Can I Trust You? A Critical Review of the Perceptual,Social and Cognitive Influences on Trust

Abstract

Trust is a complex psychological phenomenon that is foundational to coordinated behaviour and societal function. This narrative review critically examines the perceptual, social, and cognitive indicators of trust, drawing on findings from multiple fields of behavioural science to identify theoretical and empirical gaps in our current understanding. Particular attention is paid to social identity cues, perceptual features like eye gaze, pupil dilation, and facial characteristics, and memory recall factors related to threat and reliability judgements, to highlight the potential domains of functional overlap that are ignored by narrowly focused research approaches. The review also identifies critical gaps in understanding the interplay of multimodal sensory cues, the impact of trust violations on memory encoding and retrieval, and the integration of trust cues in ecologically valid settings. To ground the importance of the dynamics of trust formation, the paper then overviews some demonstrable effects trust has on other domains of decision-making, ranging from financial to legal. Recommendations for future research emphasise the need for interdisciplinary approaches to account for these complexities. Pursuing these empirical directions could lead to significant practical advances in the understanding of the cognitive mechanisms underpinning trust, and their application to real-world decision-making settings where trust is essential.

Trust is essential to coordinated behaviour and the functioning of society at large. Trust is influenced by numerous factors such as past experiences, social norms, contextual factors, and individual differences. This narrative review surveys and critically evaluates the perceptual, social, and cognitive factors linked to trust, highlighting gaps in the literature and suggesting future directions for research. The need for more reliable and ecologically valid findings in this field speaks directly to current challenges in translating psychological research on trust into consequential real-world applications, such as the targeted building of trust in social and institutional contexts, and resisting the erosion of positive-sum civic engagement (Hamm et al., 2024). The authors’ intentions in reviewing and critiquing this range of disparate fields in a single paper, in addition to their threads of influence in applied decision-making domains, are to illustrate the neglected grounds for overlap between the different areas of study, and how the variables of interest in one field are often missing confounding influences in the methodological assumptions of another field. The penultimate section of this review highlights a subset of the well-examined influences of trust judgements on decision-making to illustrate some of the practical implications of shifting levels of trust and to summarise for readers what we stand to gain from a more multifaceted understanding of trust.

What is Trust?

From a multidisciplinary perspective, the most general and universally applicable definition of trust can be summarised as ‘a psychological state comprising the intention to accept vulnerability based on positive expectations of the intentions or behaviour of another’ (Rousseau et al., 1998, p. 395). Trust, in this broad and inclusive sense, has several important functions, including facilitating cooperative behaviour, collaboration, open communication, and reducing conflicts (Chen et al., 2021; Conchie & Donald, 2008; Rousseau et al., 1998). The construct of trust is a foundational component of most forms of consequential social interaction and has been conceptualised in various ways to serve the interests of disciplines as diverse as philosophy, social sciences, psychology, and neuroscience (Thielmann & Hilbig, 2015). The foundational trade-offs between accepted vulnerability and positive expectations have also been modelled in game theoretic terms, which help quantify the corrosive effects of dispositional tendencies such as paranoia and protracted loneliness (Bellucci, 2022). Trusting information at face value is the default state of mind for most individuals as long as no conflicting interests are present (Schul & Peri, 2015), resulting in perceptions of similarity and congruence, and a habitual overlooking of potential alternatives (Posten & Mussweiler, 2013). Indicators of unreliable systems and past experiences of trust violations often lead to individuals distancing themselves from those judged to be untrustworthy, increasing reactive scepticism (Gobin & Freyd, 2014; Platt & Freyd, 2015). There are significant individual differences that influence the thresholds of these judgements for extending and withdrawing presumed trust. While a full review of the specific personal and dispositional factors that have been shown to influence trust cognition is beyond the scope of this review, examples range from age cohort effects (older adults judge others to be more trustworthy, Bailey & Leon, 2019), to common psychopathological influences (such as individuals with higher levels of social anxiety judge others as less trustworthy; Gutiérrez-García & Calvo, 2016), to developmental influences connected to attachment and upbringing (supportive parental relationships increase proclivity towards forming bonds of trust; Yilmaz et al., 2023; Zhao et al., 2024).

