What Childhood Leaves Behind: Cognitive Ability and Trust in Adulthood

Abstract

This article challenges the idea that cognitive ability uniformly predicts prosocial traits. Using data from a large, nationally representative U.K. sample (N = 24,140), we test a moderated mediation model in which childhood disadvantage is associated with generalized trust both directly and indirectly via cognitive ability, while also moderating the association between cognitive ability and trust. We find that childhood disadvantage is associated with lower cognitive ability—measured across memory, verbal fluency, fluid reasoning, and numerical reasoning—and with lower generalized trust in adulthood. We also find that cognitive ability is positively associated with generalized trust; however, this relationship is significantly attenuated among those who experienced childhood disadvantage. These results persist after adjusting for current socioeconomic factors. The pattern whereby early-life disadvantaged environments are associated with differences in cognitive development and with constrained social returns to cognitive ability is likely to reinforce social immobility.

Keywords

generalized trust childhood disadvantage cognitive ability social mobility

For to every one who has will more be given, and he will have abundance; but from him who has not, even what he has will be taken away.

—Matthew 25:29, Revised Standard Version

Introduction

Generalized trust—general beliefs about the extent to which other people can be trusted—is essential for cooperation (Gächter et al., 2004), economic development (Knack & Keefer, 1997), income equality (Uslaner, 2002), lower rates of criminality (Sampson et al., 1997), and overall well-being (Helliwell et al., 2017). Given its wide-ranging societal implications, understanding what underpins generalized trust remains a central concern across multiple scholarly fields.

A large body of research identifies two powerful predictors of generalized trust: early-life environments and cognitive abilities. Empirical work has shown that childhood adversity and low parental socioeconomic status (SES) are associated with lower trust in adulthood (Kim et al., 2025; Mell et al., 2022; Stamos et al., 2019). From a life-history perspective, resource-rich settings encourage a “slow” strategy associated with long-term gains, whereas childhood stress, adversity, and resource scarcity function as a signal of environmental harshness, directing individuals toward “fast” strategies characterized by short time horizons, heightened vigilance, self-protective behavior, and heightened sensitivity to threat (Gladden et al., 2009; Griskevicius et al., 2011, 2013; Kraus et al., 2011; Simpson et al., 2012). Indeed, as generalized trust is a future-oriented strategy—requiring individuals to forgo short-term gains in favor of sustained cooperation (Curry et al., 2008) and to believe that others will not exploit them for short-term benefit (Righetti & Finkenauer, 2011)—it may be less adaptive in harsh environments (Petersen & Aarøe, 2015; Stamos et al., 2019). Lower trust among those who experienced early-life adversity may therefore represent an adaptive response to environmental uncertainty (Pepper & Nettle, 2017).

Childhood socioeconomic disadvantage is not only associated with trust-related life-history strategies but also with cognitive development and lower cognitive function in later life, primarily through stress and cortisol exposure, deficiencies in various micronutrients, and limited access to cognitively enriching experiences (Benton, 2008; Bradley & Corwyn, 2002; Britto et al., 2017; Dórea, 2021; Gellci et al., 2019; Hackman et al., 2010; Metcalfe et al., 2013; von Stumm, 2012). Cognitive ability, in turn, robustly predicts generalized trust and other prosocial behaviors (Corgnet et al., 2016; Hooghe et al., 2012; Rindermann, 2008). Individuals with higher cognitive ability are thought to better understand the long-term benefits of cooperation (Jones, 2008; Lohse, 2016; Millet & Dewitte, 2007; Proto et al., 2019; Segal & Hershberger, 1999), to be more likely to use prosocial behavior as a costly signal of status (Millet & Dewitte, 2007; Zahavi, 1977), to be more informed about opportunities for prosocial engagement (Wiepking & Maas, 2009), and to more accurately evaluate others’ trustworthiness—meaning that they tend to have relationships with people who are unlikely to betray their trust (Cosmides et al., 2010; Yamagishi, 2001). The two-system framework of decision-making, from the “heuristics and biases” literature (Kahneman & Frederick, 2005), further suggests that low trust is a universal property of System 1 (Corgnet et al., 2016), the intuitive, emotional, and affect-driven system that operates with sense of voluntary control (Tversky & Kahneman, 1996). Those low in cognitive ability are typically regarded as less able to override the emotional system and engage the deliberative, logical, and analytical system, System 2, which can override distrustful emotional responses (Stanovich & West, 2008).

Taken together, these strands of research suggest that childhood disadvantage may influence generalized trust both directly, through the adoption of life-history strategies, and indirectly, via its effects on cognitive development. While these parallel lines of research have largely overlooked the indirect pathway, these research strands also implicitly assume an independence model—whereby the effects of different variables on an outcome are separate and additive. However, there are strong theoretical reasons to expect that this indirect pathway is itself conditional on early-life context—that is, the association between cognitive ability and prosocial attitudes may depend on childhood socioeconomic conditions.

For instance, within the broader literature on socioeconomic inequality, debate persists regarding whether resources operate through multiplication or substitution processes (Andersson, 2016; Ross & Mirowsky, 2011). Resource multiplication—often referred to as the Matthew effect, whereby early-life resource advantages compound over time, leading to disproportionate rewards—suggests that advantaged environments amplify the returns to individual resources. In contrast, resource substitution implies that individual resources matter more in disadvantaged contexts, as they may compensate for a lack of environmental support.

As an example of resource multiplication, Damian et al. (2015) found that children raised in high-SES households benefited more from intelligence in terms of educational attainment and occupational prestige. Applied here, “slow” strategy advantaged early-life environments—characterized by low crime, effective institutions, and cooperative norms—may provide opportunities for cognitively skilled individuals to recognize and learn that generalized trust is a socially adaptive and often rewarded strategy.¹ In contrast, “fast” strategy harsh environments may constrain the translation of cognitive resources into trust, either by limiting exposure to cooperative interactions or by fostering emotional dysregulation, chronic stress, and negative affect (Evans & Kim, 2013; M. R. Gunnar, 2000; M. Gunnar & Quevedo, 2007; Kim et al., 2013; Potter et al., 2025)—which are associated with lower trust (Dunn & Schweitzer, 2005; Myers & Tingley, 2016) and greater reliance on low-trust, intuitive System 1 processing (Blanchette & Richards, 2010; Eysenck et al., 2007; Oldrati et al., 2016; Simonovic et al., 2017). Indeed, in a recent study, Reiter et al. (2023) found that irritability was higher among participants who had experienced childhood adversity, and this irritability explained their attenuated developmental gains in learning to trust during a multi-round trust game. Panel A of Figure 1 illustrates this resource multiplication hypothesis, showing how childhood disadvantage attenuates the positive association between cognitive ability and generalized trust.

Figure 1.

Hypotheses for the interaction between early-life disadvantage and cognitive ability on generalized trust: (A) multiplication, (B) adaptive calibration, and (C) substitution.

At the extreme, and consistent with Adaptive Calibration Models (Del Giudice et al., 2011; Del Giudice, 2014), adverse childhood environments may lead to “fast” life-history strategies via the calibration of cognitive systems toward heightened threat detection—resulting in more vigilant and protective strategies amongst individuals with higher cognitive ability (Belsky & Pluess, 2009; Frankenhuis & de Weerth, 2013; Frankenhuis et al., 2016). Panel B of Figure 1 illustrates the adaptive calibration hypothesis. Under this hypothesis, the association between cognitive ability and generalized trust is not merely attenuated in adverse environments; rather, cognitive resources are deployed to identify and enact the optimal strategy for a given ecology. Among individuals oriented toward “fast” life-history strategies, cognitive resources are adaptively directed toward vigilance and protection, whereas in stable, resource-rich environments, they are directed toward long-term cooperation.

In contrast, a resource substitution perspective suggests that for individuals from low-SES backgrounds—who often lack the structural supports (e.g., safe, cooperative, and trusting community environments) that typically foster generalized trust (Chetty et al., 2014)—cognitive ability may serve as a critical compensatory mechanism. In this view, high cognitive ability becomes more predictive of trust among the disadvantaged because it acts as a substitute for missing social resources. Specifically, for those “fast” strategy environments that lack environmental cues which encourage trust, higher cognitive abilities compensate by enabling individuals to better discern the strategic benefits of cooperation or to more precisely calibrate their assessments of others’ trustworthiness (Cosmides et al., 2010; Jones, 2008; Lohse, 2016; Millet & Dewitte, 2007; Proto et al., 2019; Segal & Hershberger, 1999; Yamagishi, 2001). Panel C of Figure 1 illustrates the resource substitution hypothesis, which could also reflect broader compensatory advantage mechanisms (Bernardi, 2014), wherein high-SES families offset their children’s lower cognitive ability through targeted investments (Ghirardi et al., 2024; Holm et al., 2019). For instance, high-SES parents may invest disproportionate time and financial resources in their cognitively disadvantaged children (Fan & Porter, 2020), fostering prosocial traits like generalized trust even in the absence of high cognitive ability.

