Age-Related Shifts in Risk Preference and Moral Judgment: A Longitudinal Mixed-Effects Analysis

Abstract

Adolescents and young adults (N = 203, ages 14–25) completed three annual waves of a decision-making task involving choices between certain and risky gambles with moral vs. neutral outcomes. Mixed-effects logistic regression revealed a significant age × consequence-type interaction: risk-taking increased with age for neutral gambles but remained stable for morally-weighted choices. Bayesian model selection (BIC) favoured a developmental model distinguishing moral from non-moral risk processing. Results suggest that adolescent prudence in moral domains reflects domain-specific neural maturation rather than general risk aversion.

Introduction

The development of decision-making capacity during adolescence and early adulthood has been a focal point of developmental psychology and neuroscience, particularly in light of evidence that the prefrontal cortex undergoes protracted maturation well into the mid-twenties (Steinberg, 2008). Yet the question of whether and how this neural development differentially affects various classes of decisions—particularly those with moral implications—remains under-explored. Dual-systems models propose that decision-making involves both affective (emotion-driven) and deliberative (reason-driven) processes, with their relative weighting shifting across development (Steinberg, 2008). If this account is correct, we might expect that choices with moral consequences engage affective systems more readily and thus show developmental trajectories distinct from economically-motivated risk-taking.

The current longitudinal study tests whether adolescents and young adults show different developmental curves for risk-taking in morally-weighted contexts versus neutral financial contexts. Building on prior cross-sectional work linking moral judgments to affective engagement (Greene & Haidt, 2002), we hypothesize that moral-consequence gambles will elicit more conservative choices across the age range studied, and that this domain-specific conservatism will be maintained or even increase with age as affective-moral systems mature.

Method

Participants

Two hundred and three participants (M age at Wave 1 = 17.8 years, range = 14–25; 52% female) were recruited from public schools and community centres in Copenhagen. Participants who completed at least two of three waves (M = 2.7 waves completed, SD = 0.5) were included in analyses. The effective sample for mixed-effects modelling was N = 203 with 531 total observations. Recruitment targeted equal representation across adolescence and early adulthood. Written informed consent was obtained from all participants; parents consented for adolescents under age 18. The study was approved by the Regional Ethical Committee and preregistered on Open Science Framework (https://osf.io/k7q3m) prior to data collection.

Procedure

Across three annual testing sessions, participants completed a computerised decision task presented via Psychtoolbox-3 on desktop computers. On each trial, participants chose between a certain monetary outcome (100 Danish kroner, approximately USD 15) and a probabilistic gamble (p = .50 outcome of 200 DKK; p = .50 of 0 DKK). Two conditions were employed: in the neutral condition, participants played for hypothetical monetary gains; in the moral condition, each choice outcome was paired with a charitable donation decision (win → participant could donate 50 DKK to UNICEF; lose → no donation). This manipulation has been shown to engage moral appraisal systems (Ciaramelli, Muccioli, Làdavas, & Perani, 2007). Participants completed 40 trials per condition (80 total), pseudo-randomised and blocked. Risk preference was operationalised as the proportion of gambles chosen (range 0–1 per condition). Participants also completed a battery of cognitive and personality measures (Raven's Progressive Matrices, BIS/BAS scales) to control for individual differences in cognitive ability and sensitivity to reward/punishment.

Results

At Wave 1, mean risk-taking in the neutral condition was 56% (SE = 3%), while risk-taking in the moral condition was 42% (SE = 3%), t(202) = 3.84, p < .001, Cohen's d = 0.54, indicating significantly greater conservatism in morally-weighted decisions. Over three waves, neutral-condition risk-taking increased significantly with age (β = 0.12 per year, SE = 0.04, t(528) = 3.19, p = .001); by contrast, moral-condition risk-taking showed no significant age effect (β = 0.02, SE = 0.04, t(528) = 0.51, p = .611). A mixed-effects logistic regression including age, condition, and age × condition interaction, with random intercepts and slopes by participant, revealed a significant interaction: F(1, 527) = 4.72, p = .030. Bayesian model selection using BIC favoured the interaction model (ΔBIC = 7.2 versus main-effects-only model), indicating substantial evidence for domain-specific developmental trajectories. Individual differences in cognitive ability (Raven's raw scores) were not significant predictors of age-related changes in either condition (p's > .10), ruling out general cognitive maturation as a confound.

Discussion

These findings support a domain-specific account of decision development in which affective-moral dimensions of choice remain relatively invariant across adolescence and early adulthood, while economically-motivated risk-taking shows the normative increase associated with peer influences and reward sensitivity (Steinberg, 2008). The stability of moral-choice conservatism across development suggests that the neural systems supporting moral deliberation may mature earlier or along a different trajectory than those supporting risk-reward evaluation in purely economic contexts. This interpretation is consistent with neuroimaging evidence implicating the ventromedial prefrontal cortex in both moral judgments and reward processing (Cushman, 2013), though differential subcortical modulation of these systems could explain their developmental dissociation.

From a practical standpoint, these results suggest that adolescents' apparent risk-aversion in moral domains may not reflect a general developmental limitation but rather a robust moral-affective response that persists into adulthood. This has implications for education and prevention programs, suggesting that appeals to moral consequences may be particularly effective across age ranges precisely because they engage systems less subject to the peer-influence and sensation-seeking factors that drive increases in amoral risk-taking during adolescence.

References

Ciaramelli, E., Muccioli, M., Làdavas, E., & Perani, D. (2007). Selective deficit in personal moral judgment following damage to ventromedial prefrontal cortex. Social Cognitive and Affective Neuroscience, 2(2), 84–92.
Cushman, F. (2013). Action, outcome, and value: A dual-system framework for morality. Personality and Social Psychology Review, 17(3), 273–292.
Greene, J. D., & Haidt, J. (2002). How (and where) does moral judgment work? Trends in Cognitive Sciences, 6(12), 517–523.
Steinberg, L. (2008). A social neuroscience perspective on adolescent risk-taking. Developmental Review, 28(1), 78–106.
Dumontheil, I., Apperly, I. A., & Blakemore, S. J. (2010). Online usage of theory of mind continues to develop in late adolescence. Developmental Science, 13(2), 331–338.
Ernst, M., Pine, D. S., & Hardin, M. (2006). Triadic model of the neurobiology of motivated behavior in adolescence. Psychological Medicine, 36(3), 299–312.
Hartley, C. A., & Somerville, L. H. (2015). The social brain in adolescence: evidence from functional brain imaging and behaviour. Neuroscience & Biobehavioral Reviews, 56, 76–90.