions influence the construction of team trustworthiness over time (Fletcher et al., 2024).

While team trust may be a desirable feature in itself, its importance stems mostly from its vital role in fostering harmonious teamwork (Costa et al., 2018; McAllister, 1995). Among other things, trust has been found to contribute to increased perceptions of collaborative culture (Barczak et al., 2010) and team performance (Costa et al., 2001; Erdem et al., 2003; see also Grossman and Feitosa, 2018), as well as greater team satisfaction (Braun et al., 2013; Chou et al., 2008) and continuance commitment (Costa et al., 2001; Matzler and Renzl, 2006; for reviews, see, e.g., Breuer et al., 2016; Costa et al., 2018; De Jong et al., 2016).

Previous research on belief updating has explored interpersonal trust in general, but has rarely addressed dynamics specific to the workplace (for exceptions, see, e.g., Campagna et al., 2022; van der Werff and Buckley, 2014). In the present study, we aimed to extend research on trust in the workplace by integrating findings on trust based on first impressions. We thus investigate building trust in a new coworker and its downstream consequences. Going beyond just interpersonal dynamics, we consider how the dynamics of building and updating trust beliefs about this coworker might then translate into different perceptions of one’s work team (as listed above), in anticipation of the coworker joining it. Indeed, previous studies have shown that the arrival of a new coworker may shake up existing team dynamics (Allen and Meyer, 1990; Bauer et al., 2007; Liu et al., 2023) and threaten trust amongst team members. Thus, this study aims to provide a thorough assessment of trust formation and belief updating in the workplace by articulating theoretical frameworks that, to date, have remained largely disconnected.

In the present study, we utilised an experimental design that allowed us to isolate the specific mechanisms of interest, all else being equal (e.g., potential confounds related to participants’ diverse experiences in their workplace). We manipulated the first impression of an alleged future coworker as more or less trustworthy before introducing new positive information about the person. This allows us to test how initial trust beliefs are updated. We then turned to team-related beliefs, assessing participants’ beliefs about how team dynamics may be affected if and when the new coworker joined their team.

A strength of this research is that our sample is composed of employees (rather than just university students), ensuring an understanding of workplace dynamics and allowing participants to project themselves into a potential future where the job candidate would join their real-life work team. This extends previous findings that only considered graduate and undergraduate students. We now outline the various components of the study, along with the related preregistered hypotheses.

The study design, hypotheses, materials, sample size, and exclusion criteria were preregistered: https://aspredicted.org/sfkb-sg4t.pdf. Data, materials, and code for analysis are publicly available on the OSF: https://osf.io/vyzkx. This research was approved by the ethics committee of the Faculty of Psychology at the University of Basel, and all participants gave informed consent prior to their inclusion in the study.

We invited employees from Northwestern Switzerland to participate in an online study. Conditions for participation were being employed in a team of at least three members and being fluent in German. A partnering company agreed to distribute the study directly to their employees. We also advertised it on social media and through the university network. Participants completed the study voluntarily and were eligible to enter a draw to win one of several vouchers as a token of appreciation. As indicated in the preregistration, we determined the sample size based on an a priori power analysis, aiming to detect small-to-medium effect sizes (i.e., Cohen’s d = 0.40). The analysis indicated that a sample of N = 200 would allow us to detect such an effect with 0.80 power (alpha = 0.05), so we set our target sample size at 200.

Two hundred and fourteen participants completed the survey. As preregistered, we excluded participants who failed an attention check embedded in the survey (n = 3), those who worked in teams smaller than three members (n = 3), and those who were not currently working (n = 4). The final sample consisted of N = 204 (83 men, 118 women, one non-binary individual, and two did not disclose their gender; M_age = 33.97, SD = 12.27).¹

The study was introduced as an investigation of team dynamics in the work context. Figure 1 depicts the study procedure. We asked participants to imagine that their work team was in the process of recruiting a new member, about whom they would now receive some information. They first watched a short video of a bogus job interview, which served as our experimental manipulation of the first impression of the (female) job candidate based on her response (see below). They were randomly allocated to either the Apology (n = 104) or Denial (n = 100) condition. They completed a brief measure of trust perception before being presented with new information about the candidate. This positive information was presented to all participants. Then, they completed a longer set of items that measured cognitive and affective trust in the candidate, perceived trustworthiness, and willingness to risk. Following this, we asked them to imagine that the job candidate had been hired and would now join their team, and to reflect on how this might affect team dynamics. They then completed the final measures of collaborative culture, perceived team performance, continuance commitment, and satisfaction with team functioning. They were finally thanked, debriefed, and allowed to enter a lottery to win a voucher.