In this article, we examine these competing hypotheses using a large sample of 24,140 respondents from a nationally representative longitudinal survey in the United Kingdom (U.K.). Specifically, we investigate whether the association between cognitive ability and generalized trust is moderated by childhood disadvantage, while also recognizing that childhood disadvantage may be associated with trust indirectly through its influence on cognitive development. Understanding how early-life environments structure both cognitive development and the expression of cognitive abilities as prosocial attitudes is important for a number of reasons. First, it challenges the implicit assumption that cognitive traits consistently foster prosocial attitudes (Proto et al., 2019). Second, it speaks to novel mechanisms in inequality research. For instance, if the prosocial expression of cognitive resources is suppressed or redirected under conditions of socioeconomic disadvantage, then it highlights the importance of childhood intervention programs on the formation of valuable noncognitive skills (Heckman & Rubinstein, 2001; Heckman et al., 2006). Third, it provides an opening for perspectives on the mechanisms through which “fast” and “slow” strategies lead to differential levels of trust, and more broadly, the mechanisms through which early-life disadvantage is associated with lower generalized trust. As hypothesized, cognitive abilities may be primed toward threat detection as an adaptive response to childhood and forecasted adult environmental instability (Del Giudice, 2014); alternatively, the longer-term effects of childhood adversity, such as chronic stress (Evans & Kim, 2013), may inhibit cognitive resources from overriding low-trust, intuitive System 1 processing (Simonovic et al., 2017); or it may be that disadvantaged environments do not produce the environmental cues necessary for cognitively skilled individuals to recognize and learn that generalized trust is a socially adaptive and often rewarded strategy. Lastly, it invites further research into whether other beneficial personality traits, attitudes, or behaviors show similar environment-dependent correlations with cognitive ability. Indeed, both cognitive ability and childhood SES are highly correlated with risk aversion and patience (Caner & Okten, 2010; Dohmen et al., 2010, 2018; Eckel et al., 2012; Falk et al., 2021; Shah et al., 2012; Shamosh & Gray, 2008; Sheehy-Skeffington, 2020), which also predict a variety of advantageous socioeconomic outcomes in adulthood (Åkerlund et al., 2016; Bonin et al., 2007; Cadena & Keys, 2015; Dohmen & Falk, 2011; DellaVigna & Paserman, 2005; Golsteyn et al., 2014).

Materials

Participants

We used data from Understanding Society (USoc) 2009 to 2024 (Waves 1–14). USoc is a nationally representative U.K. annual longitudinal survey of some 40,000 households. USoc covers a broad range of subjects including household finances, attitudes and opinions, economic activity, and cognitive tasks, among other things. The sample used for our empirical analysis is restricted to those who gave valid responses to the dependent, independent, and control variables used in the subsequent analyses. This yielded a final cross-section of 24,140 individuals. The mean age is approximately 47 years. Just over 43% of the sample is male and 24% report holding a university or college degree.

Measures

Generalized Trust

The dependent variable in all analyses is generalized trust, measured by the standard question: “In general, would you say that most people can be trusted, or that you can’t be too careful these days?” This item was administered exclusively in Wave 1 of USoc. We treat responses as categorical (nominal), where respondents selecting the “most people can be trusted” (37.28%) option are coded 2, those who impulsively responded “it depends” (23.38%) are coded as 1, and those selecting “you can’t be too careful” (39.35%) are coded 0. This classic measure—originally developed by Rosenberg (1956) as part of the faith-in-people Guttman scale—captures an individual’s general attitude toward others and belief in human nature. Although widely used, this single-item approach has been criticized for reducing a complex construct with distinct dimensions—such as cognition-based and affect-based trust—into a single-dimensional concept (McAllister, 1995; Robbins, 2022). However, the standard generalized trust question, unlike broader 11-point scales that often suffer from midpoint clustering (Uslaner, 2008) and are often subject to cultural interpretations of scale points (Bond & Lang, 2019), reduces the likelihood of misclassifying respondents and obscuring substantively meaningful differences between trusting and distrusting groups (Uslaner, 2008). Indeed, Dawson (2019) demonstrates that responses to the standard generalized trust question are highly consistent over time, suggesting that it effectively captures a stable, long-term psychological trait rather than measurement error or transient sentiment.

Cognitive Ability

In Wave 3 of USoc, five measures of cognitive function were collected which assessed memory, semantic verbal fluency, working memory, fluid reasoning, and practical numerical knowledge. Specifically, the five tasks include: (1) Word Recall—where, after being read a series of 10 words, participants were asked immediately afterwards and later in the interview to recall as many words as possible, in any order. Scores from the immediate and delayed word recall task were summed together to produce a single measure; (2) Verbal Fluency—where participants were given 1 min to name as many animals as possible. The final score on this item is based upon the number of unique correct responses; and (3) Subtraction Test—where participants were asked to give the correct answer to a series of subtraction questions. Starting at 100, participants were asked to subtract 7, then to subtract 7 again, and so on. There were a sequence of five subtractions and the number of correct responses out of a maximum of five was recorded; (4) Fluid Reasoning—where participants were asked to write down a number sequence, as read by the interviewer. The number series consists of several numbers with a blank number in the series. The participant was asked which number goes in the blank. Participants were given two sets of three number sequences, where the difficulty of the second set was determined by performance in the first set. The final score is based on the correct responses from the two sets of questions, accounting for the difficulty level of the second set of problems; (5) Numerical Reasoning—where participants were asked up to five questions that were graded in complexity. Based on performance on the first three items, participants could receive two additional (more difficult) questions or one additional (simpler) question. The types of questions asked included: “In a sale, a shop is selling all items at half price. Before the sale, a sofa costs £300. How much will it cost in the sale?” and “If the chance of getting a disease is 10 percent, how many people out of 1,000 (one thousand) would be expected to get the disease?” The final score is based on a simple count of the number of correct items.²

From these five cognitive function scores, we created a general cognitive ability factor (i.e., composite score). This composite measure standardizes and then combines respondents’ scores from each of the cognitive tasks. Internal consistency reliability across the five items was acceptable to good (α = .72; average interitem correlation of .34).

We also address a potential measurement concern arising from the timing of key variables. Generalized trust is measured in Wave 1, whereas cognitive ability is measured two annual waves later. As cognitive ability varies systematically with age—and prior research documents substantial, nonlinear life-cycle patterns (Whitley et al., 2016)—using the general cognitive ability factor could introduce age-related measurement error. Consistent with this literature, we observe pronounced nonlinearity in cognitive ability across the life course, with levels peaking in midlife and declining at older ages. To mitigate this concern, we construct an age-effect-free measure of cognitive ability. Specifically, we regress the general cognitive ability factor on a fourth-order polynomial in age. The resulting residuals are then standardized and used as our age-effect-free measure of cognitive ability in all subsequent analyses. This approach ensures that differences in cognitive ability are not mechanically driven by age and are therefore comparable across individuals despite the two-wave measurement gap.

Childhood Disadvantage

We follow the approach of Ronda et al. (2022) and consider childhood disadvantage across four dimensions. All four dimensions are based on respondents’ retrospective assessments of their childhood environment at age 14. Responses are recorded in Wave 1 and in later waves for new survey respondents who had never been interviewed before (excluding rising 16-year-olds). We therefore use all available responses regardless of the timing of data collection. The first dimension captures family instability, measured as a dichotomous variable equal to one if the respondent was not living with both biological parents, experienced living in a single-parent household, or experienced local authority care up to the age of 16, and zero otherwise (12.40% of the sample). The second dimension reflects human capital disadvantage, also measured as a dichotomous variable equal to one if the respondent reported that neither parent left school with qualifications or certificates, and zero otherwise (34.64% of the sample). The third and fourth dimensions pertain to family resources. The third dimension is equal to one if the respondent reported that neither parent was working at age 14, and zero otherwise (6.21% of the sample). The fourth dimension is equal to one if the respondent reported that neither parent was employed in a higher-level occupation at age 14—defined as any occupation above the routine or semi-routine level according to the Standard Occupational Classification 2010—and zero otherwise (25.58% of the sample). In the sample, 48.92% of respondents experienced no dimension of disadvantage, 31.58% experienced one dimension of disadvantage, 17.21% experienced two dimensions of disadvantage, and 2.29% experienced three dimensions of disadvantage.

To capture the multidimensional nature of childhood disadvantage, we construct a composite indicator rather than relying on any single measure, which would fail to reflect the breadth of disadvantaged experiences (Galobardes et al., 2006). We classify individuals as disadvantaged if they experienced two or more dimensions of disadvantage, a criterion that applies to 19.50% of the sample. While composite measures provide broader coverage of early-life conditions, they may be subject to interpretational confounding insofar as different components can influence outcomes through partially distinct pathways (e.g., exposure to stress versus material deprivation).