One commonly studied type of trust in the literature is general (or dispositional) trust, defined as the attitude towards any individual and/or institution (Acedo-Carmona & Gomila, 2014; Jasielska et al., 2021). This type of trust is believed to play an important role in facilitating cooperation, communication, and reputation, by enabling a default prosocial trusting position in untested situations (Acedo-Carmona & Gomila, 2014). Individuals also build up particularised judgements of personal trust, which constitute one’s feelings of certainty that a specific counterpart would be loyal and choose to cooperate (influencing one’s own risk assessments and inclination to cooperate; Acedo-Carmona & Gomila, 2014). In other words, personal trust between two people is based on previous positive interactions and often leads to a prosocial attitude between them (Acedo-Carmona & Gomila, 2014). The current review mainly focuses on cognitive factors influencing general trust but will address particularised judgements of personal trust in some domains where it is influential.

The Current Review: Aims and Unique Contributions

A range of social, perceptual, and cognitive factors influence levels of trust. The definitions and groupings of these factors typically cluster together based on the research area of interest. However, a synthesis of these key findings is lacking in the literature, and therefore the current review makes two unique contributions to the literature: (a) it summarises the existing findings on perceptual, social, and cognitive factors that impact the judgements of trust and related trusting behaviours and (b) reviews these previous studies through a critical lens to highlight empirical gaps and provide potential directions for future research. Below, we first review the factors identified as contributing to trust and distrust in three major domains: social-emotional context, perception, and memory, concluding each section with an overview of the major limitations in that literature. This is followed by a brief overview of the tangible impacts trust levels have on several consequential domains of decision-making (including legal and financial decision-making), and the review concludes by highlighting existing problems and gaps in the trust literature, along with suggestions for future interdisciplinary research and theoretical clarifications.

The Socio-Emotional Context of Trust Judgements

Before exploring the specific cognitive contributions of perception, memory, and decision-making, it is crucial to first review the influence emotional states and perceived social context can have on both general and specific judgements of trust. As later sections will explore in greater depth, socio-emotional context cues related to group membership, social threat, and concealed intention are some of the most widely overlooked potential confounds in studies examining trust judgements. A deeper understanding of these cues is essential to designing more ecologically valid studies of trust.

Emotions

Emotional states, both of the target and the perceiver, influence judgements of trustworthiness that extend beyond the predictable impact of positive and negative affect. Empirical evidence suggests that happy faces are viewed as more trustworthy than faces displaying other emotions, such as anger or disgust (Galinsky et al., 2020; Sutherland et al., 2017). The emotion of anger reduces trust, arguably because individuals displaying anger are less likely to engage in affiliative behaviour (Dunn & Schweitzer, 2005; Said et al., 2009). However, this relationship between facial expressions of emotion and trust is complicated and appears to be dynamic. For example, Calvo et al. (2018) found that dynamic expressions ending with a smile were judged as both happier and more trustworthy than those ending with a neutral mouth, irrespective of the eye expressions. A momentary happy mood in the perceiver is associated with higher attributed trust levels (Mislin et al., 2015). Momentary affect can also lead to the erosion of trust, with empirical evidence showing that negative momentary mood is associated with higher levels of attributed distrust in perceivers (Schmitz et al., 2021). Similar findings have been shown in relation to decision-making about other’s behaviour. For example, judges in a negative mood were significantly more likely to decide that a person was guilty than those with neutral or positive moods, suggesting that negative mood had undermined the perceived trustworthiness of the mitigating factors for those being sentenced (Forgas & East, 2008). Similarly, a negative correlation has been found between feelings of subjective disgust and trustworthiness ratings (Kugler et al., 2020). It must be noted, however, that emotional valence does not always map onto trust in predictable ways. Baer and colleagues (2014) demonstrated that the perception that another party is willing to place their trust in you can lead to feelings of exhaustion and rumination around performance, suggesting that being trusted can give rise to emotional states that might otherwise be expected to erode bonds of trust.