At the theoretical level, the primary goal of the present research was to shed light on the dynamics of trust formation in the work context, specifically testing how first impressions and new information interact to determine interpersonal trust. Consistent with confirmation bias research (e.g., Cvetkovich et al., 2002; Pilditch et al., 2020; Poortinga and Pidgeon, 2004), we expected participants to treat the new positive information in a way that was congruent with their first impression based on the candidate’s behaviour during the interview (i.e., denial or apology). This implies that trust would increase to a greater extent when the first impression was positive. Although the apology response led to a more positive first impression than denial, consistent with our first preregistered hypothesis (H1) and directly replicating past literature (Ferrin et al., 2007; Kim et al., 2004, 2006), the results revealed no confirmation bias in treating the new information (contradicting H2). Rather, they showed an additive effect, as the piece of positive information ameliorated trust to a similar extent in both experimental conditions.

This additive effect appears elsewhere. For example, Chang et al. (2010) found that a trustee’s consistent trustworthy behaviour in a trust game (as the experimental manipulation of ‘new information’) led to high levels of trust, but also that facial features of high vs. low trustworthiness (as the proxy for ‘first impression’) continued to influence trust.

Different explanations may account for these findings. First, it is likely that the timing of receiving the different pieces of information matters. Early work on confirmation bias considered participants’ pre-existing beliefs, combined with a new piece of information (Cvetkovich et al., 2002; Poortinga and Pidgeon, 2004; Slovic, 1993). In contrast, in the present study (as in Chang et al., 2010), the information that contributed to the first impression of the candidate and the new information were presented consecutively. It is possible that dynamics of motivated reasoning (Kunda, 1990) and biased processing of new information (Hilbert, 2012) become stronger when the initial beliefs are well consolidated – explaining why we could not identify them in the present research, as trust beliefs may still have been very malleable. It is worth noting, however, that other work has identified a confirmation bias in trust in a stranger when first impressions (i.e., information about the person’s credibility) and new information were both gathered during the short timeframe of the experiment (Pilditch et al., 2020).

More broadly, this speaks to the ongoing debate in the literature as to whether first impressions only matter during the early stages of a relationship (e.g., Kramer, 1999; Lewicki and Bunker, 1996) or whether they have a lasting influence (Ferrin and Dirks, 2003; McKnight et al., 1998). In a recent series of experiments contrasting short-term and long-term effects, Campagna et al. (2022) successfully demonstrated that initial positive information of trustworthiness buffers against negative updating in a direct paradigm. However, this effect disappears if a two-week gap is introduced after the trust violation, and then, only the effect of the violation remains. The notion of timing is also discussed in recent reviews of trust repair, with Sharma et al. (2022) highlighting that the effect of introducing a delay between the trust violation and trust repair efforts remains unclear. In summary, further research is required to improve our understanding of the temporal dynamics of trust belief updating (Korsgaard et al., 2018; van der Werff and Buckley, 2014).

A second explanation might be that different dynamics underlie the updating of a negative first impression in light of new positive information (as in the present study) and that of a positive first impression in light of new negative information. Researchers have suggested that trust is a dominant response tendency, implying that people are generally more prone to trust than distrust (Katzir and Posten, 2023). Consistent findings from an aforementioned study indicated that a first positive impression of someone’s trustworthiness continues to positively affect perceived benevolence and cooperative behaviour even after untrustworthy behaviour: the effect of the first positive impression endures (Campagna et al., 2022). In another study investigating trust in the recommendations of credible vs. non-credible sources, Pilditch et al. (2020) also found that the tendency to believe (false) information from credible sources was stronger than the tendency to reject (true) information from non-credible sources, which may point again to a general motivation to trust. However, other research has yielded opposite findings. In Chang et al. (2010), participants trusted the least the partner who had initially looked trustworthy (positive first impression) but had later betrayed their trust in an economic game (negative new information).

