Skip to main content
NCBI home page
Learning & Memory logoLink to Learning & Memory
. 2025 Jan;32(1):a054047. doi: 10.1101/lm.054047.124

The influence of exposure to early-life adversity on agency-modulated reinforcement learning

Hayley M Dorfman 1, Bryan JW Dong 1, Katie A McLaughlin 1,2, Elizabeth A Phelps 1,3,✉
PMCID: PMC11801475  PMID: 39870488

Abstract

Agency beliefs influence how humans learn from different contexts and outcomes. Research demonstrates that stressors, such as exposure to early-life adversity (ELA), are associated with both agency beliefs and learning, but how these processes interact remains unclear. The current study investigated whether exposure to ELA influences agency and interacts with reinforcement learning in adults. Replicating prior behavioral and computational work, ELA resulted in decreased learning, while increased adversity severity was associated with decreased latent agency beliefs. These findings suggest that exposure to adversity in childhood has a nuanced impact on reinforcement learning and agency beliefs in adulthood.

Previous work has shown that stressful experiences occurring during childhood and adolescence are associated with fundamental learning and decision-making processes (McLaughlin et al. 2019; Raio et al. 2022). Childhood exposure to physical, emotional, and sexual abuse, neglect, and violence, collectively termed early-life adversity (ELA), might be particularly influential for learning and decision-making outcomes given that it occurs during sensitive developmental time frames. In addition, prior work has shown that exposure to stressful circumstances is related to diminished beliefs about controllability, or agency (Bryant and Trockel 1976; Chorpita and Barlow 1998), and that exposure to adversity in childhood is associated with an external locus of control in adolescence (Culpin et al. 2015). But the precise mechanisms by which the association between ELA, learning, and agency interact remain unclear.

Over time, exposure to uncontrollable learning environments (e.g., noncontingent actions and outcomes) results in a phenomenon called learned helplessness, where humans and animals develop a lack of motivation, trouble with learning, and depression-like symptoms (Maier and Seligman 1976). Learned helplessness can occur even in environments where uncontrollable outcomes are rewarding versus punishing (Tennen et al. 1982). This has led theorists to argue that what matters most is not the probability of rewards and punishments in a given environment, but rather a person's explanatory style, or their beliefs about the cause of outcomes (Abramson et al. 1978; Skinner et al. 1998). Previously, we showed that beliefs about agency can explain the extent to which people learn from positive and negative outcomes, and that this agency-modulated learning process can be accounted for by a novel Bayesian reinforcement learning model (Dorfman et al. 2019). The current study investigates whether retrospectively reported exposure to ELA in childhood impacts these learning and agency belief biases in adulthood.

To assess ELA experiences retrospectively, participants completed the Childhood Trauma Questionnaire (CTQ; Bernstein et al. 1997), the Childhood Experiences of Care and Abuse assessment (CECA; Bifulco et al. 1994), the Questionnaire on Unpredictability in Childhood (QUIC; Glynn et al. 2019), the Life Events Checklist (LEC-5; Weathers et al. 2013), and the Juvenile Victimization Questionnaire (JVQ; Finkelhor et al. 2005).

For analyses using binary calculations of exposure to ELA (i.e., exposure to adversity vs. no exposure to adversity), participants were coded as being exposed to ELA if they met the following criteria: (1) scored greater than the validated cutoff scores (Bifulco et al. 2005) on any of the subscales on the CECA; (2) scored greater than the validated cutoff scores on any of the subscales on the CTQ (Bernstein et al. 1997); (3) endorsed any single relevant question on the JVQ-R2 (see Supplemental Material for more details). Many participants met multiple ELA criteria (see Supplemental Fig. S1). A continuous measure of ELA severity was also calculated by summing and standardizing the JVQ and adding it to a standardized combination of the three CTQ abuse subscales, similar to methods used in existing work (Lambert et al. 2017; Sumner et al. 2019).

Adult participants were recruited online through Prolific (www.Prolific.com). Eligibility requirements included residence in the United States, fluent English, and a Prolific approval rating between 90% and 100%. All recruitment and ELA grouping procedures were preregistered (https://osf.io/bt74x). Participants completed all sessions of the study online. They were paid $2 for a recruitment prescreen (Session 1; JVQ) and $15 plus a performance-based bonus for Session 2 (behavioral task and remaining ELA questionnaires). Participants were paid an additional $15 to complete an exploratory Session 3 to assess self-reported adulthood psychopathology (see Supplemental Material; Supplemental Fig. S2; Table S1). Participants gave informed consent before each study session, and the Harvard University Committee on the Use of Human Subjects approved the procedures.

We had full data sets (JVQ prescreen, ELA measures, and behavioral task) for 193 participants. Due to an unexpected change in recruitment procedures, oversampling was unavoidable (see Supplemental Material). Four participants’ data were not analyzed due to oversampling, and nine participants were excluded for inaccurate behavior in the task (not choosing the higher-value option for >51% of trials). This left us with our target total of 170 participants (102 women, 65 men, three other, ages 18–78, mean = 32.97) split into two equal groups (85 ELA, 85 no-ELA). The mean choice accuracy of this final sample was well above chance (73%).

Participants completed a previously validated reinforcement learning task that manipulates beliefs about agency (Dorfman et al. 2019), called the Agency-Modulated Feedback Learning task (AMFL). Participants were instructed to imagine they were mining for gold in the Wild West. On each trial, participants had to choose between one of two different-colored mines by clicking on a button underneath the mine of their choice (Fig. 1B, left). After making a choice, participants either received a reward of gold or a loss of rocks (Fig. 1B, right). Each mine in a pair produced a reward with either 70% or 30% probability. Each reward yielded a small amount of real bonus money (5 cents) and each loss resulted in a subtraction of real bonus money (5 cents). Bonuses were summed, revealed, and paid out at the end of the task. Participants completed three blocks of 50 trials (150 total trials) in different “mining territories” (Fig. 1A).