Social Identity Cues

The expectation that an individual will reciprocate or honour a social norm draws in part on group-level expectations and generalisations (Tanis & Postmes, 2005). In-group members are broadly trusted to behave more prosocially, particularly when mutual acknowledgement of shared group membership is possible (Platow et al., 2012). Out-group members are broadly viewed as less trustworthy, and this effect is enhanced by relevant negative stereotypes and recent exposure to betrayals of trust by others who share out-group social identity cues (Chou et al., 2022). Individual-level cues of trustworthiness are more influential for in-group members as compared to out-group members (Olcaysoy Okten et al., 2020). Further, when the individual targets of trustworthiness judgements are entirely unknown or unidentifiable, cues of social identity, and coalitional group membership are relied upon for proxy judgements of trustworthiness (Tanis & Postmes, 2005).

Future Directions for Socio-Emotional Influences on Trust Research

Emotional displays, particularly happiness and anger, have been shown to influence judgements of trustworthiness, but the complexities of these relationships, including the roles of moderating factors, are not fully understood. Given the embodied nature of much of emotional experience, and our visceral reliance on empathically modelling the perceived emotions of others when making emotionally informed social decisions, it is at this point unclear how the emotional influences on trust judgements interact with cues such as pupil dilation, roaming gaze, and vocal modulation. While it is possible that the observed emotional impacts on trust judgements are partially comprised of these body-language cues that have been studied in isolation, it is also possible that the context of expected emotional judgement may strongly influence the interpretation of the specific cues, something which could only be disambiguated with carefully designed future studies. Finally, concerns relating to the prospect of emotion-signal miscommunication across sociocultural lines of difference (e.g. cultural variations in the expression norms of aggression and displeasure) add a further complication that the present literature is yet to be addressed.

Perception and Trust

Judgements of trust are intimately informed by the perception of our interlocutors, the context of interactions, and at times even our own internal sensations. While the following areas of this subsection will mostly overview the direct cognitive impacts that specific perceptual cues make on judgements of trustworthiness, it is important to first clarify the global affects that preceding mental states of trust and distrust have on perceptual processes. Mounting evidence suggests that people adapt information-processing strategies under trust and distrust to the needs of these fundamentally different states of mind (Mayo, 2015; Posten & Gino, 2021). Interpersonal judgements, particularly those involving social comparisons, can both influence and be influenced by evaluations of trust in ways that magnify initial impressions and biases. For example, individuals who felt higher levels of trust judged members of a shared social group as more alike in their traits and characteristics compared to those who experienced relative distrust. This enhanced stereotype-reliance and inflated perception of similarity can subsequently deepen those initial judgements of trustworthiness, and create situations where more accurate (or at least, less stereotype-reliant) social perception is notably more likely only under conditions of mistrust (Posten & Mussweiler, 2013). With that caveat provided, the remainder of this section will summarise studies that examined the links between perception and trust via eye gaze or pupil dilation, interoception and bodily cue integration, vocal confidence/audio features, age, facial structures/features, and multimodal integration.

Pupil Dilation and Gaze Cuing

Pupil dilation is considered a significant biomarker of trustworthiness. In a recent study, dilated pupils produced positive impressions, suggesting that pupil dilation could be linked with signals of trustworthiness (Okubo et al., 2023). Indeed, two studies found that participants tend to view individuals with dilated pupils as more trustworthy as compared to those with constricted pupils (Kret & De Dreu, 2019; Kret et al., 2015), and pupil dilation does not require conscious awareness to influence judgement of trustworthiness (Prochazkova et al., 2022). It has also been shown that pupil dilation mimicry consistently promotes trust among partners (Prochazkova et al., 2018).

