ABSTRACT Time preference plays a central role in decision‐making, shaping how individuals value outcomes that occur at different points in time. While most research on time preference focuses on one‐off payoffs, many real‐world decisions generate recurring consequences that unfold continuously over time. Despite their prevalence, little is known about how recurring payoff structures influence intertemporal preferences. We propose that recurring payoffs reduce discounting by promoting temporal integration. By highlighting the continuity of future consequences, recurring payoffs encourage individuals to form a more temporally integrated representation of outcomes rather than evaluating consequences as isolated events. Drawing on the choice isolation argument and the literature on choice bundling and choice bracketing, we argue that this representation reduces the salience of immediate‐versus‐delayed trade‐offs and ultimately lowers discount rates. We test this hypothesis in two experiments involving 284 individuals. Participants completed intertemporal choice tasks in either the gain or loss domain, with choices varying along two additional dimensions: payoff structure (one‐off vs. recurring) and outcome magnitude (small vs. large). Results revealed significantly lower discounting in recurring payoff conditions than in comparable one‐off conditions. The findings are consistent with the proposition that recurring payoff structures encourage a more temporally integrated evaluation of future outcomes, leading individuals to place less weight on immediacy when making intertemporal choices. This study extends the time‐preference literature by introducing recurring payoff structures as a distinct class of intertemporal outcomes and identifying temporal integration as a potential mechanism through which payoff structure influences discounting. It further demonstrates that the magnitude effect extends to recurring payoff environments, suggesting that outcome size continues to shape discounting even when consequences unfold repeatedly over time. These findings suggest that making the recurring nature of future consequences salient may be a promising avenue for promoting more future‐oriented decision‐making.
ABSTRACT Behavioral Measures of Risk Taking (BMRT), such as the Balloon Analogue Risk Task (BART) and the Angling Risk Task (ART), are widely used to assess risk‐taking propensity, yet most of the existing literature focuses on environments in which risk increases over time. The present study examines how behavior changes when risk instead decreases, and whether sensitivity to such changes depends on the availability of probabilistic information. Participants ( N = 272) completed three versions of the ART in which risk increased, remained constant, or decreased, under conditions where outcome probabilities were either known (“ Sunny ”) or unknown/ambiguous (“ Cloudy ”), followed by the BART. Censored mixed‐effects Bayesian regression models were used to analyze the data. Participants took more actions in decreasing‐risk environments than in constant‐ or increasing‐risk environments, particularly when the reduction in risk was explicitly known. Under ambiguity, however, participants were more cautious and showed little sensitivity to improvement in the environment. Correlations across tasks also depended on information. The ART and the BART showed stronger and more consistent associations in the Sunny condition than in the Cloudy condition. These findings suggest that performance on BMRT reflects sensitivity to the structure of the decision environment, particularly to improvements in risk and the availability of information about probabilities, and that BMRT summary scores should not be interpreted as direct measures of a single, domain‐general risk‐taking propensity or risk preference.
The date-delay effect, a well-established anomaly in intertemporal choice, describes the tendency for individuals to prefer future rewards more when they are described using calendar dates (e.g., "July 1, 2026") compared to delays (e.g., "100 days later"). Although prior research has primarily attributed this effect to reduced perceived temporal distance under the date frame, evidence for this mechanism comes almost exclusively from short delay distance. Through five experiments systematically manipulating delay distance (15, 105, and 450 days) and description frame, the present study suggests that the date-delay effect is robust across both short and long delays. Critically, however, the underlying cognitive mechanism shifts with delay distance. Under short delays (15 days), the effect is mediated by reduced perceived temporal distance, supporting the Temporal Distance Perception Hypothesis. Under long delays (450 days), where frame-induced differences in perceived temporal distance disappear, the effect is instead driven by increased perceived temporal ambiguity, validating the Temporal Ambiguity Perception Hypothesis. Study 5 further validated the robustness of the findings from Studies 3 and 4, demonstrating that the two mechanisms operate respectively across different delay distances. These findings support a context-dependent dual-process account aligned with the attribute-comparison model, revealing that the date-delay effect may be underpinned by distinct temporal perception mechanisms operating at different temporal distances. This refined theoretical framework advances our understanding of how temporal framing shapes intertemporal decisions.
