Belief in a just world or belief in just others? a study on the object of belief in a just world

Abstract

Background

While belief in a just world has been extensively studied, the object of this belief still requires further specification. This study distinguishes two sources of uncertainty in future returns—nature and other people—and investigates whether belief in a just world is specifically directed toward human-sourced uncertainty but not nature-sourced uncertainty.

Methods

To test this hypothesis, an experiment was conducted in which participants decided whether to make an investment based on their beliefs about the return they would receive from another player. The identity of the other player was varied (computer or human, representing nature-sourced uncertainty and human-sourced uncertainty, respectively), and participants’ level of belief in a just world was manipulated using a priming method. Multiple statistical analyses were conducted to examine the differences in investment behavior and expected returns between computer and human conditions.

Results

When interacting with human players, participants in the just-world priming condition showed significantly higher investment rates and expected returns compared to those in the unjust-world priming condition. In contrast, when interacting with computer players, the differences between just-world and unjust-world conditions were non-significant, both for investment rates and expected returns. Mediation analysis further revealed that expected return mediated the relationship between priming condition and investment behavior in the human player condition.

Conclusion

These findings demonstrate that belief in a just world influences decision-making when interacting with human players but not with computer players, supporting our hypothesis that belief in a just world mainly targets human-sourced uncertainty but not nature-sourced uncertainty. This research advances our theoretical understanding of belief in a just world and contributes to our understanding of its functions for both individuals and society.

Introduction

In real life, investments are usually made in the present, whereas returns often occur in the future. This temporal gap creates uncertainty about whether one’s investments will yield deserved returns. When faced with this uncertainty, people’s investment decisions largely depend on their expectation of returns. Belief in a just world (BJW)—the belief that individuals live in a just world where they will get what they deserve—plays a crucial role in addressing this uncertainty and shaping expectations of returns [1, 2]. When people have a strong BJW, they believe that their efforts will be appropriately rewarded, making them more willing to invest in the future [3]. In contrast, when people lack a BJW, they do not expect fair returns for their efforts and, as a result, are discouraged from making future investments.

According to Lerner and Miller (1978), BJW stems primarily from people’s basic psychological needs [1]. Specifically, people need to perceive the world as controllable and stable, and BJW serves to satisfy this need. Additionally, BJW has been linked to a number of psychological benefits. For example, it is associated with higher life satisfaction [4–6], greater optimism [6, 7], more positive affect [8], higher self-esteem [9], less depression [6, 10, 11], lower anxiety [6, 12], and reduced stress [6].

Despite these psychological benefits, BJW also needs to be adaptive at the behavioral level to ensure long-term survival from an evolutionary perspective. However, research suggests that BJW is not an accurate reflection of reality but rather a “positive illusion” [13–15]. Such a cognitive bias could potentially lead individuals to overestimate future returns, resulting in inefficient investments and losses [16, 17]. This raises a paradox: why does BJW, as a cognitive bias potentially leading to inefficient decisions, survive natural selection? Drawing on this, we propose that a clearer understanding is needed of the environments in which BJW is adaptive and likely to survive, and those in which it is not.

Given that BJW is a belief related to future uncertainties, examining the sources of these uncertainties offers a promising approach to understanding its adaptiveness. Based on these considerations, we distinguish between two sources of uncertainty in future payoffs—nature and other people.1 These two types of uncertainty differ in terms of the factors that primarily determine the outcomes of the uncertainty. Nature-sourced uncertainty refers to uncertainty where outcomes are determined by factors following fixed probability distributions that cannot be influenced by human actions.2 For example, the outcome of a coin toss is an example of nature-sourced uncertainty: the odds of heads or tails are determined by the physical properties of the coin. Similarly, a computer-generated random number is also a type of nature-sourced uncertainty, as it is determined by the algorithm. A lottery draw conducted using a mechanical machine is another example: the result depends on natural factors such as the physical properties of the balls. In contrast, for human-sourced uncertainty, the outcomes are primarily determined by other people. For instance, in a principal-agent relationship, a principal may delegate a task to an agent, but the outcome, which is uncertain for the principal, largely depends on the agent’s time, effort, and actions. This paper aims to investigate:

RQ1: What type of uncertainty does BJW target?

RQ2: Through what mechanism does BJW influence behavior under targeted uncertainty?