In addition to pupil dilation, the eye conveys trustworthiness via direction of gaze. Gaze-cueing of attention can be defined as the process of shifting attentional resources in response to the eye-gaze direction of another person (Rogers et al., 2014). It has been consistently reported that attentional shifts in a direction cued by another person’s gaze can affect how that individual is perceived. Individuals whose gaze accurately predicts the location of a target are considered more cooperative and trustworthy than individuals with gaze that does not predict a target location (Barbato et al., 2020). Further, once trust has been established, people have been shown to allocate more attention in the form of gaze following to those who are perceived as trustworthy (Süßenbach & Schönbrodt, 2014). This in turn influences subsequent judgements concerning the maintenance or even deepening of these trust evaluations. This ability to infer and learn trust from gaze can also be extended at a collective level. In other words, one can infer and make trustworthiness judgements based on not only the gaze of an individual but also a collective group’s eye gaze patterns (Sun et al., 2020).

Interoception and Bodily Cue Integration

A perceptual domain that also influences perceptions of trustworthiness relies on interoception and the integration of bodily cues. An example of this is cardiac rhythm. One study found that faces presented in synchrony with the participant’s cardiac systole were chosen less often as trustworthy and explicitly judged as less trustworthy than those presented asynchronously (Azevedo et al., 2022). Similarly, increasing cardiac signals generates heightened bodily arousal, which is argued to generate neural noise, reducing one’s confidence in internal signals and shifting weight to external social information (von Mohr et al., 2023). This finding suggests a meaningful role for interoceptive cues alongside conventional perception in informing judgements of trustworthiness. Interoceptive accuracy appears to play a role in social connection and trust in others. Specifically, better accuracy in sensing internal signals is linked with experiencing more cognitive and affective empathy for others (Grynberg & Pollatos, 2015), more social understanding and connection (Hess & Fischer, 2013), and the degree to which individuals experience an emotional contagion for a positive, but not negative, emotion (Terasawa et al., 2014). Likewise, covert presentation of positive faces can increase trust in neutral faces (an effect known as ‘affective realism’), and this effect is more pronounced in individuals with poorer interoceptive sensitivity (Feldman et al., 2022), suggesting that interoception can influence subjective experience and affective meaning imbued on an event. Therefore, use of bodily signals may facilitate adaptive learning from social and affective experiences. In fact, there is evidence linking the neuronal markers of interoception to trusting behaviours (Kirk et al., 2011). Specifically, when comparing Buddhist meditators to controls, meditators not only accept more unfair offers in the ‘Ultimatum game’ but also have increased activity in the low-level interoceptive representations of the posterior insula and attenuated activity in high-level emotional representations of the anterior insula – a brain area associated with violations of social norms and mistrust (Rilling et al., 2002).

Vocal Confidence and Audio Features

People who orally express confidence in speech are viewed as more believable (i.e. more trustworthy; Jiang et al., 2020). Further, certain acoustic characteristics are associated with greater perceived trustworthiness, including accelerated tempo, low harmonic-to-noise ratio, more shimmer, low fundamental frequency, more jitter, large intensity range, and positive speaker valence (Schirmer et al., 2020). For instance, a study found that hello, which consisted of a high starting intonation (f0 trajectory), followed by a significant decrease at mid-utterance, and then finishing on a strong rise, was rated higher on trustworthiness (Belin et al., 2017). Overall, the audio quality of individual voices tends to strongly influence initial judgements about trustworthiness.