Adolescent risk-taking generally occurs in ambiguous situations where decisions must rely on experiential learning. Although adolescence is characterized by heightened exploration in such situations, this does not systematically lead to optimal risk adjustment based on feedback learning. This study examines initial exploration and risk adjustment through feedback in ambiguous situations. The Risk Learning Card Task (RLCT), distinguishing three levels of risk (low, medium, and high probability of loss), was used to assess risk-taking in ambiguous situations by adolescents (N = 285, aged 10-17 years, M = 13.39, SD = 1.63, 45.96% female) and young adults (N = 162, aged 18-25 years, M = 19.10, SD = 1.29, 70.99% female). Generalized additive mixed models (GAMM), considering age as a continuous variable, revealed that (1) the youngest adolescents showed the highest level of exploration, (2) a majority of participants exhibited a correct explicit risk level estimation regardless of age (i.e., they were able to estimate risk levels at the end of the task), (3) the younger the participants were, the less risk they took and the more points they earned in the low-risk condition; the more risk they took, the fewer points they earned in the high-risk situation. Thus, adolescents' adjustment is more optimal in low-risk situations and less optimal in high-risk situations.
Cognitive overload can impair professional skepticism in high-stakes contexts such as auditing. In these settings, sustaining professional skepticism is essential. Default nudges, or pre-selected options, may offset these effects by reducing cognitive demands. We conducted an online pilot study followed by the main experiment to examine how cognitive load and default nudges influence professional skepticism in auditing decisions. A pilot study validated a dot memory task manipulation of cognitive load and identified low and high load conditions for the main experiment. The main experiment embedded this manipulation in the Phillips' audit task, used to measure professional skepticism in auditing. Results showed that cognitive load only slowed responses. However, default nudges accelerated responding and improved accuracy under load, but only when aligned with the most probable response; misaligned nudges reduced accuracy. These findings suggest that defaults act as conditional scaffolds under cognitive strain, supporting judgment and decision-making in some contexts but introducing risks in others.
As robots become increasingly integrated into daily life, researchers have begun to examine how they shape social influence and conformity in human-robot hybrid groups. However, prior work has focused almost exclusively on decision outcomes-the answers group members provide-while overlooking how decision time may also influence conformity. Across three studies, we investigate how the response time of human and robot group members shapes individuals' conformity. We find that faster group decisions increase conformity, whereas slower decisions reduce it. Importantly, fast responses exert a stronger influence when they come from human rather than robot group members, an effect driven by differences in perceived confidence. These findings underscore the critical role of response time and responder identity (human or robot) in shaping conformity, expanding our understanding of social influence in human-robot interactions, and opening new avenues for future research.
Organizations are increasingly using powerful AI aids to support decision-making. Yet, performance improvements are difficult to predict, in part because employees vary in how they use them. We propose that such behavioral variation may stem from differences in how employees appraise these aids. Applying an opportunity-threat framework to AI augmentation (positing two primary cognitive AI appraisals-as either an opportunity or a threat to employees), we examine the effect of appraisal type on professionals' decision-making when using powerful AI aids. Specifically, we examine the independent effort they invest in deliberate, vigilant information-seeking during their initial decisions and their behavioral compliance with AI recommendations that diverged from their initial judgments during final decision-making. Four simulation-based experiments, in which we manipulated appraisal type and used the Wizard of Oz method to simulate recommendations from powerful AI aids, yielded consistent findings: Study 1, conducted among employees participating in a weight estimation simulation, revealed that opportunity (vs. threat) appraisals reduced participants' individual effort in making their initial decisions and increased their compliance with fictitious AI recommendations that contradicted their initial judgments in their final decisions. Studies 2-4, conducted among HR professionals and medical students engaged in realistic tasks within their work domains, further revealed that these effects of opportunity appraisals on decision-making were driven by an increased preference for using the powerful AI aid rather than making decisions alone (an effect evident among those with higher domain experience; Studies 3-4). Our findings provide important implications for organizational decision-making in hybrid human-AI environments.
Decisions about how much we help others have been quantified recently as general social discounting functions or as individual-specific welfare tradeoff ratios (WTRs). The WTR model integrates kinship (based on kin-selected altruism), which should be relatively inflexible as a contributor to overall WTR value, whereas more general social discounting functions minimize kinship as a distinct factor. Using monetary forced-choice measures of social discounting, this research compared valuations of close kin versus friends and distant kin versus friends when that individual exhibits extreme disrespect of the participant. Participants (143 undergraduates in Experiment 1; 210 in Experiment 2) significantly devalued distant versus close relationships and post- versus pre-disrespect scenarios. In Experiment 2, there was also a significantly lower valuation of friends relative to kin. The WTR prediction of an interaction between kin/friends and pre-/post-disrespect, however, did not emerge. This indicates instead a more general social discounting process. There are possible adjustments to the present methodology, which further research may explore, or to the WTR model, which might account for the present results.