In addition to this conceptual consideration, there is also a measurement constraint: the four components of the childhood disadvantage index are not jointly observed for all respondents, as the availability of specific indicators depend to some extent on childhood circumstances. In particular, parental occupation is undefined when neither parent was working. Similarly, respondents who experienced family instability are systematically more likely to respond “don’t know” to other dimensions of disadvantage, such as parental employment, occupation, or education. Missing values therefore reflect a level of conditional observability rather than random nonresponse. In creating the composite indicator, missing values are treated as zero—82.43% of respondents provided valid responses to all four disadvantage dimensions, 15.63% had one missing response, while 1.94% had more than one missing response. The coding of the composite indicator is therefore conservative, as it increases the likelihood of false negatives—classifying some disadvantaged individuals as non-disadvantaged—and therefore attenuates estimated effect differences between advantaged and disadvantaged individuals toward zero. However, to account for missingness, in all analyses that follow, we apply the missing indicator method—where we add a dichotomous control variable equal to one if the respondent had one or more missing response, and zero otherwise. Moreover, we conduct a range of robustness checks using alternative operationalizations of childhood disadvantage.

Control Variables

We include a number of control variables that may explain the association between cognitive ability, childhood disadvantage, and generalized trust. These control variables, measured at Wave 1, are age (in linear and quadratic form), sex, educational attainment, housing tenure, marital status, square root of household size, logarithm of household income (which is adjusted by the Organization for Economic Co-operation and Development-modified equivalence scale and deflated by the Consumer Price Index), economic activity, and region of residence. These control variables were chosen as age, sex, resources, and education have been shown to be related to generalized trust (Bailey & Leon, 2019; Balliet et al., 2011; Brehm & Rahn, 1997; Charron & Rothstein, 2016; Li & Fung, 2013; Poulin & Haase, 2015; Sutter & Kocher, 2007; Uslaner, 2002; Van Den Akker et al., 2020). Supplemental Table S1 presents an overview of the sample characteristics by childhood disadvantage status.

Analytic Strategy

As childhood disadvantage may be associated with the development of cognitive ability and also moderate the association between cognitive ability and generalized trust, we adopt a moderated mediation framework. We first estimate the association between childhood disadvantage and cognitive ability using ordinary least squares (OLS). For individual i, we estimate:

{CognitiveAbility}_{i} = α + b_{1} {Disadv}_{i} + \sum_{j} b_{j} {Controls}_{j i} + ϵ_{i}

(1)

where ${CognitiveAbility}_{i}$ is the standardized age-effect-free general cognitive ability factor, ${Disadv}_{i}$ is a binary indicator of childhood disadvantage—equal to one if the individual experienced two or more dimensions of disadvantage, and zero otherwise—and $b_{1}$ captures the association between childhood disadvantage and cognitive ability. ${Controls}_{j i}$ denotes the vector of control variables described above, and $ϵ_{i}$ is an error term.

Next, we estimate the association between cognitive ability and generalized trust and test whether this association is moderated by childhood disadvantage. Generalized trust is measured as a three-category outcome with categories $k \in {0, 1, 2}$ corresponding to “can’t be too careful,” “it depends,” and “most people can be trusted.” We therefore estimate multinomial logistic regression models. First, we estimate a baseline model without interaction:

\begin{matrix} {Trust}_{i k}^{*} = c_{1 k} {CognitiveAbility}_{i} + c_{2 k} {Disadv}_{i} \\ + \sum_{j} c_{j k} {Controls}_{j i k} + ϵ_{i k} \end{matrix}

(2)

We then estimate an interaction model in which childhood disadvantage moderates the association between cognitive ability and generalized trust:

\begin{matrix} {Trust}_{i k}^{*} = d_{1 k} {CognitiveAbility}_{i} + d_{2 k} {Disadv}_{i} + \\ d_{3 k} ({CognitiveAbility}_{i} \times {Disadv}_{i}) + \\ \sum_{j} d_{j k} {Controls}_{j i k} + ϵ_{i k} \end{matrix}

(3)

where ${Trust}_{i k}^{*}$ is the latent propensity to select outcome k. In contrast to ordered logistic regression—which relies on the proportional odds (parallel regression) assumption and therefore constrains predictors to have identical effects across outcome categories—multinomial logistic regression allows predictors to have outcome-specific effects. We follow the recommended practice of presenting the results in terms of outcome-specific average marginal effects (AMEs; Norton & Dowd, 2018; Williams, 2012).

Finally, we assess conditional indirect effects—that is, the extent to which childhood disadvantage is associated with generalized trust through cognitive ability, allowing the association between cognitive ability and trust to vary by disadvantage status. As generalized trust is modelled using a nonlinear probability model, indirect effects are defined as the product of the estimated association between childhood disadvantage and cognitive ability ( $b_{1}$ from Equation 1) and the group-specific (i.e., disadvantaged or non-disadvantaged) AMEs of cognitive ability on generalized trust obtained from Equation 3. We additionally compute the difference between these two quantities—the Index of Moderated Mediation—which captures any group level heterogeneity in the association between childhood disadvantage and generalized trust via cognitive ability. Statistical inference for conditional indirect effects and their differences is obtained using bias-corrected bootstrap resampling.

In all estimations, we consider estimated associations across nested models. Specifically, Model 1 includes exogenous controls (age and sex). Model 2 adds sociodemographic controls and Model 3 further includes educational attainment, covariates that could in principle have been affected by childhood disadvantage and cognitive ability. Conditioning on these variables, we can better isolate the residual associations between generalized trust, childhood disadvantage, and cognitive ability (Cinelli et al., 2024).

Results

To begin our moderated mediation framework, we first estimate the association between childhood disadvantage and cognitive ability. Table 1 presents the main findings. As our parameters are estimated using ordinary least squares, we report OLS coefficients, which capture the change in standardized cognitive ability—our age-effect-free general cognitive ability factor—associated with a discrete change in childhood disadvantage from zero to one. We consider our estimates across nested models with increasing sets of control variables.

Table 1.

Association between childhood disadvantage and cognitive ability.

Model:	1.	2.	3.
Dependent variable:	Cognitive ability
Disadv.	−.348*** [−.381, −.315]	−.225*** [−.257, −.194]	−.113*** [−.144, −.083]
Observations	24,140	24,140	24,140
Exogenous controls	Yes	Yes	Yes
Sociodemographic controls	No	Yes	Yes
Educational attainment	No	No	Yes

Note. OLS coefficients; 95% confidence intervals using robust standard errors in brackets. Cognitive ability is our standardized age-effect-free general cognitive ability factor. Disadv. is a dichotomous variable that takes on the value of one for those who experienced two or more dimensions of childhood disadvantage, and zero otherwise. Exogenous controls include age (in linear and quadratic form) and sex. Sociodemographic controls include housing tenure, marital status, square root of household size, logarithm of household income (which is adjusted by the Organization for Economic Co-operation and Development-modified equivalence scale and deflated by the Consumer Price Index), economic activity, and region of residence. Educational attainment represents the highest level of attainment. All models include a missingness indicator for incomplete childhood disadvantage information. OLS = ordinary least square.

***

p < .001.

Model 1 of Table 1 reports the association between childhood disadvantage and cognitive ability, controlling only for exogenous characteristics (i.e., age and sex). Consistent with developmental perspectives, childhood disadvantage is associated with lower cognitive ability. Specifically, childhood disadvantage is associated with a .348 standard deviation lower level of cognitive ability. In Model 2, we include sociodemographic controls. Model 3 further adjusts for educational attainment—a key correlate of cognitive ability (Lynn & Mikk, 2009). The estimated associations, although attenuated, remain statistically and economically meaningful.

Second, we estimate the association of generalized trust with cognitive ability and childhood disadvantage. Table 2 presents the main findings. As our parameters are estimated via multinomial logistic regression, we report outcome-specific AMEs, which capture the change in the predicted probability of each generalized trust outcome associated with a one standard deviation increase in our age-effect-free general cognitive ability factor or a discrete change (from zero to one) in childhood disadvantage.

Table 2.

Associations Between Cognitive Ability, Childhood Disadvantage, and Generalized Trust.