Perception of Age

Preliminary research provides mixed evidence for a relationship between perception of age and trustworthiness. One study found that young faces were rated as more trustworthy than middle-aged and older faces (Pehlivanoglou et al., 2023). In two further studies, faces of older adults were rated as more trustworthy than faces of younger adults (Bailey et al., 2015; Zebrowitz et al., 2013). In a subsequent trust game, Bailey et al. (2015) found that, despite the greater reported trust in older adults, participants did not invest more money with older adults relative to young adults. This suggests that ratings of trustworthiness may not be a good indicator of actual trust, or that trust in older adults may be domain specific and may not extend to financial investments. It may also be that the influence of perceived age on trust depends on additional factors. For example, recent work examining moderators of the association between age of face and perceptions of trustworthiness found that people trust older faces more than young faces when the faces displayed low attractiveness and happy or sad emotions (Li et al., 2021). There is also some evidence for an own-age bias in the association between perceived age and trustworthiness. For example, older adults were more likely to think that someone telling a lie was telling the truth if the speaker was an older adult as opposed to a young adult (Slessor et al., 2014). However, this own-age bias was not evident among the young adults, and a further study found no own-age bias in ratings of facial trustworthiness (Zebrowitz et al. (2013). The precise nature of interactions between perceptions of age and trustworthiness, and their relevance to activation of the high warmth/low competence stereotype of older adults (Fiske et al., 2002), requires further empirical exploration.

Facial Structure and Features

The assessment of facial trustworthiness in the literature is often believed to be based on facial features rather than visual context. Facial width-to-height ratio has been shown to impact the judgements of trustworthiness. Specifically, faces that are perceived to have smaller heights (compared to widths) are viewed as less trustworthy and more aggressive (Costa et al., 2017; see also Foo et al., 2022). Moreover, faces that are perceived as attractive are considered more trustworthy than faces with low levels of attractiveness, also known as the Halo Effect (Schmidt et al., 2012). Most studies use isolated facial expressions, a rare occurrence in the real world (Brambilla et al., 2018). The limited evidence suggests that visual context (e.g. threatening vs. neutral scenes) indeed impacts ratings of trustworthiness, such as threatening contexts tend to increase the levels of untrustworthiness for untrustworthy faces (Mattavelli et al., 2022).

Thinking through the lens of the COVID-19 pandemic, wearing face masks is believed to have several health benefits. As a result, studies have begun to examine the impact of face masks on trustworthiness. Empirical evidence suggests that participants tend to judge masked faces as less approachable and less trustworthy (Biermann et al., 2022; Bylianto & Chan, 2023). However, another study found that participants viewed masked faces as more trustworthy and approachable (Guo et al., 2021). This contradictory finding could be explained by the mandated practices and social conformity related to COVID-19. Further, before the pandemic, surgical masks were often worn by those who work in national health services (e.g. doctors, nurses, and dentists), who are often perceived as trustworthy and approachable (Guo et al., 2021).

Multimodal Integration

Multimodal integration refers to attending and combining cues encountered in different sensory modalities (e.g. emotions and eye gaze) and several communication channels simultaneously (Tsiourti et al., 2019). A recent study examined cooperation (behaviour) and trust perceptions of virtual agents displaying smiles either congruently across face and voice or incongruently in only one channel. Incongruent expressions reduced perceptions of trustworthiness and knowledgeability, which in turn lessened the agent’s social influence on participants’ cooperative behaviour (Torre et al., 2021). Another study found that faces were viewed as untrustworthy when accompanied by a threatening auditory context compared to faces with a non-threatening auditory context (Brambilla et al., 2021). These preliminary studies support a link between multimodal sensory integration and trust.

Future Directions for Perceptual Influences on Trust Research

Cues signalling social identity play crucial roles in trust judgements, but in ecologically valid settings, it is unclear how indicators of group membership may interact with the social dimensions of visual context. For example, situational ambiguities that can be interpreted as threatening may be processed differently if targets show visual signifiers of ingroup membership, something that is rarely even considered in the methodologies of studies not directly assessing intergroup conflict. The limited research on how threatening versus neutral scenes impact trustworthiness ratings call for more comprehensive studies that can distinguish between threats that are concretely implied in a presented scene, and more subtle implied threats that can be inferred from cues of potentially antagonistic social relationships (e.g. being outnumbered by outgroup members in a context that may turn competitive or hostile). This approach could also be effectively combined with signals from gaze cueing, but it is worth noting that previous studies have not examined whether the perception of an immediate threat influences the effect of gaze-cueing on trust and whether this would amplify sensitivity to the gaze of trustworthy or untrustworthy faces. This relationship may need to be disambiguated before such elements can be meaningfully incorporated into broader studies.