Targeting behaviors that conflict with deeply held values remains a persistent challenge for behavioral science, particularly as polarization increasingly defines both political and everyday decision contexts. Prior research suggests that nudges (interventions that simplify decision-making) often struggle in value-incongruent domains. The potential of sludges (interventions that introduce friction) remains less well understood, especially when such interventions are implemented transparently and disclosed to decision-makers. Across two experiments spanning multiple behavioral domains, we examine whether decisiveness, a dispositional tendency toward rapid and confident decision-making, moderates the effectiveness of transparent nudges and sludges in value-congruent versus value-incongruent contexts. We find a systematic divergence. Among highly decisive individuals, transparent sludges increase behavioral intentions more than nudges when interventions target value-incongruent behaviors, but not when they target value-congruent behaviors. Among less decisive individuals, nudges and sludges are similarly effective across contexts. These findings indicate that transparent friction can facilitate engagement with value-incongruent options among individuals most inclined to reach closure quickly, without steering choice or obscuring influence. By identifying decisiveness as a moderator of intervention effectiveness in polarized settings, this research contributes to growing evidence of heterogeneity in behavioral interventions and highlights the importance of aligning choice architecture tools with both individual decision styles and value alignment. From a policy perspective, the results suggest that transparent, nondirectional frictions may offer a viable and ethical design option in contested domains where autonomy concerns are salient.
ABSTRACT Products and services are commonly rated on either a 5‐point scale or a binary scale. Using a pilot experiment, two between‐subjects experiments, and a survey, this paper investigates how individuals evaluate products differently depending on the scale used, how they categorize ratings on a 5‐point scale into binary bins, and how they estimate 5‐point distributions from binary scales. Individuals perceive products as higher in quality when ratings are presented on a binary scale if reviews have been assigned to positive or negative categories based on whether they are above or below the midpoint of a 5‐point scale. Individuals perceive products as being of equivalent quality across scales only when ratings of four and five are taken as positive, and the remainder as negative. However, when individuals are asked to generate a 5‐point distribution from binary ratings, they do not account for this skew unless the 5‐point scale is explicitly labeled so that the “neutral” point of the scale is defined as four rather than three. These findings extend the literature that aims to understand how people evaluate products given different forms of ratings information.
Cases in which a costly helping act by a benefactor can greatly improve the outcomes of a beneficiary abound. Such help conforms with prevalent social norms and can increase collective welfare. We study how the surplus created through such help is shared by the benefactor and the beneficiary. We hypothesized that aspects of the situation, such as relative wealth, social distance, whether the help increased gains or prevented losses, and perspective may result in variations in this division. In two experiments, we compare participants' judgments (N = 695) and participants' bargaining choices (N = 617 dyads). In both tasks, the proposed compensation for the cost incurred, while the benefits of helping exceeded the cost and was affected similarly by the manipulated factors. We observed pro-social equalizing of outcomes, less indebtedness to family benefactors than to commercial ones, and sensitivity to the fact that help-seeker was suffering a loss. However, whereas the judgments allocated to the beneficiary about 80% of the surplus generated, in (consequential) minimal bargaining, the surplus was split more equally, as expected by economic theory. This gap raises doubts about the validity of using nonconsequential measures.
This research examines how evaluability is influenced by the interplay of attribute framing, magnitude information, and reference points. In two experiments, participants evaluated scenarios varying in framing valence (positive vs. negative), magnitude polarity, and the availability of a base-rate reference point. The results indicated that framing valence systematically biased evaluations regardless of reference-point availability and magnitude polarity. However, the availability of a reference point improved calibration, revealing higher sensitivity of evaluations to magnitude information. Experiment 2 extended these findings by manipulating reference-point value. The results demonstrated that evaluations are highly sensitive to the relative value of magnitudes compared with the reference point, yet absolute magnitude made an additional contribution. These findings are consistent with fuzzy-trace theory, insofar as they align with its dual-route framework, and may be interpreted as compatible with the idea that evaluability is associated with dual cognitive routes: gist-based processes may contribute to susceptibility to framing bias, whereas verbatim-based processes may support magnitude sensitivity and improve calibration. Importantly, the present research did not directly test fuzzy-trace theory against alternative accounts; therefore, it cannot determine whether fuzzy-trace theory uniquely explains the observed findings or whether other accounts could explain them. The research highlights the intricate nature of evaluability by demonstrating that providing reference-point information can enhance magnitude sensitivity without mitigating framing bias. Practical implications for information presentation are discussed, and future directions are outlined for promoting theoretical understanding of evaluability mechanisms.