	Dependent variable: Generalized trust
Model/Predictors	Can’t be too careful	It depends	People can be trusted
Model 1: Exogenous controls
Disadv.	.068*** [.052, .085]	−.036*** [−.050, −.023]	−.032*** [−.047, −.017]
Cognitive ability	−.079*** [−.085, −.073]	.001 [−.005, .006]	.079*** [.072, .085]
Model 2: Exogenous and sociodemographic controls
Disadv.	.054*** [.037, .070]	−.036*** [−.049, −.022]	−.018* [−.033, −.003]
Cognitive ability	−.064*** [−.070, −.058]	.006* [.000, .012]	.058*** [.051, .065]
Model 3: Exogenous, sociodemographic, and educational attainment controls
Disadv.	.033*** [.017, .049]	−.027*** [−.041, −.013]	−.006 [−.022, .009]
Cognitive ability	−.043*** [−.049, −.036]	−.003 [−.009, .003]	.046*** [.039, .053]
Observations	24,140	24,140	24,140

Note. Main entries are outcome-specific average marginal effects (AMEs); 95% confidence intervals using robust standard errors are reported in brackets. Disadv. is a dichotomous variable that takes on the value of one for those who experienced two or more dimensions of childhood disadvantage, and zero otherwise. Cognitive ability is our standardized age-effect-free general cognitive ability factor. Exogenous controls include age (in linear and quadratic form) and sex. Sociodemographic controls include housing tenure, marital status, square root of household size, logarithm of household income (which is adjusted by the Organization for Economic Co-operation and Development-modified equivalence scale and deflated by the Consumer Price Index), economic activity, and region of residence. Educational attainment represents the highest level of attainment. All models include a missingness indicator for incomplete childhood disadvantage information.

p < .05. ***p < .001.

In Model 1 of Table 2, we control for our exogenous regressors (i.e., age and sex). Consistent with previous studies, cognitive ability is positively associated with the probability of reporting “most people can be trusted,” while childhood disadvantage is negatively associated with this probability (Corgnet et al., 2016; Kim et al., 2025; Hooghe et al., 2012; Mell et al., 2022; Rindermann, 2008; Stamos et al., 2019). Here, a one standard deviation increase in cognitive ability is associated with a 7.9 percentage point increase (and corresponding decrease) in the probability of reporting “most people can be trusted” (“can’t be too careful”). To illustrate the substantive magnitude regarding the probability of reporting “most people can be trusted,” respondents with low cognitive ability (−2 standard deviations from the mean) have a predicted probability of 22.20%, whereas those with high cognitive ability (+2 standard deviations from the mean) have a predicted probability of 53.48%. For the probability of reporting “can’t be too careful,” the respective predicted probabilities are 55.59% and 24.07%. Cognitive ability has a much smaller association with the probability of choosing the “it depends” category.

In terms of childhood disadvantage, a discrete change from the reference group (i.e., non-disadvantaged) to the disadvantaged group is associated with a 3.2 (6.8) percentage point decrease (increase) in the probability of reporting “most people can be trusted” (“can’t be too careful”). Childhood disadvantage is also significantly associated with a reduced probability of choosing the “it depends” category. In Model 2, we add sociodemographic controls, and in Model 3, we further adjust for educational attainment. The estimated associations are attenuated but remain largely statistically and economically significant.

Third, our analysis now focuses on the moderation of the association between cognitive ability and generalized trust by childhood disadvantage. As marginal effects cannot be computed for interaction terms in nonlinear models like multinomial logistic, we report group-specific AMEs of cognitive ability on generalized trust for the non-disadvantaged and disadvantaged groups. We also, in Figure 2, provide a graphical representation of these AMEs, plotting the predicted probability of reporting “most people can be trusted” across the cognitive ability distribution for both the non-disadvantaged and disadvantaged groups.

Figure 2.

Moderation of the association between cognitive ability and generalized trust by childhood disadvantage. Panel A (Model 1, Table 3) includes exogenous controls. Panel B (Model 2, Table 3) includes exogenous and sociodemographic controls. Panel C (Model 3, Table 3) includes exogenous, sociodemographic, and educational attainment controls.

In Model 1 of Table 3 (Panel A, Figure 2), we control for our exogenous regressors (i.e., age and sex). The group-specific AMEs provide clear evidence that the association between cognitive ability and generalized trust is moderated by childhood disadvantage. Specifically, for the non-disadvantaged group, a one standard deviation increase in cognitive ability is associated with an 8.8 (8.9) percentage point increase (decrease) in the probability of reporting “most people can be trusted” (“can’t be too careful”). For the disadvantaged group, a one standard deviation increase in cognitive ability is associated with a 4.2 (4.5) percentage point increase (decrease) in the probability of reporting “most people can be trusted” (“can’t be too careful”), with the difference in group-specific AMEs being statistically significant.

Table 3.

Moderation of the Association Between Cognitive Ability and Generalized Trust by Childhood Disadvantage.

	Dependent variable: Generalized trust
Model/Predictors	Can’t be too careful	It depends	People can be trusted
Model 1: Exogenous controls
Cognitive ability \| Disadv. = 0	−.089*** [−.095, −.082]	.000 [−.006, .007]	.088*** [.081, .095]
Cognitive ability \| Disadv. = 1	−.045*** [−.059, −.032]	.003 [−.009, .015]	.042*** [.029, .056]
Δ Cognitive ability	−.043*** [−.059, -.028]	−.003 [−.016, .011]	.046*** [.030, .061]
Model 2: Exogenous and sociodemographic controls
Cognitive ability \| Disadv. = 0	−.074*** [−.081, −.067]	.006 [−.000, .013]	.067*** [.060, .075]
Cognitive ability \| Disadv. = 1	−.030*** [−.043, −.016]	.006 [−.006, .018]	.024*** [.010, .038]
Δ Cognitive ability	−.044*** [−.059, −.029]	.000 [−.013, .014]	.044*** [.028, .059]
Model 3: Exogenous, sociodemographic, and educational attainment controls
Cognitive ability \| Disadv. = 0	−.051*** [−.059, −.044]	−.003 [−.010, .004]	.054*** [.047, .062]
Cognitive ability \| Disadv. = 1	−.012 [−.026, .001]	−.002 [−.014, .011]	.014* [.000, .028]
Δ Cognitive ability	−.039*** [−.054, -.0240]	−.001 [−.015, .012]	.040*** [.025, .056]
Observations	24,140	24,140	24,140

Note. Main entries are outcome-specific average marginal effects (AMEs); 95% confidence intervals using robust standard errors are reported in brackets. Disadv. is a dichotomous variable that takes on the value of one for those who experienced two or more dimensions of childhood disadvantage, and zero otherwise. Cognitive ability is the standardized age-effect-free general cognitive ability factor. Cognitive ability | Disadv. = 0 reports the AME for the non-disadvantaged group, Cognitive ability | Disadv. = 1 reports the AME for the disadvantaged group, and ∆ Cognitive ability denotes the AME of the non-disadvantaged group minus the AME of the disadvantaged group. Exogenous controls include age (in linear and quadratic form) and sex. Sociodemographic controls include housing tenure, marital status, square root of household size, logarithm of household income (which is adjusted by the Organization for Economic Co-operation and Development-modified equivalence scale and deflated by the Consumer Price Index), economic activity, and region of residence. Educational attainment represents the highest level of attainment. All models include a missingness indicator for incomplete childhood disadvantage information.

p < .05. ***p < .001.

To illustrate the substantive magnitude, Panel A of Figure 2 shows that the predicted probability of reporting “most people can be trusted” for respondents with high cognitive ability (+2 standard deviations from the mean) is 55.98% for the non-disadvantaged group and 42.41% for the disadvantaged group. For respondents with low cognitive ability (−2 standard deviations from the mean), the respective predicted probabilities are 21.02% and 25.46%.

In Model 2 (Panel B, Figure 2), we add sociodemographic controls, and in Model 3 (Panel C, Figure 2), we further adjust for educational attainment. Despite these adjustments, the moderated association between cognitive ability and generalized trust by childhood disadvantage remains economically and statistically significant.³

Fourth, we examine the conditional indirect effects: the extent to which childhood disadvantage is associated with generalized trust through cognitive ability, conditional on disadvantage status. These indirect effects are therefore conditional on the observation that childhood disadvantage moderates the association between cognitive ability and generalized trust. Table 4 reports these conditional indirect effects, which combine the association between childhood disadvantage and cognitive ability (Table 1; Equation 1) with the group-specific AMEs of the association between cognitive ability and generalized trust (Table 3; Equation 3).

Table 4.

Conditional Indirect Effects of Childhood Disadvantage on Generalized Trust via Cognitive Ability.