We propose that the difference in the primary determinants of uncertainty outcomes has significant implications for the adaptiveness of BJW in these two types of uncertainty. In situations of nature-sourced uncertainty, it is obvious that the most adaptive belief for individual is the one that aligns with the objective likelihood of getting deserved return: if an individual holds an upward-biased belief about the likelihood of receiving a deserved return, he or she will make inefficient investments, leading to losses; conversely, if an individual holds a downward-biased belief about the likelihood of receiving a deserved return, he or she will forgo efficient investment opportunities, missing out on potential gains. As BJW tends to overestimate the likelihood of receiving the deserved return [13–15], it is expected to lead inefficient investments and to be less adaptive in the context of nature-sourced uncertainty.

However, BJW can be adaptive in situations of human-sourced uncertainty. Unlike nature-sourced uncertainty, the outcomes in social interactions can be shaped by one’s belief and subsequent behavior through a self-fulfilling prophecy mechanism. Specifically, human-sourced uncertainty differs from nature-sourced uncertainty in that the distribution of possible outcomes can be influenced by one’s own actions, as one’s actions can alter the behaviors of others. For example, social exchange theory suggests that interpersonal interactions are guided by reciprocal exchanges: individuals tend to respond to kindness with positive outcomes and unkindness with negative ones [20]. Thus, when an individual holds a strong belief that they will receive deserved returns from others, he or she is more likely to contribute more; as the individual contributes more, others may respond with greater rewards. This creates a self-fulfilling prophecy, where BJW actually helps create the just outcomes it assumes. This makes BJW adaptive from an individual perspective. Furthermore, BJW can also contribute to an increase in collective welfare. Since an individual’s contribution typically positively impacts social outcomes, BJW can increase collective welfare by motivating people to contribute more to society [21–23]. Thus, holding a BJW can also improve the adaptiveness of the group to which one belongs. In summary, from both individual and group perspectives, BJW is adaptive in situations of human-sourced uncertainty.

From an evolutionary perspective, the adaptive value of BJW in human-sourced uncertainty and its maladaptive nature in nature-sourced uncertainty suggests that BJW has evolved to be specifically directed towards human- but not nature-sourced uncertainty. Therefore, we hypothesize:

H1: BJW mainly targets human-sourced uncertainty but not nature-sourced uncertainty.

To test this hypothesis, we conducted an experiment in which subjects decided whether to make an investment, which depended on their beliefs about the return they would receive from another player. We varied the identity of the player so that it could be either a computer (representing nature-sourced uncertainty) or a human (representing human-sourced uncertainty). Additionally, we manipulated the subjects’ level of BJW. Our hypothesis indicates that BJW plays a role in the case of human-sourced uncertainty but not in the case of nature-sourced uncertainty. This further predicts that the manipulation of BJW would affect subjects’ investment behavior when the other player was a human, but would not affect subjects’ investment behavior when the other player was a computer. By testing these predictions, we can evaluate our hypothesis.3

Furthermore, we propose that the effect of BJW on investment behavior under human-sourced uncertainty is mediated through expected returns. As discussed above, the temporal gap between investment and returns makes investment decisions heavily dependent on expected returns. BJW, by definition, shapes people’s expectations about receiving deserved outcomes. These expectations about returns then guide investment decisions. Building on this analysis, we hypothesize:

H2: Expected return mediates the effect of BJW on investment behavior under human-sourced uncertainty.

Methods

Participants

A prior power analysis using G*Power 3.1 indicated that a sample size of 197 would ensure 80% power for the chi-square test, assuming α = 0.05 and a small to medium effect size (Cramer’s V = 0.2) [24]. In total, 223 participants were recruited from a university in China. Participants completed the experiment by accessing the experimental webpage from their mobile phones. The experimental webpage was generated through a professional survey platform called “Wenjuanxing”, which provides functionality similar to “Qualtrics”. Each participant took part in the experiment only once. These participants were randomly assigned to one of two conditions: the just-world priming condition or the unjust-world priming condition. 19 participants were excluded because they did not correctly answer the control questions. Consequently, there were 102 participants in the just-world priming condition and 102 participants in the unjust-world priming condition. The participants had a mean age of 21.27 years (SD = 2.17), and 125 (61.27%) of them were female. Additionally, 57.35% of the participants were from urban areas, and the average family income level of the participants fell within the range of 4,001 to 8,000 Chinese yuan per month per capita. This study was conducted with institutional IRB approval (SDU-CER-2024010; 28 March 2024). All participants in this study gave their informed consent.