Responses to COVID-19 in the United States were split along partisan political lines throughout the pandemic. People on the political left tended to take the medical threat more seriously and were more likely to adopt preventive health behaviors than those on the political right, resulting in clear differences in state-level policies and health consequences. Here, we examine the origins of these differences, investigating whether they are based on interindividual differences or in party dynamics. Two groups of participants (), who had voted either for Donald Trump or Hillary Clinton in 2016, played the Transmission Game. Players made repeated decisions in a simulated pandemic that involved a trade-off between increasing their expected personal payoff and reducing the chance of a total payoff-loss for themselves and others. In four experimental conditions, the study framing (neutral or pandemic) was crossed with the presence (or absence) of a normative intervention aimed at reducing risk-taking. We observed systematic partisan differences in all conditions, with Republicans taking more risk than Democrats, even in neutrally framed conditions-supporting the idea of interindividual differences between voter groups beyond party dynamics. At the same time, both normative interventions and the pandemic framing reduced risk-taking and expected infection rates in both voter groups. Moreover, we examined individual predictors of risk-taking and demonstrated that game behavior, conservatism, and psychological reactance predict intentions to adopt preventive health behaviors outside the laboratory. We discuss implications for the framing of studies conducted during ongoing crises and lessons for future pandemic preparedness.
We conducted three experiments to examine how organizations' responses to sexual harassment influence individuals' investment decisions, and whether moral values moderate these effects. In Study 1, the organization either addressed or ignored a harassment case; in Study 2, it either retaliated against (or not) the reporting employee; and in Study 3, the perpetrator was either the CEO or a regular employee. In all studies, we found robust interactions between organizational culture and moral values-particularly fairness and harm. Individuals high in fairness were less willing to invest when the organization failed to address harassment (Study 1) or retaliated against the victim (Study 2), regardless of perpetrator role (Study 3). Individuals high in harm were less willing to invest when harassment was ignored (Studies 1 and 3). These patterns also extended to punitive behavior: stronger moral values increased participants' willingness to harm the organization. Overall, individuals with stronger fairness and harm values were more sensitive to how organizations respond to harassment, and this sensitivity shaped both their financial decisions and punitive intentions.
Eliciting the weights of attributes is a key step in multi-attribute decision-making methods. The weights usually represent the relative importance of the attributes or the tradeoffs among them in forming a decision. Various weight elicitation methods exist, each based on different assumptions and procedures. Still, many of these methods do not explicitly account for the potential influence of cognitive biases in their design. This study examines the anchoring bias, a well-known cognitive bias, in the weight elicitation step (the Tradeoff procedure) of multi-attribute value theory (MAVT). We developed the following three hypotheses: (i) Using the most important (best) attribute to construct the indifference pairs in the Tradeoff procedure leads to higher weights for the best and worst attributes and lower weights for the other attributes, (ii) using the least important (worst) attribute to construct the indifference pairs in the Tradeoff procedure leads to lower weights for the best and worst attributes and higher weights for the other attributes, and (iii) using both best and worst attributes to construct the indifference pairs (i.e., the best-worst tradeoff: BWT) mitigates the anchoring bias. To test the hypotheses, we conducted an experiment by designing a questionnaire based on MAVT and collected data from 336 participants for a decision problem. The findings indicate that the anchoring bias has a significant impact on the Tradeoff procedure and that the BWT is effective in mitigating this bias.
Consequentialist theories of judgment and choice hold that individuals and actions should be evaluated in terms of the outcomes they produce, but not on how they bring about (otherwise equivalent) outcomes. This paper demonstrates a striking violation of consequentialism in judgment when fatal martyrdom-sacrificing one's life for a cause-is introduced. Across six experiments (N-total = 4861), including one preregistered replication, US participants judged scenarios in which a protagonist takes actions to save members of his group from an attack. They evaluated the protagonist and his actions more positively when he (voluntarily) sacrificed his life in the process, compared with when he achieved the same goal without dying. This is despite the fact that the former scenario-which adds self-sacrifice to an otherwise identical chain of events-is clearly worse for the protagonist (and his fellow group members). Moreover, fatal martyrdom (self-sacrifice) boosted evaluations even when the protagonist belonged to a despised group and his actions produced harmful outcomes that served an aversive cause. These results show that people praise fatal martyrdom (self-sacrifice), regardless of its consequences, and regardless of whether they generally support or oppose the martyr and the martyr's cause. The experiments also examined several potential mechanisms and boundary conditions of this fatal martyrdom effect, and they show that the effect can occur even in the absence of human intergroup conflict.