	Dependent variable: Generalized trust
Model/Predictors	Can't be too careful	It depends	People can be trusted
Model 1: Exogenous controls
Indirect effect \| Disadv. = 0	.031*** [.027, .035]	−.000 [−.002, .002]	−.031*** [−.035, −.027]
Indirect effect \| Disadv. = 1	.016*** [.011, .021]	−.001 [−.005, .003]	−.015*** [−.020, −.010]
Δ Indirect effect	.015*** [.009, .021]	.001 [−.004, .006]	−.016*** [−.022, −.010]
Model 2: Exogenous and sociodemographic controls
Indirect effect \| Disadv. = 0	.017*** [.014, .019]	−.001 [−.003, .000]	−.015*** [−.018, −.012]
Indirect effect \| Disadv. = 1	.007*** [.003, .010]	−.001 [−.004, .001]	−.005** [−.009, −.002]
Δ Indirect effect	.010*** [.006, .014]	−.000 [−.003, .003]	−.010*** [−.014, −.006]
Model 3: Exogenous, sociodemographic, and educational attainment controls
Indirect effect \| Disadv. = 0	.006*** [.004, .008]	.000 [−.000, .001]	−.006*** [−.008, −.004]
Indirect effect \| Disadv. = 1	.001 [−.000, .003]	.000 [−.001, .002]	−.002 [−.003, .000]
Δ Indirect effect	.004*** [.002, .007]	.000 [−.001, .002]	−.005*** [−.007, −.002]
Observations	24,140	24,140	24,140

Note. Main entries are conditional indirect effects, calculated as the product of the coefficient for childhood disadvantage on cognitive ability and the outcome-specific average marginal effects (AMEs) of cognitive ability on generalized trust; 95% bias-corrected confidence intervals based on 1,000 bootstrap resamples are reported in brackets. Disadv. is a dichotomous variable that takes on the value of one for those who experienced two or more dimensions of childhood disadvantage, and zero otherwise. Indirect effect | Disadv. = 0 reports the indirect effect for the non-disadvantaged group, indirect effect | Disadv. = 1 reports the indirect effect for the disadvantaged group, and Δ Indirect effect denotes the indirect effect of the non-disadvantaged group minus the indirect effect of the disadvantaged group. Exogenous controls include age (in linear and quadratic form) and sex. Sociodemographic controls include housing tenure, marital status, square root of household size, logarithm of household income (which is adjusted by the Organization for Economic Co-operation and Development-modified equivalence scale and deflated by the Consumer Price Index), economic activity, and region of residence. Educational attainment represents the highest level of attainment. All models include a missingness indicator for incomplete childhood disadvantage information.

p < .01. ***p < .001.

As shown in Model 1 of Table 4—where we control for exogenous regressors—the lower cognitive ability associated with childhood disadvantage corresponds to a 3.1 percentage point decrease in the probability of reporting that “people can be trusted” among the non-disadvantaged. In contrast, for the disadvantaged group, this indirect association is significantly attenuated; here, the same cognitive ability difference is associated with a 1.5 percentage point decrease in the probability of reporting that “people can be trusted,” with the difference in group-specific indirect effects being statistically significant.

This pattern of results remains robust across Models 2 and 3—where we add sociodemographic controls and further adjust for educational attainment, respectively—although the substantive magnitude of the associations is somewhat reduced as additional controls are introduced. Taken together, these findings indicate a significantly attenuated conditional indirect effect: although childhood disadvantage is negatively associated with cognitive ability, the weaker association between cognitive ability and generalized trust among the disadvantaged group means that this developmental pathway is associated with markedly smaller differences in generalized trust. This indicates that the mediating role of cognitive ability in the association between childhood disadvantage and generalized trust is weaker among those who experienced early-life adversity.

Lastly, we assess whether our results are robust to alternative specifications. As noted above, the coding strategy for childhood disadvantage used in the main analyses is conservative, as it increases the likelihood of false negatives—classifying some disadvantaged individuals as non-disadvantaged—and may therefore attenuate estimated differences between advantaged and disadvantaged individuals toward zero.

First, we address missing information on childhood disadvantage using multiple imputation by chained equations. Specifically, we impute missing values on the four binary childhood disadvantage indicators using logistic regression models conditioned on generalized trust, cognitive ability, and all exogenous, sociodemographic, and educational controls. We generate 20 imputed datasets, construct the binary childhood disadvantage indicator—categorizing individuals as disadvantaged if they experienced two or more dimensions of disadvantage—passively within each imputation, and combine estimates across imputations following Rubin’s rules (Rubin, 1976, 1987). As expected, this procedure classifies a larger share of respondents as disadvantaged—as the categorization of disadvantage in the main analyses increases the likelihood of false negatives—suggesting that the estimates from the main analysis represent conservative lower bounds. The results, reported in Supplemental Tables S2 to S5, are fully consistent with this interpretation, and indeed, the main findings.

Second, we implement a complete-case restriction, limiting the sample to respondents with valid responses on all four childhood disadvantage dimensions. This specification also serves as a bounding exercise, as missingness in these indicators reflects a degree of conditional observability tied to childhood circumstances rather than random nonresponse, meaning that a disproportionate number of disadvantaged respondents may be excluded. Consistent with this, the restriction reduces the sample size to 19,898 individuals, of whom 17.90% (n = 3,562) are classified as disadvantaged—experienced two or more dimensions of disadvantage. Again, the results, reported in Supplemental Tables S6 to S9, are fully consistent with this interpretation, and indeed, the main findings.

Third, we replicate our main analysis using alternative constructions of the childhood disadvantage measure. Specifically, we use: (a) a cumulative index of disadvantage, defined as the number of distinct childhood disadvantage dimensions experienced, allowing us to assess whether associations are additive and incremental (Supplemental Tables S10–S13); (b) a proportion-based measure, defined as the number of observed childhood disadvantage dimensions divided by the number of non-missing indicators, with disadvantage classified as a proportion of .5 or greater (20.45%, n = 4,936; Supplemental Tables S14–S17); (c) a less restrictive classification defining respondents as disadvantaged if they experienced at least one dimension of disadvantage (51.08%, n = 12,330; Tables S18 to S21 of the Supplemental Material); and (d) separate component analyses in which each dimension of childhood disadvantage is examined individually (Supplemental Tables S22 to S28 and Supplemental Figures S1 to S4). All alternative specifications yield results consistent with the main analyses.

General Discussion

Our findings indicate that childhood disadvantage is associated with lower cognitive ability in adulthood and that it conditions the association between cognitive ability and generalized trust. This challenges the common assumption of independent effects in trust research and underscores how social inequality structures the association between cognitive resources and social attitudes. The magnitudes are economically meaningful. First, conditional on age and sex, childhood disadvantage is associated with a .348 standard deviation lower level of cognitive ability. Second, the association between cognitive ability and generalized trust differs by disadvantage status: a one standard deviation increase in cognitive ability is associated with an 8.8 percentage point higher probability of reporting that “most people can be trusted” among the non-disadvantaged, compared to a 4.2 percentage point higher probability among the disadvantaged. Third, taken together, these patterns indicate a substantially attenuated indirect association, whereby the mediating role attributed to cognitive ability in the association between childhood disadvantage and generalized trust is weaker among those who experienced early-life adversity.

These findings contribute to broader theoretical frameworks of social immobility. Immobility is commonly attributed to multiple mechanisms, including human capital transmission, whereby families with greater resources have better access to education and can invest more in their children (Becker & Tomes, 1986; Chetty et al., 2014; Loury, 1981); neighborhood effects, in which place-based disadvantage is associated with long-term outcomes (Chetty & Hendren, 2018); social capital, particularly connections to more educated or affluent individuals (Chetty et al., 2022); and aspiration failures, where limited opportunities are associated with lower goal formation (Genicot & Ray, 2017). Our results suggest a complementary mechanism: disadvantaged childhood environments are associated with attenuated social returns to cognitive ability. This raises concerns about the underutilization of human potential due to resource constraints within families. These findings are also consistent with evidence from childhood intervention programs, which indicates that enriched early environments are associated with higher levels of noncognitive skills and improved social and economic outcomes (Borghans et al., 2008; Cunha et al., 2006; Heckman et al., 2010). Relatedly, Stansbury and Rodriguez (2024) documented that lower social class is associated with reduced career attainment in academia, partly through differences in cultural and social capital. Taken together, an important direction for future research is to examine whether other beneficial traits with cognitive components—such as economic preferences (Dohmen et al., 2010, 2018) or personality dimensions (Stanek & Ones, 2023)—also exhibit environment-dependent associations with cognitive ability.

Our analyses also provide an opening for perspectives on the mechanisms through which “fast” and “slow” strategy environments lead to differential levels of trust. Specifically, our finding that disadvantaged backgrounds attenuate the positive relationship between cognitive ability and trust aligns with the perspective that early adversity may impose pervasive constraints—such as emotional dysregulation, chronic stress, and negative affect—that limit the expression of cognitive resources (Reiter et al., 2023). This finding is also consistent with the resource multiplication perspective and the Matthew Effect, whereby higher levels of one resource are associated with greater benefits from other resources. Specifically, “slow” strategy advantaged early-life environments—often characterized by low crime, effective institutions, and cooperative social norms—may provide contexts in which individuals with higher cognitive ability are likely to recognize and learn that generalized trust is a socially adaptive and often rewarded strategy. In contrast, “fast” strategy disadvantaged early-life environments—characterized by frequent exposure to corrupt or unreliable institutions and high-crime neighborhoods—may lack the environmental context that supports the positive association between cognitive ability and generalized trust, reinforcing persistent inequalities in social capital. Our results are therefore consistent with attenuation-based accounts and, as such, do not provide direct support for adaptive calibration models, in which cognitive resources are strategically allocated in response to ecological conditions. However, future empirical research should aim to isolate and estimate the relative contributions of these theoretical mechanisms.