Procedure and measurement

The participants were first randomly assigned to either the just-world priming condition or the unjust-world priming condition using a between-subject design. In the just-world priming condition, participants were asked to recall and write about an experience in which their efforts or contributions were rewarded. In the unjust-world priming condition, participants recalled and wrote about an experience where their efforts or contributions were not rewarded. This recall and writing task aimed to manipulate participants’ level of BJW. Although BJW is generally considered stable, previous studies have shown that it can be temporarily manipulated using this method [25, 26].

After the recall task, we assessed the participants’ BJW to evaluate the effectiveness of the manipulation. The participants’ BJW was assessed via Dalbert’s (1999) Personal Belief in a Just World Scale [15]. This scale consists of seven questions, for example, “I think that important decisions that are made concerning me are usually just”. The participants responded on a 6-point Likert scale, where 1 represented “strongly disagree” and 6 represented “strongly agree”. Each participant’s level of BJW was calculated by averaging the scores on the seven scale items (Cronbach’s α = 0.84).

Additionally, we measured participants’ emotional state to control for potential affective responses triggered by the recall task that might influence subsequent behavioral decisions. Emotional state was measured using a well-established nonverbal scale developed by Bradley and Lang (1994), in which participants chose one of nine manikins representing their current emotion, ranging from very negative (coded as 1) to very positive (coded as 9) [27].

Participants then played the investment game, also known as the trust game, which was first developed by Berg et al. (1995) and has since become a widely used and frequently replicated measure of behavioral trust [28]. In our version of this game, there were two roles, Player A and Player B, interacting in a single round. The participants, as Player A, were initially given 20 experimental currency units (ECUs) (5 ECUs = 1 Chinese yuan). They then decided whether to invest this amount in Player B. If they chose not to invest, they would keep the 20 ECUs, and their final payoff would be 20 ECUs. If they chose to invest, the 20 ECUs would triple to 60 ECUs and be given to Player B, who would then decide how much of the 60 ECUs to return to Player A. In this case, their final payoff would depend on Player B’s decision, specifically the amount returned by Player B.4

The participants were informed that Player B could be either a computer or a human (another participant). These two types of Player B were designed to represent nature-sourced and human-sourced uncertainty, respectively. When Player B was a computer, it would make choices according to a fixed probability distribution that could not be influenced by Player A’s actions—a key characteristic of nature-sourced uncertainty. Specifically, the computer would randomly select a return amount between 0 and 60 ECUs. In contrast, when Player B was human, the return decision would be made by another participant, who could potentially respond to Player A’s investment decision based on social considerations such as reciprocity and fairness—a fundamental feature of human-sourced uncertainty. The participants did not know the identity of Player B in advance and were required to make decisions for both cases. After deciding whether to invest in either case, the participants were asked to report their expectations about how much Player B would return if they were to choose to invest. After this, the participants answered two control questions to test their understanding of the rules of the investment game.

Notably, participants were told that the identity of Player B would be randomly determined and that their payment would be based on their choice in the realized case. This eliminated potential wealth effects (earnings from one task affecting decisions in the subsequent task) and ensured that the participants made honest choices for both cases. In addition, the order of these two cases was randomized to ensure that any difference in the effects of BJW between the two scenarios was not due to order effects.

After the game, we measured participants’ risk preference to control for its potential influence on investment behavior. Following the approach of Dohmen et al. (2011), participants rated their general willingness to take risks on a scale from 1 (completely unwilling) to 10 (completely willing) [29]. This method has been shown to provide an effective measure of risk preferences. Additionally, we collected demographic information including gender, age, family location, and family income level. A balance test using logistic regression showed that none of these characteristics significantly predicted experimental condition, indicating that subjects in the two conditions were balanced on these characteristics (LR χ²(5) = 2.43, p = 0.787). Figure 1 provides an overview of the experimental procedure described above.

Fig. 1.