There is preliminary evidence for the impacts of multimodal integration, the combination of cues from different sensory modalities, on general judgements of trust. However, research on this topic is scarce, particularly regarding whether these judgements occur implicitly or explicitly and how they influence downstream behaviours. One particularly viable direction, given the findings on threatening versus neutral visual contexts, would be to explore whether untrustworthy faces processed in a threatening context (cross-modal) are more likely to foster physical avoidance by the perceiver compared to targets processed in a single mode (only visual for faces or auditory for vocal cues).

Memory and Trust

In this section, we will focus on links between different facets of memory, including general recall, and working memory updating, and subsequent evaluations of trust.

Prior Learning and Future Trust Decisions

Prior learning tends to influence decisions related to trust. For instance, Chou et al. (2022) found that strangers who resembled trustworthy individuals from the past were trusted more, whereas those who resembled individuals from the past who were untrustworthy were trusted less. Similarly, individuals who perceive a recent betrayal by a trustee are less likely to trust a new entity that shared nominal group membership with the previous trust transgressor (Chou et al., 2022). This is also known as the Betrayal Spillover Effect, which is prevalent both in economic game environments and real-world charitable contexts (Chou et al., 2022). Thus, this finding suggests that prior learning/betrayals could influence trust in novel contexts. However, it is unclear how long individuals would generalise past betrayals to novel situations (i.e. the time frame is unclear) and what magnitude of betrayal would influence future trust-related decision-making. Overall, research focussing on the association between learning and trust is relatively scarce outside these few memory-focused studies, offering a fruitful avenue for future research across a broader context.

General Recall

The perceived and attributed trustworthiness of a target has a range of influences on recall of information about their behaviour. Empirical evidence suggests that incongruent associations activate even when the domain of distrust is not related to the recalled message, and recipients are unable to formulate a strategic response (Schul et al., 2004). Generally, untrustworthy faces are remembered better than trustworthy faces (Mattarozzi et al., 2015; Rule et al., 2012), which is a robust finding that has been demonstrated in both lab and online settings (Giraudier et al., 2022). These findings indicate that subtle cues of untrustworthiness are automatically learnt and remembered.

These memory effects appear to be long term. Specifically, a study found that untrustworthy faces are recognised better than trustworthy faces after a week’s delay, which also resulted in larger event-related potential (ERP) differences during delayed recognition (Weymar et al., 2019). Further, people in a general state of distrust, those primed to be on the lookout for deceitful individuals, tend to perform better at memory recognition tasks than those in a trustful or neutral state (Posten & Gino, 2021). This memory advantage for untrustworthy faces may serve an adaptive function to avoid potential exploitation and harm in future social interactions.

Working Memory Updating

Another facet of memory where trust is relevant is working memory updating (WMU). This type of memory could be defined as the process wherein incoming information is monitored and coded for relevance to the task at hand, then appropriately replaced when that information is no longer relevant to the current situation (Morris & Jones, 1990). Trust violations appear to impair our ability to hold and update information in working memory. A study found that after a trust violation occurred in a trust game, participants had poorer performance in an n-back task than the control group (Yuan et al., 2023). This is supported by ERP evidence that trust violations impair emotional memory processing (Yuan et al., 2022). These results could be explained by the strength model of self-control (Baumeister & Vohs, 2016). This model posits that cognitive resources are limited, and individuals are unable to allocate resources to subsequent cognitive tasks once they experience the consumption of such resources. When experiencing trust violations, people deploy resources for social adaptation, and thus, the deficits in WMU after trust violations may be related to the exhausted self-control resources (Yuan et al., 2023). When dealing with trust dilemmas, older adults are less likely to rely on the frontoparietal network linked to working memory and executive functions, and instead their propensity to trust is driven by activity in the default mode network and cingulo-opercular network (Chen et al., 2023, 2024). Overall, the attentional biases and cognitive load of anticipated and actual trust violations significantly impact working memory function.

