Individuals at higher risk of death are expected to prioritize near-term benefits over delayed benefits (Wilson & Daly, 1997). This hypothesis, which has empirical support at the between-population level (Pepper & Nettle, 2017; Sng et al., 2018), has a basis in evolutionary logic (Del Giudice et al., 2015): With a long expected time-horizon for future reproduction, it can be adaptive to forego immediate fitness-linked benefits (e.g., mating opportunities, material resource consumption) in order to invest in components of embodied capital (e.g., growth, skills, resource accumulation) that will increase lifetime fitness via positive effects on, e.g., phenotypic quality, fertility, mating success, and social status (Kaplan, Hill, Lancaster, & Hurtado, 2000). With a shorter expected time-horizon for future reproduction, however, it can be most adaptive to prioritize immediate fitness-linked benefits, given that investments in future reproduction are relatively unlikely to pay off. The link between mortality risk and time horizon also plays a key role in some behavioral research that does not invoke evolutionary biology. For example, evidence from economics suggests that part of the reason why poor countries tend to remain poor is that their shorter life expectancies disincentivize investment in physical and human capital (Lorentzen et al., 2008).

Via applications of life history theory to explain individual differences within species, it has been hypothesized that present- vs. future-orientation is a component of an integrated continuum of “life history strategy”. Life history theory (Del Giudice et al., 2015; Kaplan et al., 2000) seeks to describe and explain energy allocation tradeoffs among growth, somatic maintenance, and reproduction. A proposed axis of life history variation contrasts “slower” strategies (prioritizing future over present reproduction, quality over quantity of offspring, and parental effort over mating effort), with “faster” strategies that prioritize their opposites (Del Giudice, 2020; Jeschke et al., 2008). As with present-orientation specifically, faster strategies are hypothesized to be adaptive responses to higher mortality risk. The hypothesis of an integrated fast-slow life history continuum has recently been cast into doubt on theoretical and empirical grounds (e.g., Dinh & Gangestad, 2024; Frankenhuis & Nettle, 2020; Richardson et al., 2017; Sear, 2020; Zietsch & Sidari, 2020). However, the hypothesis that traits related to present vs future-orientation will be calibrated to mortality risk stands regardless of whether an integrated fast-slow continuum of life history strategy exists.

Operationalizing within-population individual differences in mortality risk (as distinct from between-group, e.g. between-neighborhood, mortality risk) has proven to be a major methodological challenge. In laboratory settings, researchers have primed perceived mortality risk, and tested hypotheses about the interaction between it and childhood SES in influencing risk preference, temporal discounting, and preferred reproductive timing (Griskevicius, Delton, et al., 2011a; Griskevicius, Tybur, et al., 2011b). However, some of the reported effects have failed to replicate (Pepper et al., 2017). Other researchers have used telomere attrition as a biomarker of accelerated aging and hence a plausible proxy for individual mortality risk (Epel et al., 2004; Schlomer, 2024). However, the relationships between exposures to environmental stress (hypothesized cues of exogenous mortality risk) and telomere attrition are weak (Pepper et al., 2018), requiring large samples to generate adequate statistical power. And perhaps the most commonly used operational definitions of mortality risk are self-ratings of overall health or perceived mortality risk (e.g., Chua et al., 2017), which obviously have dubious accuracy. The development of a valid operational definition of individual mortality risk would therefore help advance research in this area.

Here we introduce and explore a novel measure of individual mortality risk: the risk ratings assigned by insurance companies to applicants for individual life insurance policies. Life insurance companies determine the price of insurance for a given applicant through the process of underwriting, wherein the applicant (i) provides personal information about, e.g., their medical history, family medical history, occupation, and smoker status, and (ii) is subjected to a paramedical exam that includes anthropometric measurements (e.g., BMI), cardiovascular indicators (e.g., blood pressure), and a blood sample (Klein, 2013). This information is then used by actuaries to assign the applicant to a mortality risk category via algorithms based on vast amounts of data regarding the predictors of all-cause mortality (e.g. Boodhun & Jayabalan, 2018; Klein, 2013). The mortality risk ratings assigned by insurance actuaries can be provisionally assumed to be actuarially sound, because of the inherent incentive for companies to price insurance accurately in order to avoid paying too many claims, which is supplemented by the market discipline imposed by competition among insurance companies (Eling, 2012; Eling & Schmit, 2012). We reasoned that insurance-based mortality risk estimates may have a number of advantages over other operational definitions. First, insurance-based risk estimates are based exclusively on information that is known to be predictive of all-cause mortality. Second, insurance-based risk estimates integrate diverse sources of predictive information, much of which policyholders themselves may be unaware of (e.g., biomarkers assayed from blood; interacting risk factors from family medical history). Third, insurance-based risk estimates indicate how likely someone is to die relative to others in the entire population; this should avoid the reference class biases present in, e.g., self-assessed mortality risk estimates, which are likely based on a comparison of oneself with a non-representative subset of the population. In sum, there are multiple reasons to believe that insurance-based risk estimates may be actuarially superior to extant measures of individual mortality risk.