Procedure flow chart

Data analysis

Our analyses proceeded in several steps to test H1 and H2. First, we conducted a manipulation check using an independent samples t-test to verify whether our priming manipulation successfully affected participants’ BJW levels. Second, to test our main hypothesis, we examined whether the manipulation of BJW had different effects in computer versus human scenarios. For investment behavior, we used chi-squared tests to compare the investment proportions between just-world and unjust-world priming conditions in each scenario. We then conducted logistic regression analyses to verify these results while controlling for emotional state, risk preference, and demographic variables. For expected return, we first conducted independent samples t-tests to compare expectations between just-world and unjust-world priming conditions in each scenario. We then performed a 2 (priming condition: just-world vs. unjust-world) × 2 (partner type: computer vs. human) ANOVA to examine the interaction effect between priming condition and partner type. Finally, we conducted ordinary least squares (OLS) regression analyses with the same control variables. Based on our hypothesis, we expected that the manipulation of BJW would affect both investment behavior and expected returns when Player B was human, but not when Player B was computer. Finally, to test whether expected returns mediated the relationship between BJW and investment behavior in the human scenario, we conducted mediation analyses using Hayes’s PROCESS macro (Model 4) with 5000 bootstrap samples. For all analyses, statistical significance was set at p < 0.05.

Results

Manipulation check

We conducted an independent samples t-test to test the effectiveness of the manipulation. The results showed that the level of BJW was significantly higher in the just-world priming condition (M = 4.46, SD = 0.84) compared to the unjust-world priming condition (M = 4.01, SD = 0.71; t = 4.06, p < 0.001; Cohen’s d = 0.56), indicating successful manipulation of BJW.

Investment behavior

We tested H1 by examining whether manipulating BJW affected participants’ behavior and expectations when Player B was a human, but not when Player B was a computer. We first examined whether manipulating BJW affected participants’ investment behavior. A chi-squared test showed that the manipulation of BJW did not affect participants’ investment behavior when Player B was a computer. The proportion of investments in the just-world priming condition was 36.27%, and that in the unjust-world priming condition was 35.29%, with no significant difference between them (χ²(1) = 0.02, p = 0.884, Cramer’s V < 0.01).

However, when Player B was a human, the results showed that manipulating BJW had a significant effect on participants’ investment behavior. The proportion of investments in the just-world priming condition was 69.61%, and that in the unjust-world priming condition was 48.04%. A chi-squared test showed that the proportion of investments in the just-world priming condition was significantly higher than that in the unjust-world priming condition (χ²(1) = 9.79, p = 0.002; Cramer’s V = 0.21). Figure 2 illustrates these differences in investment proportions between conditions for both computer and human scenarios visually.

Fig. 2.

Results of investment decision. Error bars indicate standard errors of the means. * p < 0.05, ** p < 0.01, *** p < 0.001

Notably, the absence of significant effect in the computer context indicates that BJW is indeed ineffective in affecting investment in this context rather than due to insufficient statistical power, as evidenced by our power analysis showing the sample size was sufficient to detect even small to medium effect sizes and the significant result obtained with the same sample size in the human context.

To further test these results while controlling for potential confounding variables, we conducted logistic regression analyses for the computer and human scenarios, respectively (see Table 1). The dependent variable was investment behavior (1 = invest, 0 = not invest), and the key independent variable was the priming condition (1 = just-world priming condition, 0 = unjust-world priming condition), while controlling for emotional state, risk preference, gender, age, family location, and family income level. As shown in Table 1, after controlling for these variables, the manipulation of BJW still had a significant effect (β = 0.91, p = 0.003) on participants’ investment behavior when Player B was a human, but showed no significant effect when Player B was a computer (β = 0.01, p = 0.951). These regression analyses confirmed the results of the above non-parametric tests.

Table 1.