A significant limitation of this study is that cognitive ability was assessed in adulthood, long after childhood background had begun shaping developmental trajectories. As adult cognitive performance reflects both underlying cognitive capacity and the cumulative impact of educational and environmental exposures (Ayoub et al., 2018), it cannot be treated as fully independent of the stratification processes under investigation. More specifically, if childhood disadvantage attenuates the expression of genetic potential—as suggested by the Scarr–Rowe hypothesis, whereby genetic associations with cognitive ability are weaker in resource-scarce environments (Rowe et al., 1999; Scarr-Salapatek, 1971; Tucker-Drob et al., 2013)—then phenotypic cognitive ability becomes endogenous to the very stratification processes being studied. It is simultaneously a product of those processes and the focal predictor whose effects are moderated by them. Even a moderated mediation approach cannot fully disentangle an individual’s latent cognitive capacity from environmental influences on its expression. Distinguishing genetic propensity from realized cognitive ability would allow a more direct assessment of whether the environment-dependent patterns documented here reflect the differential expression of underlying potential or the cumulative environmental impact on the development and maintenance of cognitive skills. Furthermore, while the current results do not directly support the adaptive calibration model, genetically informed designs would provide a more stringent test of such accounts, consistent with the view that selection may favor phenotypic plasticity, whereby a single genotype supports a range of outcomes depending on ecological conditions (Del Giudice, 2014).

Lastly, although not within the scope of this article, our findings suggest a wider perspective on whether these data patterns might translate to developed and developing countries. As a simple test of this interpretation, we follow the strategy in Chowdhury et al. (2022) and test for a positive relationship between cognitive ability and trust in high-income countries and an attenuated relationship in low- and middle-income countries. To implement this strategy, we accessed the data from the Global Preferences Survey of Falk et al. (2018) and applied the World Bank’s classiﬁcation of countries into high-income and low-to-middle-income groups. We then used math skills as a proxy for cognitive ability and linked this proxy to trust attitudes—a self-assessed measure based upon the response to “I assume that people have only the best intentions.” The associations are summarized in Figure 3. Consistent with our main results, for high-income countries, we see a positive and statistically significant relationship between cognitive ability and trust ( $β = . 587, 95 % CI [0.322, 0.852], p < . 001$ ); for low-to-middle-income countries, there is an attenuated and nonsignificant relationship ( $β = . 203, 95 % CI [- 0.085, 0.491], p = . 167) .$ A Wald test following seemingly unrelated estimation indicated that the difference between these coefficients is statistically significant ( $χ^{2} (1) = 3.70, p = . 054$ ).

Figure 3.

Relationship between cognitive ability and trust, conditional on income level of country.

Supplemental Material

sj-docx-1-psp-10.1177_01461672261439412 – Supplemental material for What Childhood Leaves Behind: Cognitive Ability and Trust in Adulthood

Supplemental material, sj-docx-1-psp-10.1177_01461672261439412 for What Childhood Leaves Behind: Cognitive Ability and Trust in Adulthood by Chris Dawson in Personality and Social Psychology Bulletin

Footnotes

ORCID iD

Chris Dawson

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Open Practices

The data that support the findings of this study are publicly available from the U.K. Data Archive as Understanding Society: Waves 1–15, 2009–2024, and Harmonised BHPS: Waves 1–18, 1991–2009. The complete Stata analysis script to replicate the results is openly available on the Open Science Framework at: . This study was not preregistered.

Supplemental Material

Supplemental material is available online with this article.

Notes

References

Åkerlund

Golsteyn

B. H.

Grönqvist

Lindahl

(2016). Time discounting and criminal behavior. Proceedings of the National Academy of Sciences, 113(22), 6160–6165.

Andersson

M. A.

(2016). Health returns to education by family socioeconomic origins, 1980–2008: Testing the importance of gender, cohort, and age. SSM-Population Health, 2, 549–560.

Ayoub

Gosling

S. D.

Potter

Shanahan

Roberts

B. W.

(2018). The relations between parental socioeconomic status, personality, and life outcomes. Social Psychological and Personality Science, 9(3), 338–352.

Bailey

P. E.

Leon

(2019). A systematic review and meta-analysis of age-related differences in trust. Psychology and Aging, 34(5), 674–685.

Balliet

N. P.

Macfarlan

S. J.

Van Vugt

(2011). Sex differences in cooperation: A meta-analytic review of social dilemmas. Psychological Bulletin, 137(6), 881–909.

Becker

G. S.

Tomes

(1986). Human capital and the rise and fall of families. Journal of Labor Economics, 4(3, Part 2), S1–S39.

Belsky

Pluess

(2009). Beyond diathesis stress: Differential susceptibility to environmental influences. Psychological Bulletin, 135(6), 885–908.

Benton

(2008). Micronutrient status, cognition and behavioral problems in childhood. European Journal of Nutrition, 47, 38–50.

Bernardi

(2014). Compensatory advantage as a mechanism of educational inequality: A regression discontinuity based on month of birth. Sociology of Education, 87(2), 74–88.

10.

Blanchette

Richards

(2010). The influence of affect on higher level cognition: A review of research on interpretation, judgement, decision making and reasoning. Cognition & Emotion, 24(4), 561–595.

11.

Bond

T. N.

Lang

(2019). The sad truth about happiness scales. Journal of Political Economy, 127(4), 1629–1640.

12.

Bonin

Dohmen

Falk

Huffman

Sunde

(2007). Cross-sectional earnings risk and occupational sorting: The role of risk attitudes. Labour Economics, 14(6), 926–937.

13.

Borghans

Duckworth

A. L.

Heckman

J. J.

Ter Weel

(2008). The economics and psychology of personality traits. Journal of Human Resources, 43(4), 972–1059.

14.

Bradley

R. H.

Corwyn

R. F.

(2002). Socioeconomic status and child development. Annual Review of Psychology, 53(2002), 371–399.

15.

Brehm

Rahn

(1997). Individual-level evidence for the causes and consequences of social capital. American Journal of Political Science, 41(3), 999–1023.

16.

Britto

P. R.

Lye

S. J.

Proulx

Yousafzai

A. K.

Matthews

S. G.

Vaivada

Perez-Escamilla

Rao

Fernald

L. C. H.

MacMillan

Hanson

Wachs

T. D.

Yao

Yoshikawa

Cerezo

Leckman

J. F.

Bhutta

Z. A.

, & Early Childhood Development Interventions Review Group, for the Lancet Early Childhood Development Series Steering Committee. (2017). Nurturing care: Promoting early childhood development. The Lancet, 389(10064), 91–102.

17.

Cadena

B. C.

Keys

B. J.

(2015). Human capital and the lifetime costs of impatience. American Economic Journal: Economic Policy, 7(3), 126–153.

18.

Caner

Okten

(2010). Risk and career choice: Evidence from Turkey. Economics of Education Review, 29(6), 1060–1075.

19.

Charron

Rothstein

(2016). Does education lead to higher generalized trust? The importance of quality of government. International Journal of Educational Development, 50, 59–73.

20.

Chetty

Hendren

(2018). The impacts of neighborhoods on intergenerational mobility I: Childhood exposure effects. The Quarterly Journal of Economics, 133(3), 1107–1162.

21.

Chetty

Hendren

Kline

Saez

(2014). Where is the land of opportunity? The geography of intergenerational mobility in the United States. The Quarterly Journal of Economics, 129(4), 1553–1623.

22.

Chetty

Jackson

M. O.

Kuchler

Stroebel

Hendren

Fluegge

R. B.

Gong

Gonzalez

Grondin

Jacob

Johnston

Koenen

Laguna-Muggenburg

Mudekereza

Rutter

Thor

Townsend

Zhang

Bailey

, . . . & Wernerfelt

(2022). Social capital I: Measurement and associations with economic mobility. Nature, 608(7921), 108–121.

23.

Chowdhury

Sutter

Zimmermann

K. F.

(2022). Economic preferences across generations and family clusters: A large-scale experiment in a developing country. Journal of Political Economy, 130(9), 2361–2410.

24.

Cinelli

Forney

Pearl

(2024). A crash course in good and bad controls. Sociological Methods & Research, 53(3), 1071–1104.

25.

Corgnet

Espín

A. M.

Hernán-González

Kujal

Rassenti

(2016). To trust, or not to trust: Cognitive reflection in trust games. Journal of Behavioral and Experimental Economics, 64, 20–27.

26.

Cosmides

Barrett

H. C.

Tooby

(2010). Adaptive specializations, social exchange, and the evolution of human intelligence. Proceedings of the National Academy of Sciences, 107, 9007–9014.