Future Directions for Memory and Learning Influences on Trust Research

While we have discussed the general role of memory in trust judgements, it is unclear whether impacts of trust and distrust on memory processes occur during the encoding phase, the decoding phase, or both phases and, if so, then to what extent. Future research should investigate boundary limits on these effects. For example, testing whether unrelated, standalone content, that is hardly relatable to other content, could constitute a boundary condition to the effect of trust on episodic memory. It is unclear, at this stage, if remembered elements are separated enough in memory, and whether the similarity processing effect will still occur. Another area worth examining is memory consolidation for specific people who have acted in trustworthy or untrustworthy ways towards the subject in the past, and whether consolidation varies for individuals high or low on general disposition to trust. Finally, another avenue for future research would be to include a period of consolidation followed by a gaze-cueing session, to see if consolidation of information about untrustworthiness can be used to strategically inhibit the automatic reorienting of attention to these cues, expanding both the memory and gaze-cueing literature on trust.

The Impact of Trust: Known Influences of Trust Judgements on Decision-Making

To clarify the consequential nature of the trust judgements outlined in the preceding sections, this section provides an overview of general decision-making, financial decision-making, temporal risk decision-making, legal decision-making, and prior learning, as well as decisions relating to future trust. Each of these domains can be considered not only a consequence of causally upstream trust evaluations, but contributing factors to future social contexts where norms of trust will either be reenforced or potentially undermined.

General Decision-Making

The limited evidence suggests that increased trustworthiness will lead to increased risk in decision-making. It has been found that when it is expected that a source of information (i.e. someone you intend to interact with) aims to harm you, then people show a stronger inclination not to respond with a decision, but when they do, they are more accurate in their decisions (Posten & Mussweiler, 2013). Further, participants are more willing to take advice after recounting an experience of distrust but are more inclined to discount the advice of others after recounting an experience of trust (Schul & Peri, 2015). In a medical scenario, people accept medical decisions more if they perceive their physicians as more trustworthy (Fortune et al., 2016).

Financial Decision-Making

Given that the majority of trustworthiness literature is based on economic trust games, there is extensive research on links between trust and financial decision-making. For instance, individuals who are perceived as less trustworthy are less likely to have their personal loans funded (Duarte et al., 2012). On the other hand, people seem to be more accepting of unfair financial distributions in games, such as the ultimatum game, when the face of a distributor seems more trustworthy (Wu et al., 2018). A few studies suggest that perceiving someone as trustworthy increases risk-taking tolerance in financial trust games (Qi et al., 2022). Similarly, people are more willing to book highly-priced Airbnb accommodation if hosts are perceived as more trustworthy from photos (Ert et al., 2016). Overall, people invest more money when their partners are viewed as more trustworthy and cooperate less with untrustworthy partners.

Temporal Decision-Making

Studies have examined willingness to delay rewards or discount potential future rewards as a function of trust (Michaelson et al., 2013). For instance, a study found that a proposer’s perceived facial untrustworthiness is associated with a higher temporal discounting rate (i.e. the perceived reduction in the value of a reward when it is delayed), indicating a lower preference toward reward postponing (Anzani et al., 2022). This effect appears to be robust and also occurs in children as young as 3 to 5 years old (Ma et al., 2018).

Legal Decision-Making

Mounting evidence suggests that perceived trustworthiness is relevant in decision-making situations where it should not be a consequential factor. An example of this is judicial decision-making. Specifically, people or faces (e.g. defendants) that are perceived as untrustworthy are often viewed as guilty, irrespective of the quality of evidence presented (Jaeger et al., 2020; Wilson & Rule, 2015; Yang et al., 2019). In another study, participants required less evidence to arrive at a guilty verdict and were more confident in this decision for an untrustworthy appearing defendant than a trustworthy defendant (Porter et al., 2010). Initial trustworthiness judgements are also associated with higher honesty ratings, even in high-stake real-life legal situations (Baker et al., 2016). Overall, facially more attractive people (which are judged as trustworthy) are found to be less guilty and culpable than their unattractive counterparts (Mazzella & Feingold, 1994).