Regression analyses of the effects of the priming condition on investment behavior in the computer and human scenarios

	Investment
	Computer scenario β (SE)	Human scenario β (SE)
Priming condition	0.01 (0.30)	0.91 (0.30)**
Emotional state	0.24 (0.11)*	0.28 (0.11)*
Risk preference	0.01 (0.07)	0.07 (0.08)
Gender	0.79 (0.33)*	0.19 (0.33)
Age	0.01 (0.07)	0.12 (0.07)
Family location	−0.15 (0.34)	−0.31 (0.35)
Family income level	−0.02 (0.14)	0.38 (0.15)*
Constant	−2.57 (1.87)	−7.60 (2.05)***
Pseudo R2	0.03	0.10
LR χ 2	LR(7) = 10.46	LR(7) = 28.49***
N	204	204

Note: The regression analyses used the logistic model. Investment (1 = invest, 0 = not invest). Priming condition (1 = just-world priming condition, 0 = unjust-world priming condition). Gender (1 = female, 0 = male). Age was measured in years. Family location (1 = urban area, 0 = rural area). Family income level was measured on a 5-point scale ranging from 1 (below 2000 Chinese yuan a month per capita) to 5 (more than 8000 Chinese yuan a month per capita). * p < 0.05, ** p < 0.01, *** p < 0.001

Expected return

We then examined whether manipulating BJW affected participants’ expectations about how much Player B would return if they were to choose to invest. The results again showed that it depended on whether Player B was a computer or a human. When Player B was a computer, there was no significant difference between the participants’ expectations in the just-world (M = 27.25, SD = 12.34) and unjust-world priming conditions (M = 27.30, SD = 11.72; t = 0.03, p = 0.976; Cohen’s d < 0.01). Conversely, when Player B was human, participants’ expectations in the just-world priming condition (M = 28.42, SD = 12.57) were significantly higher than those in the unjust-world priming condition (M = 22.99, SD = 12.09; t = 3.14, p = 0.002; Cohen’s d = 0.44). A 2 (priming condition: just-world vs. unjust-world) × 2 (partner type: computer vs. human) mixed ANOVA showed a significant interaction effect between priming condition and partner type, F(1, 202) = 9.21, p = 0.003, η_p² = 0.04. This result confirms that the effect of just-world beliefs on expected returns depends on whether participants were interacting with a human or computer partner. Figure 3 presents these results visually. Similar to the investment behavior result, the absence of significant effect in the computer context indicates BJW’s genuine ineffectiveness in affecting expected return rather than due to insufficient statistical power.

Fig. 3.

Results of expected return. Error bars indicate standard errors of the means. * p < 0.05, ** p < 0.01, *** p < 0.001

Moreover, regression analyses in Table 2 confirmed the results of the above non-parametric tests. In these OLS regression analyses, the dependent variable was expected return, and the key independent variable was the priming condition (1 = just-world priming condition, 0 = unjust-world priming condition). As shown in Table 2, after controlling for other variables, the manipulation of BJW still had a significant effect (β = 5.19, p = 0.003) on participants’ expectations when Player B was a human, but not when Player B was a computer (β = 0.19, p = 0.911). These results further supported our hypothesis by demonstrating that BJW specifically influences expectations in human-sourced uncertainty but not in nature-sourced uncertainty.

Table 2.

Regression analyses of the effects of the priming condition on the amount of expected return in the computer and human scenarios

	Expected return
	Computer scenario β (SE)	Human scenario β (SE)
Priming condition	0.19 (1.70)	5.19 (1.74)**
Emotional state	−1.05 (0.63)	1.43 (0.64)*
Risk preference	0.27 (0.44)	−0.01 (0.45)
Gender	−1.62 (1.81)	0.41 (1.84)
Age	−0.21 (0.41)	−0.03 (0.41)
Family location	−0.67 (1.94)	−1.08 (1.99)
Family income level	0.01 (0.81)	0.58 (0.82)
Constant	37.26 (10.47)***	4.03 (10.69)
R 2	0.02	0.07
F	F(7, 196) = 10.46	F(7, 196) = 2.23*
N	204	204

Note: The regression analyses used the OLS model. Expected return was participants’ expectations about the amount returned by Player B. Other variables were the same as those in Table 1. * p < 0.05, ** p < 0.01, *** p < 0.001

The mediating role of expected return

We further tested whether expectation played a mediating role in the effect of BJW on investment behavior in the human scenario (H2). As reported in Sect. 3.2 and 3.3, priming condition positively predicted investment behavior (β = 0.91, p = 0.003) and expected return (β = 5.19, p = 0.003). When we included expected return in the regression model (see Table 3), priming condition still significantly predicted investment behavior (β = 0.69, p = 0.034), and expected returns also significantly predicted investment behavior (β = 0.05, p < 0.001). These results indicated that expected returns partially mediated the relationship between the priming condition and investment behavior.