27.

Cunha

Heckman

J. J.

Lochner

Masterov

D. V.

(2006). Interpreting the evidence on life cycle skill formation. Handbook of the Economics of Education, 1, 697–812.

28.

Curry

O. S.

Price

M. E.

Price

J. G.

(2008). Patience is a virtue: Cooperative people have lower discount rates. Personality and Individual differences, 44(3), 780–785.

29.

Damian

R. I.

Shanahan

Trautwein

Roberts

B. W.

(2015). Can personality traits and intelligence compensate for background disadvantage? Predicting status attainment in adulthood. Journal of Personality and Social Psychology, 109(3), 473–489.

30.

Dawson

(2019). How persistent is generalised trust?. Sociology, 53(3), 590–599.

31.

DellaVigna

Paserman

M. D.

(2005). Job search and impatience. Journal of Labor Economics, 23(3), 527–588.

32.

Del Giudice

(2014). Early stress and human behavioral development: Emerging evolutionary perspectives. Journal of Developmental Origins of Health and Disease, 5(4), 270–280.

33.

Del Giudice

Ellis

B. J.

Shirtcliff

E. A.

(2011). The adaptive calibration model of stress responsivity. Neuroscience & Biobehavioral Reviews, 35(7), 1562–1592.

34.

Dohmen

Falk

(2011). Performance pay and multidimensional sorting: Productivity, preferences, and gender. American Economic Review, 101(2), 556–590.

35.

Dohmen

Falk

Huffman

Sunde

(2010). Are risk aversion and impatience related to cognitive ability?. American Economic Review, 100(3), 1238–1260.

36.

Dohmen

Falk

Huffman

Sunde

(2018). On the relationship between cognitive ability and risk preference. Journal of Economic Perspectives, 32(2), 115–134.

37.

Dórea

J. G.

(2021). Exposure to environmental neurotoxic substances and neurodevelopment in children from Latin America and the Caribbean. Environmental Research, 192, 110199.

38.

Dunn

J. R.

Schweitzer

M. E.

(2005). Feeling and believing: The influence of emotion on trust. Journal of Personality and Social Psychology, 88(5), 736–748.

39.

Eckel

C. C.

Grossman

P. J.

Johnson

C. A.

De Oliveira

A. C.

Rojas

Wilson

R. K.

(2012). School environment and risk preferences: Experimental evidence. Journal of Risk and Uncertainty, 45, 265–292.

40.

Eysenck

M. W.

Derakshan

Santos

Calvo

M. G.

(2007). Anxiety and cognitive performance: Attentional control theory. Emotion, 7(2), 336–353.

41.

Evans

G. W.

Kim

(2013). Childhood poverty, chronic stress, self-regulation, and coping. Child Development Perspectives, 7(1), 43–48.

42.

Fan

Porter

(2020). Reinforcement or compensation? Parental responses to children’s revealed human capital levels. Journal of Population Economics, 33(1), 233–270.

43.

Falk

Becker

Dohmen

Enke

Huffman

Sunde

(2018). Global evidence on economic preferences. The Quarterly Journal of Economics, 133(4), 1645–1692.

44.

Falk

Kosse

Pinger

Schildberg-Hörisch

Deckers

(2021). Socioeconomic status and inequalities in children’s IQ and economic preferences. Journal of Political Economy, 129(9), 2504–2545.

45.

Frankenhuis

W. E.

de Weerth

(2013). Does early-life exposure to stress shape or impair cognition?. Current Directions in Psychological Science, 22(5), 407–412.

46.

Frankenhuis

W. E.

Panchanathan

Nettle

(2016). Cognition in harsh and unpredictable environments. Current Opinion in Psychology, 7, 76–80.

47.

Gächter

Herrmann

Thöni

(2004). Trust, voluntary cooperation, and socio-economic background: Survey and experimental evidence. Journal of Economic Behavior & Organization, 55(4), 505–531.

48.

Galobardes

Shaw

Lawlor

D. A.

Lynch

J. W.

Smith

G. D.

(2006). Indicators of socioeconomic position (part 1). Journal of Epidemiology & Community Health, 60(1), 7–12.

49.

Gellci

Marusak

H. A.

Peters

Elrahal

Iadipaolo

A. S.

Rabinak

C. A.

(2019). Community and household-level socioeconomic disadvantage and functional organization of the salience and emotion network in children and adolescents. NeuroImage, 184, 729–740.

50.

Genicot

Ray

(2017). Aspirations and inequality. Econometrica, 85(2), 489–519.

51.

Ghirardi

Gil-Hernández

C. J.

Bernardi

van Bergen

Demange

(2024). Interaction of family SES with children’s genetic propensity for cognitive and noncognitive skills: No evidence of the Scarr-Rowe hypothesis for educational outcomes. Research in Social Stratification and Mobility, 92, 100960.

52.

Gladden

P. R.

Figueredo

A. J.

Jacobs

W. J.

(2009). Life history strategy, psychopathic attitudes, personality, and general intelligence. Personality and Individual Differences, 46(3), 270–275.

53.

Golsteyn

B. H.

Grönqvist

Lindahl

(2014). Adolescent time preferences predict lifetime outcomes. The Economic Journal, 124(580), F739–F761.

54.

Griskevicius

Delton

A. W.

Robertson

T. E.

Tybur

J. M.

(2011). Environmental contingency in life history strategies: The influence of mortality and socioeconomic status on reproductive timing. Journal of Personality and Social Psychology, 100(2), 241–254.

55.

Griskevicius

Ackerman

J. M.

Cantú

S. M.

Delton

A. W.

Robertson

T. E.

Simpson

J. A.

Thompson

M. E.

Tybur

J. M.

(2013). When the economy falters, do people spend or save? Responses to resource scarcity depend on childhood environments. Psychological Science, 24(2), 197–205.

56.

Gunnar

M. R.

(2000). Early adversity and the development of stress reactivity and regulation. In Nelson

C. A.

(Ed.), The effects of early adversity on neurobehavioral development (pp. 163–200). Psychology Press.

57.

Gunnar

Quevedo

(2007). The neurobiology of stress and development. Annual Review of Psychology, 58(1), 145–173.

58.

Hackman

D. A.

Farah

M. J.

Meaney

M. J.

(2010). Socioeconomic status and the brain: Mechanistic insights from human and animal research. Nature Reviews Neuroscience, 11(9), 651–659.

59.

Helliwell

J. F.

Huang

Wang

(2017). New evidence on trust and well-being. In Uslaner

E. M.

(Ed.), The Oxford Handbook of Social and Political Trust (pp. 409–446). Oxford University Press.

60.

Heckman

J. J.

Moon

S. H.

Pinto

Savelyev

P. A.

Yavitz

(2010). The rate of return to the HighScope Perry Preschool Program. Journal of Public Economics, 94(1–2), 114–128.

61.

Heckman

J. J.

Rubinstein

(2001). The importance of noncognitive skills: Lessons from the GED testing program. American Economic Review, 91(2), 145–149.

62.

Heckman

J. J.

Stixrud

Urzua

(2006). The effects of cognitive and noncognitive abilities on labor market outcomes and social behavior. Journal of Labor Economics, 24(3), 411–482.

63.

Holm

Hjorth-Trolle

Jæger

M. M.

(2019). Signals, educational decision-making, and inequality. European Sociological Review, 35(4), 447–460.

64.

Hooghe

Marien

De Vroome

(2012). The cognitive basis of trust. The relation between education, cognitive ability, and generalized and political trust. Intelligence, 40(6), 604–613.

65.

Jones

(2008). Are smarter groups more cooperative? Evidence from prisoner's dilemma experiments, 1959–2003. Journal of Economic Behavior & Organization, 68(3–4), 489–497.

66.

Kahneman

Frederick

(2005). A model of intuitive judgment. In Holyoak

K. J.

Morrison

R. G.

(Eds.), The Cambridge handbook of thinking and reasoning (pp. 267–293). Cambridge University Press.

67.

Kim

Evans

G. W.

Angstadt

S. S.

Sripada

C. S.

Swain

J. E.

Liberzon

Phan

K. L.

(2013). Effects of childhood poverty and chronic stress on emotion regulatory brain function in adulthood. Proceedings of the National Academy of Sciences, 110(46), 18442–18447.

68.

Kim

Y. I.

VanderWeele

T. J.

Johnson

B. R.

(2025). Childhood predictors of perceptions of social trust across 22 countries in the global flourishing study. Scientific Reports, 15(1), 14358.

69.

Knack

Keefer

(1997). Does social capital have an economic payoff? A cross-country investigation. The Quarterly Journal of Economics, 112(4), 1251–1288.

70.

Kraus

M. W.

Horberg

E. J.

Goetz

J. L.

Keltner

(2011). Social class rank, threat vigilance, and hostile reactivity. Personality and Social Psychology Bulletin, 37(10), 1376–1388.