Overall, evidence is mixed regarding trust belief updating, and it appears that the dynamics at play depend on the direction of that updating (from negative to positive and vice versa), as well as on the nature of the information being processed (e.g., facial cues of trustworthiness, reputation, behaviour, etc.). We noted that research on belief updating has remained somewhat disconnected from the trust repair literature (for recent reviews, see, e.g., Kähkönen et al., 2021; Sharma et al., 2022). Although trust repair constitutes a specific case of belief updating (focusing on ways to rebuild positive beliefs after a trust breach), some insights could be better connected to belief updating in general, especially in terms of trust repair strategies and processes (see also Gillespie and Dietz, 2009).

In general, updating seems to be coloured by a broad motivation to trust and therefore to look for cues of trustworthiness even when doubt creeps in. The apparent dominant nature of trust (Katzir and Posten, 2023) may express itself more strongly when the trustee is a person that participants would expect (or imagine) to interact with again in the future. No motivated reasoning is necessary when one plays a trust game against an anonymous partner whom they will never encounter again. However, when the person is for example a new colleague, in reality (van der Werff and Buckley, 2014) or in an alleged scenario (as in the present study), people might be motivated to convince themselves that this new person is better than they might have initially thought and actively look for cues of trustworthiness to correct their initial negative impression.

The present study found a positive effect of apology over denial and a constant positive effect of the additional piece of positive information, resulting in a sustained effect of the positive first impression on trust, trustworthiness, and willingness to make oneself vulnerable. The effect was more pronounced for cognitive trust than for affective trust, which is consistent with the view that cognitive trust is more relevant at the early stages of a relationship while affective trust develops later (McAllister, 1995; Tomlinson et al., 2004). Apart from affective trust, all effects were of medium size (0.41 ≤ Cohen’s d_s ≤ 0.68).

The same pattern of results emerged as we examined team dynamics: here, too, we found a sustained effect of the positive first impression on perceived collaborative culture (H3a), anticipated team performance (H4a), and satisfaction with team functioning (H6a). Only the effect on continuance commitment (H5a) was non-significant. These effects were relatively smaller (0.35 ≤ Cohen’s d_s ≤ 0.50), which is potentially attributable to the nature of the variables, as they move further away from the core concept of trust and are, rather, consequences of it. Participants’ ratings were also most likely influenced by their perceptions of their own team, beyond the manipulated information about the projected new team member. This is reassuring in a way as the difference in effect sizes suggests the team-related outcomes do not merely reflect attitudes towards the new team member, but rather participants’ integrated view of how the member might affect the team. However, these outcomes remain a projection in a fictional scenario, which is discussed in the Limitations section.

Our mediation analyses further supported the hypothesis that the observed effects on subjective team dynamics were due to a differential level of trust in the new team member (mediation through trust was significant for collaborative culture, team performance, and satisfaction; H3b, H4b, H6b). We also replicated previous findings that the effect on satisfaction is serially mediated through trust first, then team performance (H6c; Costa et al., 2001). Overall, our findings therefore strengthen the idea that interpersonal trust within a team is important for team perceptions.

At the applied level, we sought to investigate how team perceptions may be challenged by the arrival of a new team member, an event that has been recognised as a potential challenge for teams (Allen and Meyer, 1990; Bauer et al., 2007; Liu et al., 2023) especially when the trustworthiness of this person may be questioned. Our results identified clear effects on team perceptions, highlighting that the anticipated arrival of a new team member can be a challenging time to negotiate. They suggest that it would be essential to provide team members with support, transparent information, and a space to express their concerns, thereby facilitating a smooth transition to the new team composition.

Trust violations are inevitable in the workplace, and at some point, one may need to disclose past trust violations, such as to a new employer. Consistent with previous research, our findings suggest that the best way to handle such disclosure is for one to own their previous mistake, provide an explanation and reassurance, and to apologise for the violation. This holds, at least, as long as the violation is competence-based (Ferrin et al., 2007; Kim et al., 2004, 2006). Importantly, we also showed that subsequent evidence of one’s trustworthiness will continue to ameliorate this first impression. Providing such evidence may be a clever strategy for both the newly hired team member (see also Kim et al., 2013) and team leaders or managers who can reassure team members of their hiring choice and ensure the newcomer is well integrated into the team (Sharma et al., 2022).