Table 3.

Mediating effect of the expected return on the relationship between the priming condition and investment in the human scenario

	Investment β (SE)
Priming condition	0.69 (0.32)*
Emotional state	0.21 (0.12)
Risk preference	0.08 (0.08)
Gender	0.14 (0.35)
Age	0.14 (0.07)
Family location	−0.34 (0.37)
Family income level	0.41 (0.16)*
Expected return	0.05 (0.01)***
Pseudo R2	0.17
LR	LR (8)   =   47.20***
N	204

Note: The regression analysis used the logistic model. * p < 0.05, ** p < 0.01, *** p < 0.001

Using Model 4 in Hayes’s PROCESS (v4.2) with SPSS 27.0, we calculated the mediating effect of expectation on the relationship between the priming condition and investment behavior (see Fig. 4). Bootstrap analysis (5000 samples) yielded a 95% confidence interval of [0.07, 0.71] for the indirect effect 0.30, which did not include 0, confirming the mediating role of expectation.5

Fig. 4.

Mediating effect of the expected return on the relationship between the priming condition and investment in the human scenario. * p < 0.05; ** p < 0.01, *** p < 0.001

Discussion

The present study investigated whether BJW mainly targets human-sourced uncertainty but not nature-sourced uncertainty by examining its differential effects on behavior and expectations under human versus nature-sourced uncertainty. From an evolutionary perspective, we proposed that BJW should be adaptive in human-sourced uncertainty through social self-fulfilling prophecy mechanisms but maladaptive in nature-sourced uncertainty where outcomes follow fixed probability distributions. Supporting our hypothesis, our results showed that the manipulation of BJW had effects on both investment behavior and expectations when Player B was a human, but not when Player B was a computer. Furthermore, the expected return played a mediating role in the effect of BJW on investment behavior when Player B was a human.

Theoretical and practical implications

Our study contributes to the literature by addressing a fundamental question about BJW: why does this seemingly inaccurate belief survive natural selection despite potentially leading to inefficient decisions? We propose that this paradox can be resolved by specifying the objects toward which BJW is directed. Our findings suggest that BJW may have evolved to influence behavior under human-sourced uncertainty, where its effects can be adaptive through social self-fulfilling prophecy mechanisms, while not extending to nature-sourced uncertainty where such mechanisms are absent.

Our study underscores the importance of specifying the concept of BJW. We demonstrate that a more nuanced understanding of this concept can be gained by distinguishing between nature-sourced uncertainty and human-sourced uncertainty. In this sense, our study can be seen as part of the research agenda to clarify the BJW concept. As BJW can be understood as people believing they will be treated fairly by the world, this concept can be further elucidated from at least three perspectives: What does “they” mean? What does “fairly” mean? And what does “world” mean? In addition to our study, which addressed the third question, several other studies have addressed the first two questions and demonstrated the theoretical importance of addressing these questions.

With respect to the first question, previous studies have distinguished the beneficiaries of justice. Studies find that people tend to view the world as being more just to themselves than to others [4, 30, 31]. Furthermore, while beliefs in a just world for oneself and for others are correlated, there are also important differences between them [32]. For example, it is BJW for oneself but not for others that is associated with higher life satisfaction [4], higher self-esteem [33], lower depression [10], and greater purpose in life [34]. Conversely, it is BJW for others but not for oneself that is associated with negative attitudes toward disadvantaged groups [4]. Moreover, the “others” can also be further distinguished, as people hold different beliefs in a just world toward different types of others. For example, Sutton et al. (2008) reported that people viewed the world as being more just for men than for women [31].

Other studies have distinguished between distributive justice and procedural justice to address the second question. Distributive justice concerns people’s perceptions of the fairness of outcomes and resource distributions [35], and procedural justice focuses on people’s perceptions of the fairness in the rules and decision-making processes that lead to these outcomes [36]. Like the self-other distinction, the distributive-procedural justice distinction also provides a framework for refining our understanding of the relationship between justice beliefs and well-being. For example, previous studies have shown that procedural justice, relative to distributive justice, may be especially associated with improved health behaviors [37] and increased receptivity to preventive health messages [38].