71.

Fung

H. H.

(2013). Age differences in trust: An investigation across 38 countries. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 68(3), 347–355.

72.

Lohse

(2016). Smart or selfish–when smart guys finish nice. Journal of Behavioral and Experimental Economics, 64, 28–40.

73.

Loury

G. C.

(1981). Intergenerational transfers and the distribution of earnings. Econometrica: Journal of the Econometric Society, 49(4), 843–867.

74.

Lynn

Mikk

(2009). National IQs predict educational attainment in math, reading and science across 56 nations. Intelligence, 37(3), 305–310.

75.

McAllister

D. J.

(1995). Affect-and cognition-based trust as foundations for interpersonal cooperation in organizations. Academy of Management Journal, 38(1), 24–59.

76.

McFall

(2013). Understanding society: UK household longitudinal study: Cognitive ability measures. Institute for Social and Economic Research, University of Essex.

77.

Mell

Safra

Demange

Algan

Baumard

Chevallier

(2022). Early life adversity is associated with diminished social trust in adults. Political Psychology, 43(2), 317–335.

78.

Metcalfe

L. A.

Harvey

E. A.

Laws

H. B.

(2013). The longitudinal relation between academic/cognitive skills and externalizing behavior problems in preschool children. Journal of Educational Psychology, 105(3), 881–894.

79.

Millet

Dewitte

(2007). Altruistic behavior as a costly signal of general intelligence. Journal of Research in Personality, 41(2), 316–326.

80.

Myers

C. D.

Tingley

(2016). The influence of emotion on trust. Political Analysis, 24(4), 492–500.

81.

Norton

E. C.

Dowd

B. E.

(2018). Log odds and the interpretation of logit models. Health Services Research, 53(2), 859–878.

82.

Oldrati

Patricelli

Colombo

Antonietti

(2016). The role of dorsolateral prefrontal cortex in inhibition mechanism: A study on cognitive reflection test and similar tasks through neuromodulation. Neuropsychologia, 91, 499–508.

83.

Pepper

G. V.

Nettle

(2017). The behavioural constellation of deprivation: Causes and consequences. Behavioral and Brain Sciences, 40, e314.

84.

Petersen

M. B.

Aarøe

(2015). Birth weight and social trust in adulthood: Evidence for early calibration of social cognition. Psychological Science, 26(11), 1681–1692.

85.

Potter

Bridger

Piotrowska

P. J.

Drewelies

(2025). Emotional reactivity to daily positive and negative events in adulthood: The role of adverse childhood experiences. Emotion, 25(6), 1502–1515.

86.

Proto

Rustichini

Sofianos

(2019). Intelligence, personality, and gains from cooperation in repeated interactions. Journal of Political Economy, 127(3), 1351–1390.

87.

Poulin

M. J.

Haase

C. M.

(2015). Growing to trust: Evidence that trust increases and sustains well-being across the life span. Social Psychological and Personality Science, 6(6), 614–621.

88.

Reiter

A. M.

Hula

Vanes

Hauser

T. U.

Kokorikou

Goodyer

I. M.

, NSPN Consortium, Fonagy

Moutoussis

Dolan

R. J.

(2023). Self-reported childhood family adversity is linked to an attenuated gain of trust during adolescence. Nature Communications, 14(1), 6920.

89.

Righetti

Finkenauer

(2011). If you are able to control yourself, I will trust you: The role of perceived self-control in interpersonal trust. Journal of Personality and Social Psychology, 100(5), 874–886.

90.

Rindermann

(2008). Relevance of education and intelligence at the national level for the economic welfare of people. Intelligence, 36(2), 127–142.

91.

Robbins

B. G.

(2022). Measuring generalized trust: Two new approaches. Sociological Methods & Research, 51(1), 305–356.

92.

Ronda

Agerbo

Bleses

Mortensen

P. B.

Børglum

Mors

Rosholm

Hougaard

D. M.

Nordentoft

Werge

(2022). Family disadvantage, gender, and the returns to genetic human capital. The Scandinavian Journal of Economics, 124(2), 550–578.

93.

Rosenberg

(1956). Misanthropy and political ideology. American Sociological Review, 21(6), 690–695.

94.

Ross

C. E.

Mirowsky

(2011). The interaction of personal and parental education on health. Social Science & Medicine, 72(4), 591–599.

95.

Rowe

D. C.

Jacobson

K. C.

Van den Oord

E. J.

(1999). Genetic and environmental influences on vocabulary IQ: Parental education level as moderator. Child Development, 70(5), 1151–1162.

96.

Rubin

D. B.

(1976). Inference and missing data. Biometrika, 63(3), 581–592.

97.

Rubin

D. B.

(1987). Multiple imputation for non-response in surveys. John Wiley & Sons.

98.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277(5328), 918–924.

99.

Scarr-Salapatek

(1971). Race, Social Class, and IQ: Population differences in heritability of IQ scores were found for racial and social class groups. Science, 174(4016), 1285–1295.

100.

Segal

N. L.

Hershberger

S. L.

(1999). Cooperation and competition between twins: Findings from a prisoner’s dilemma game. Evolution and Human Behavior, 20(1), 29–51.

101.

Shah

A. K.

Mullainathan

Shafir

(2012). Some consequences of having too little. Science, 338(6107), 682–685.

102.

Shamosh

N. A.

Gray

J. R.

(2008). Delay discounting and intelligence: A meta-analysis. Intelligence, 36(4), 289–305.

103.

Sheehy-Skeffington

(2020). The effects of low socioeconomic status on decision-making processes. Current Opinion in Psychology, 33, 183–188.

104.

Simonovic

Stupple

E. J.

Gale

Sheffield

(2017). Stress and risky decision making: Cognitive reflection, emotional learning or both. Journal of Behavioral Decision Making, 30(2), 658–665.

105.

Simpson

J. A.

Griskevicius

Kuo

S. I.

Sung

Collins

W. A.

(2012). Evolution, stress, and sensitive periods: The influence of unpredictability in early versus late childhood on sex and risky behavior. Developmental Psychology, 48(3), 674–686.

106.

Stamos

Altsitsiadis

Dewitte

(2019). Investigating the effect of childhood socioeconomic background on interpersonal trust: Lower childhood socioeconomic status predicts lower levels of trust. Personality and Individual Differences, 145, 19–25.

107.

Stanek

K. C.

Ones

D. S.

(2023). Meta-analytic relations between personality and cognitive ability. Proceedings of the National Academy of Sciences, 120(23), e2212794120.

108.

Stanovich

K. E.

West

R. F.

(2008). On the relative independence of thinking biases and cognitive ability. Journal of Personality and Social Psychology, 94(4), 672–695.

109.

Stansbury

Rodriguez

(2024). The class gap in career progression: Evidence from us academia. Technical report. Working Paper.

110.

Sutter

Kocher

M. G.

(2007). Trust and trustworthiness across different age groups. Games and Economic Behavior, 59(2), 364–382.

111.

Tucker-Drob

E. M.

Briley

D. A.

Harden

K. P.

(2013). Genetic and environmental influences on cognition across development and context. Current Directions in Psychological Science, 22(5), 349–355.

112.

Tversky

Kahneman

(1996). On the reality of cognitive illusions. Psychological Review, 103(3), 582–591.

113.

Uslaner

E. M.

(2002). The moral foundations of trust. Cambridge University Press.

114.

Uslaner

E. M.

(2008). Where you stand depends upon where your grandparents sat: The inheritability of generalized trust. Public Opinion Quarterly, 72(4), 725–740.

115.

Van Den Akker

O. R.

van Assen

M. A.

Van Vugt

Wicherts

J. M.

(2020). Sex differences in trust and trustworthiness: A meta-analysis of the trust game and the gift-exchange game. Journal of Economic Psychology, 81, 102329.

116.

von Stumm

(2012). You are what you eat? Meal type, socio-economic status and cognitive ability in childhood. Intelligence, 40(6), 576–583.

117.

Wiepking

Maas

(2009). Resources that make you generous: Effects of social and human resources on charitable giving. Social Forces, 87(4), 1973–1995.

118.

Williams

(2012). Using the margins command to estimate and interpret adjusted predictions and marginal effects. The Stata Journal, 12(2), 308–331.

119.

Whitley

Deary

I. J.

Ritchie

S. J.

Batty

G. D.

Kumari

Benzeval

(2016). Variations in cognitive abilities across the life course: Cross-sectional evidence from Understanding Society: The UK Household Longitudinal Study. Intelligence, 59, 39–50.

120.

Yamagishi

(2001). Trust as a form of social intelligence. In Cook

(Ed.), Trust in society. Russell Sage Foundation.

121.

Zahavi

(1977). The cost of honesty (further remarks on the handicap principle). Journal of Theoretical Biology, 67(3), 603–605.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.62 MB

0.00 MB