Future Directions for Research on the Influence of Trust on Decision-Making

The relationship between trust and decision-making has been extensively studied in financial contexts, but there is a need for more research on how trust impacts decisions in other areas, such as health maintenance. For example, specific research is needed to separate the face-perception components of wearing a face mask from the social connotations of group membership and ideological commitments (particularly in the context of public health requirements during the COVID-19 pandemic), as well as in the context of evaluating highly regarded medical professional such as doctors, nurses, and dentists. Masks are likely also associated with potential sickness, and this association could be expected to depend on the differential normalisation of casual mask-wearing in different countries. More broadly, the effects of trust violations on decision-making accuracy and the willingness to take risks are not well-documented outside of controlled experimental settings. The existing evidence suggests that people might be more tolerant of cheating and be willing to take higher risks for a facially trustworthy person, another under-explored interaction between risk assessment and the facial cues of trustworthiness. It is also unclear the extent to which these kinds of manipulations would show effects outside of financial trust games, as the abstract nature of these games and gambling scenarios intentionally avoid some of the social cues that help contextualise other trust judgements.

There are also some specific applied domains of legal decision-making that are in need of further examination, particularly if said studies can incorporate he aforementioned dimensions of social identity cues and memorial biases. For example, police officers often look to establish trust in forensic interviews by building rapport with individuals. However, empirical evidence suggests this may decrease memory performance (Mazzella & Feingold, 1994), and may be differentially effective based on the presence or absence of social identity overlaps with the interviewee. Estimating when this effect occurs and to what extent it can be prevented would be a valuable focus for future studies.

Looking Back and Moving Forward: Gaps and Future Directions

The current review has provided a detailed analysis of key perceptual, social, and cognitive indicators of trust, revealing important insights on which domains of emotion, social context, perception, and memory directly inform judgements of trustworthiness. There are, however, significant gaps in the literature within each of these areas, which represent not only opportunities to advance the understanding of individual effects, but also for interdisciplinary research that can enhance understanding of trust (e.g. the discussed definitional issues identified in social identity research are partially addressed by the behavioural operationalisations discussed from Decision Science game-based research). Part of the authors’ intentions in reviewing and critiquing these disparate fields in a single paper is to illustrate how the variables of interest in one field are often missing confounding influences in the methodological assumptions of another field.

To that end, an overarching concern that presents an additional potential barrier to meaningful integration between fields are variations in the quality and specificity of methods of measuring trust. While some areas prioritise demonstrable behavioural commitments to vulnerable action as their core measures of trust (notably, game theoretic approaches like those in Bellucci, 2022; Wu et al., 2018), others rely on self-report scales that focus on specific applied domains of trust judgements (e.g. disposition to social trust, Acedo-Carmona & Gomila, 2014), or multi-dimensional constructs where overall trust relies on more narrow perceptions of ability, responsivity, or value-alignment (as in Patent, 2022). Both behavioural commitment and scale construct measurements have uncertain connections to the most widely used measurements outside of the directly trust-focused literature, which are single-item self-report evaluations that vary widely in their predictive convergence with other measurement methods (see Castro et al., 2023). More comprehensive theoretical and psychometric work connecting disparate indicators and constructs could significantly improve the standardisation of measurement approaches, and render results between research areas more comparable and mutually accessible.

Overall, these identified gaps through targeted research will not only advance our theoretical knowledge but may also offer practical solutions for enhancing opportunities for trust across different domains. Continued interdisciplinary exploration of the cognitive indicators of trust is crucial for fostering cooperation, improving communication, and reducing potential conflicts in a wide range of consequential real-world settings.