Our study addresses the third question by distinguishing the sources of uncertainty. To our knowledge, Stroebe et al. (2015) is the only other study that also distinguished between the sources of uncertainty [39]. Their aim was to construct a new scale of BJW that incorporated multiple dimensions of sources of uncertainty. They reported that the levels of BJW in different dimensions of sources of uncertainty were associated with different types of behavior. Although our study and Stroebe et al. (2015) both involve the distinction of sources of uncertainty, our study differs significantly from theirs. First, in terms of research objectives, we aimed to explore what type of uncertainty BJW mainly targets, while Stroebe et al. (2015) focused on constructing a new scale of BJW [39]. Given these different objectives, our study and Stroebe et al. (2015) also employed distinctly different research designs and methodologies. Specifically, we directly manipulated subjects’ BJW and their exposure to different sources of uncertainty by varying the identity of the other player in investment game (human or computer) to represent different sources of uncertainty, and compared the differential effects of BJW on behavior across these sources to identify BJW’s target. In contrast, Stroebe et al. (2015) adopted a conventional scale construction approach, developing and validating items to assess individuals’ BJW levels across different sources of uncertainty through principal component analysis, and subsequently examining the relationships between these dimensions and behavioral responses to significant life events and societal attitudes [39]. Finally, different from Stroebe et al. (2015), we innovatively provided an evolutionary perspective to explain BJW’s differential responses to different sources of uncertainty. Thus, our study advances beyond measurement to provide theoretical insights into when and why BJW influences human behavior. Taken together, the aforementioned studies and our research highlight the need to enhance our understanding of BJW by further specifying the elements within this concept. Our study contributes to this goal by specifying the object of BJW.

Meanwhile, this study provides evidence on the function of BJW in motivating investment in the long-term. While it is widely proposed that BJW has a function in encouraging investment in the long-term, the direct evidence has been limited. Laurin et al. (2011) manipulated subjects’ BJW and provided evidence on the impact of BJW on investing in career pursuits [40]. However, they focused primarily on the willingness to invest in hypothetical scenarios (without real incentives) rather than actual investment behavior (with real incentives). Our study provides causal evidence on the function of BJW in motivating long-term investment and demonstrates that its effect extends beyond the willingness to invest to actual investment behavior. Moreover, our study further suggests that the function of BJW in motivating long-term investment may exist only when people perceive the future uncertainty as human-sourced, but not when they perceive it as nature-sourced.

Based on these findings, our research has important practical implications for the design of modern interactions. In today’s digital age, there is an increasing trend toward replacing human-to-human interactions with human-machine interactions, and human interactions are increasingly mediated by automated systems. However, our findings suggest that this trend may have unintended consequences. BJW’s unique role in promoting reciprocity and trust specifically in human interactions indicates that direct human engagement has distinctive value that may not be readily replicated by automated systems. The human element in social exchanges appears to activate psychological mechanisms that foster reciprocity and cooperation, potentially leading to higher social welfare. Therefore, while technological advancement is inevitable and beneficial in many ways, maintaining opportunities for meaningful human interaction may be crucial for maximizing social welfare. This suggests that rather than completely automating all interactions, a balanced approach that preserves human elements, such as in situations involving trust and reciprocity, might be more beneficial for society.

Limitations and future research

This study also has certain limitations that warrant further investigation. First, while we make a dichotomy regarding the source of uncertainties, there may be situations where the source is ambiguous, such as in interactions with AI and algorithmic decision-making systems. People’s perceptions of these anthropomorphic but non-human entities are more complex and may challenge the simple dichotomy between human and natural sources of uncertainty. On the one hand, people often unconsciously apply human social heuristics when interacting with AI systems, treating them as social entities with human-like characteristics [41], such as applying gender stereotypes to robots and responding to emotional expressions in healthcare chatbots [42, 43]. On the other hand, since AI systems fundamentally differ from humans in their lack of true autonomy, emotions, and consciousness [44], people often employ distinct “machine heuristics” when dealing with AI, attributing less agency and responsibility to algorithmic decisions compared to human decisions [45]. Given these conflicting perspectives, how to classify anthropomorphic but non-human entities requires empirical investigation, and future research could examine how BJW influences behavior in these cases where the boundary between human and natural sources becomes blurred. Beyond this classification challenge with anthropomorphic entities, real-world uncertainties are often mixed, with outcomes determined by both nature and other people. Future research could explore how people’s behavior is affected by their BJW under such mixed uncertainties. Furthermore, there may be other ways to classify uncertainties beyond the dimension of the uncertainty source. Future research could explore the impact of BJW under alternative classifications.