In the two studies reported here, we test associations among insurance-based mortality risk, self-estimated mortality risk (Dunkel & Decker, 2010), and a set of personality traits, some of them directly related to the present vs. future benefits tradeoff (e.g. conscientiousness and concern for future consequences), and others that are hypothesized (Belsky et al., 1991; Patch & Figueredo, 2017; but see Dinh & Gangestad, 2024; Richardson et al., 2017) to be indicators of psychometric life history strategy (e.g. aggression and sociosexuality).

Two non-mutually exclusive causal pathways have been proposed to explain how early life stress (e.g., neighborhood violence) causes, as an adaptive response, the adoption of a “faster” behavioral life history strategy (i.e., risk-proneness, impulsivity, present orientation, and short-term mating orientation). In one type of model, childhood stress has, during human evolution, reliably predicted later-life environmental harshness and thus higher rates of extrinsic mortality, favoring adoption of present-oriented (or “faster”) strategies based on exposure to external cues of harshness (Belsky et al., 1991; Ellis et al., 2009; Sng et al., 2018). In the other type of model, childhood stress damages the individual's phenotype, producing poorer health (and therefore higher mortality risk) throughout life, thereby favoring adoption of present-oriented (or “faster”) strategies based on a sampling of internal somatic cues of mortality risk (Nettle et al., 2013). The latter models generate the prediction that relationships between early life stress and faster indicators of present-versus future-orientation and life history strategy will be mediated by self-perceived or objectively measured health, a prediction for which some evidence has been found (Chua et al., 2017).

Both our studies recruited U.S. residents via the internet. We discovered that the risk rating classification systems used by different insurance companies are extremely similar to each other, permitting the unproblematic pooling of data for analysis. Study 1 measured personality using two self-report instruments: the Life History Rating Form (LHRF) (Dunkel et al., 2016), and four items from the Sociosexual Orientation Inventory (SOI) (Jackson & Kirkpatrick, 2007). We predicted that higher insurance-based risk ratings (i.e., higher estimated mortality risk) would be associated with “faster” scores on the LHRF, and with less interest in long-term mating relationships and greater interest in short-term mating relationships. Study 2 built on Study 1 by using a larger number of construct-specific self-report measures of focal traits, in addition to measures of recalled childhood environmental harshness. Also, Study 2 had greater statistical power than Study 1, included participants spanning a wider range of life insurance risk ratings, and verified participants' identities and insurance policy information via face-to-face interviews. Our predictions for Study 2 are pre-registered at https://osf.io/ynm9b. We predicted that higher insurance-based risk ratings would be associated with (1) higher self-estimated mortality risk; (2) higher levels of risk propensity, impulsiveness, aggression, mating effort, and interest in short-term mating relationships; and (3) lower levels of concern for future consequences, conscientiousness, and parental effort. We also predicted that three measures of recalled childhood environmental harshness, two pertaining to the family environment and the other to neighborhood violence and disorder, would be positively associated with all measures of mortality risk. Our analyses included age, sex, and smoker status as covariates. Insurance companies have separate risk rating tables for tobacco users and non-users, respectively, because smoking itself is a critical mortality risk factor (Klein, 2013). Because smoking is a form of behavior that has a causal impact on mortality risk, we always controlled for smoker status in analyses assessing associations between personality variables and mortality risk ratings across the entire sample. Additionally, in Study 2, where this was possible, we analyzed associations both across the entire sample and separately within tobacco users and non-users, respectively.