Delay discounting refers to a decrease in subjective value of an outcome as its receipt is delayed. In discounting studies, properties of the delay are important, but understudied, methodological details. One aspect of the delay is opportunity costs, which refers to alternative reinforcement that is forgone while waiting for a reward. Research shows that verbal narratives specifying high opportunity costs lead to steeper discounting for money and cigarettes. However, opportunity costs have not been examined with food discounting. The current study examined the extent to which opportunity costs affect discounting for money (Experiment 1) and food (Experiment 2) using within-subjects designs. Participants were presented with narratives specifying differing magnitudes of opportunity costs prior to completing four discounting tasks for two magnitudes of money (Exp 1) or food (Exp 2). Results indicated that as opportunity costs increased, discounting for both money and food also increased. Magnitude effects were also found; the smallest magnitudes of money and food were discounted most steeply. Lastly, systematic responding was lower, but more consistent across opportunity costs for food compared to money, indicating choice stability with money is more sensitive to opportunity costs. These results then replicate and extend opportunity costs to food discounting.
For long-term hazards, the likelihood of a hazardous event can be communicated in the format of the probability of occurrence per unit time (probability-based, e.g., a flood with a 1% chance of occurring annually) or the average time of risk occurrence (time-based, e.g., a 100-year flood). This study investigates how the two presentation formats influence risk perception and related behavior in long-term hazard scenarios, such as typhoons, healthcare, high-risk occupations, and investments, and further examines the moderating role of probability magnitude. First, we demonstrated that at an extremely low probability level, time-based formats reduced perceived risk and protective tendency compared to probability-based formats (Studies 1 and 2). However, as probability increased to a moderately low level, this effect diminished and then reversed, such that time-based formats increased risk perception and protective tendency compared to probability-based formats (Study 2). The difference between the two formats diminished again when probabilities reached the moderate-to-high range and moved toward 50% (Studies 1 and 2). Second, we revealed that presentation formats also influenced perceived risk differences between options (Study 3) and altered individuals' risk preference between options (Studies 3, 4a, and 4b), with these effects also moderated by probability magnitude. For lower probability events, time-based formats amplified perceived risk differences between options, promoting preferences for options with lower loss-probabilities, whereas for higher probability events, this effect weakened and eventually reversed. The findings contribute to related theory by identifying probability magnitude as a key boundary condition under which time-based and probability-based formats have different influences on risk perception and behavior. Practically, the results offer implications for risk communication in long-term hazard contexts, indicating how to choose between time-based and probability-based formats according to the size of the probability and the intended persuasive purpose.
Expertise is multifaceted. This project builds on the difference between subjective expertise (the confidence one holds in understanding a particular topic) and objective expertise (one's actual understanding of that topic), which can vary independently. We examine how subjective and objective expertise differentially relate to making better choices and seeking additional choice-relevant information. We hypothesized that some participants would be poorly calibrated in their ability to assess the value of more difficult-to-evaluate nonalignable attributes (in our setting, categorical attributes that are present or absent, such as torque-vectoring) compared with easier-to-evaluate alignable attributes (e.g., miles per gallon). We measure and manipulate subjective and objective expertise and find that participants with greater objective expertise (either naturally or through provided information) tend to base their choices on the values of nonalignable attributes and that this pattern is generally not predicted by measures of subjective expertise. High subjective expertise, instead, predicts a reduced willingness to seek additional information or take advice, as well as a greater willingness to pay for customized items, independent of objective expertise. These studies point to important differences between these facets of expertise, underscoring the importance of incorporating measures of each when examining the role of expertise in decision-making.
Decision makers employ either a selection strategy to choose desirable options or an elimination strategy to remove undesirable options. While previous research has revealed how these two strategies influence decision processes and outcomes, the interpersonal influences remain understudied. This research fills this gap by documenting that people are less willing to interact with decision makers who use an elimination (vs. selection) strategy due to the belief that elimination (vs. selection) strategy users are more critical. However, the interpersonal cost of using an elimination strategy is mitigated when negative decisions are made to identify the worst option, when negative feedback is desired, and when decision makers are forced to use the strategy. Our research contributes to the literature on decision strategy by revealing the social effect of selection and elimination, and to the literature on choice perception by showing that decision strategy shapes the way people view decision makers.