Second, several limitations regarding the generalizability of our findings should be noted. Prior research has shown that priming effects tend to decay rapidly over time [46, 47]. While our results verified the validity of the priming effect through the recall task, the relatively short interval between the BJW manipulation and the gaming task limits our understanding of its sustained effect. This temporal limitation suggests that our BJW manipulation might have limited applicability in long-term or complex real-world situations where the effects need to persist over extended periods. Future research could employ alternative methodological approaches to investigate BJW’s role in naturalistic settings, where multiple contextual factors interact dynamically over longer timeframes. Additionally, our investment game paradigm focused primarily on monetary decisions in a simplified experimental setting, whereas real-world situations where BJW might influence behavior are often more complex and diverse. For example, people make decisions about investing time in relationships, effort in education, or resources in community projects. Future research could test whether BJW’s specific influence on human-sourced uncertainty extends beyond monetary contexts.

Third, all participants in our study were students from a Chinese university, characterized by relatively young age and homogeneous socioeconomic backgrounds. Given that BJW formation is closely tied to socialization processes, this single sample may limit the generalizability of our findings. We anticipate potential heterogeneity in BJW’s differential effects across uncertainty sources among different cultural and age groups, based on two key considerations. First, according to our theoretical analysis, BJW’s adaptiveness in human-sourced uncertainty relies on people’s tendency toward spontaneous reciprocity (gift exchange). As spontaneous reciprocity represents an intuitive response shaped by cultural environment and requires social learning throughout life development [48, 49], we expect the differential effects of BJW between nature-sourced and human-sourced uncertainty to be more pronounced among highly socialized individuals (e.g., adults) and in cultures with stronger cooperative norms. Conversely, this difference might be minimal or non-existent among individuals with lower levels of socialization (e.g., children) and in cultures with weaker cooperative traditions. Furthermore, in some cultures where natural events are attributed to supernatural agents rather than viewed as objective occurrences, the distinction between nature-sourced and human-sourced uncertainty becomes blurred. In contrast to our findings, for individuals from these cultures, BJW might influence behavioral responses to both human-sourced and nature-sourced uncertainty. Future research should investigate these possibilities through cross-cultural and cross-age studies to better understand how BJW operates across different sociocultural contexts.

Finally, while we focus on a sequential interaction scenario, as captured by the investment game in our experiment, simultaneous interaction scenarios are also common in real life. Simultaneous interaction scenarios differ substantially from sequential interaction scenarios in that others do not know one’s behavior when making decisions, so one’s contribution cannot increase others’ contributions through reciprocal mechanisms. On the one hand, this feature may make believing in a just world less adaptive for the individual in simultaneous interaction scenarios because it may increase one’s contribution without receiving reciprocity from others. However, on the other hand, from a group selection perspective, it is adaptive because it can increase the group’s total welfare. Future research could employ simultaneous-action games to examine whether BJW exists in such scenarios and to explore potential new insights.

Conclusions

This study distinguishes between two potential objects of BJW and demonstrates that this belief mainly targets human-sourced uncertainty but not nature-sourced uncertainty. This research advances our theoretical understanding of the concept of BJW and contributes to our understanding of its functions for both individuals and society.

Abbreviations

BJW

Belief in a just world

ECUs

Experimental currency units

OLS

Ordinary least squares

Author contributions

Conceptualization, X.Z. and Y.Z.; methodology, X.Z. and Y.Z.; software, X.Z. and Y.Z.; validation, X.Z. and Y.Z.; formal analysis, X.Z. and Y.Z.; investigation, X.Z. and Y.Z.; resources, X.Z. and Y.Z.; data curation, X.Z.; writing—original draft preparation, X.Z.; writing—review and editing, X.Z. and Y.Z.; visualization, Y.Z.; supervision, Y.Z.; project administration, Y.Z.; funding acquisition, Y.Z. All authors have read and agreed to the published version of the manuscript.

Funding

Not applicable.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Shandong University (SDU-CER-2024010, 28 March 2024). Informed consent was obtained from all subjects involved in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.