Figure 1.

Figure 1.

Open in a new tab

Behavioral task schematic. (A) Participants were instructed that different hidden agents frequent each territory: a bandit steals gold from the mines and replaces it with rocks (adversarial condition), a tycoon leaves extra gold in the mines (benevolent condition), and a sheriff will redistribute gold and rocks in the mines (neutral condition). Participants are told the hidden agent intervention condition (adversarial, benevolent, or neutral) at the start of each block. (B) Participants make a choice on a two-armed bandit (left). They then receive either positive (gold) or negative (rocks) feedback and judge whether or not they believe that the hidden agent intervened on the current trial. (Dorfman et al. 2019; reprinted by permission of SAGE Publications.)

Participants were instructed that different hidden agents frequent each territory: a bandit steals gold from the mines and replaces it with rocks (adversarial condition), a tycoon leaves extra gold in the mines (benevolent condition), and a sheriff will redistribute gold and rocks in the mines (neutral condition). Participants complete each of the three conditions in randomized order. Participants were told that the agent intervenes “sometimes.” They do not know whether the agent intervened on any particular trial. While the underlying reward distributions (i.e., absent intervention) for the mines were 70% or 30%, the hidden agent intervenes on 30% of trials (or 15 out of 50 trials). For example, the benevolent intervention produces rewards on 15 out of 50 trials, the adversarial intervention produces losses on 15 out of 50 trials, and the neutral intervention produces either losses or rewards for 15 out of 50 trials. After feedback on each trial, participants provided a subjective estimate of their agency beliefs and were asked whether they believed the outcome they received was a result of hidden agent intervention (binary response of “Yes” or “No”; Fig. 1B). This subjective trial-by-trial agency judgment was used to calculate average explicit agency beliefs for each participant (subsequently referred to as subjective agency).

We hypothesized that adults in our sample would be best fit by a Bayesian reinforcement learning model (or one of its variations), developed and used in prior work (Dorfman et al. 2019, 2021; Cohen et al. 2020). To test this hypothesis, we compared five variations of a novel Bayesian reinforcement learning model (Adaptive Bayesian, Fixed Bayesian, Empirical Bayesian; described below) to two variations of standard reinforcement learning models: a model in which we modeled separate, fixed learning rates for positive and negative outcomes (Two Learning Rate), and a model with separate, fixed learning rates for positive and negative outcomes in each of the three experimental conditions (Six Learning Rate). See Supplemental Table S2 for parameter descriptions for all comparison models. Below we briefly outline the intuition of the Bayesian reinforcement learning models. Further mathematical details can also be found in our prior work (Dorfman et al. 2019) and in the Supplemental Material.

To derive model parameters for subsequent model-based analyses, we fit our comparison models to the choice data and then extracted the relevant parameters from the winning model. Overall, the best-fitting model for the data is a variation of our Bayesian reinforcement learning model (Empirical Bayesian model; protected exceedance probability [PXP] = 0.74). Average learning rates from the winning model show the expected pattern of learning asymmetries (Supplemental Fig. S3) and learning trajectories for this model fit participants’ choice behavior (Supplemental Fig. S4).

The problem facing the participant during the AMFL task is to choose the option yielding the highest reward. Because they do not know the reward probabilities of the two options, they must estimate them from experience, while taking into account possible intervention from the hidden agent. The Bayesian models jointly infer the hidden agent interventions and the reward probabilities. After choosing an action and observing reward rt on trial t, the participant updates their estimate of the action's intrinsic reward probability θt according to a reinforcement learning equation that depends on inferences about latent causes: θt+1 = θt + αt(rt − θt), where αt is a learning rate. In the winning model (Empirical Bayesian), the prior probability of hidden-agent intervention is initially set at 30% to replicate the ground-truth intervention probability in the task. The learning rate then changes across trials depending on the posterior probability of the hidden agent intervention. Specifically, the learning rate and prediction error are scaled by a parameter that encodes the posterior probability of beliefs about agency over outcomes, ψ, as follows:

θt+1=θt+αt(rt−θt)⋅ψ (1)

In addition to participants’ task-based subjective agency beliefs, we can also use this agency belief parameter, ψ, to investigate the relationship between ELA and model-derived, latent agency (subsequently referred to as latent agency).

The learning rate (αt), which is modulated by latent agency beliefs, is derived analytically to be consistent with Bayesian updating (Dorfman et al. 2019). In particular, it depends on beliefs about the two possible sources of the outcome: the intrinsic reward distribution of the action and the intervention of the hidden agent. The posterior belief ψ encodes the degree to which the outcome should be attributed to the intrinsic reward distribution rather than to the hidden agent, which is affected by the experimental condition, whether feedback was positive or negative, as well as the trial history. Importantly, this latent agency (ψ) is based solely on choice behavior and probabilistic inference and is not dependent on the task-based subjective agency judgments.

Essentially, when the posterior probability is high, the learning rate is low. Intuitively, the model predicts that the participant should suspend learning about the reward probabilities when they believe that the outcome was generated by an external force. In other words, the learning rate in the Bayesian model is calculated by integrating one's cumulative, past beliefs about agency into one's value estimate of a particular choice. Critically, the learning rate will differ in magnitude for positive and negative outcomes depending on how much agency the participant believes they have. For example, when the agent is adversarial, positive outcomes can only be generated from the intrinsic reward probabilities, whereas negative outcomes can be generated by either the hidden agent or the intrinsic reward probabilities. Consequently, negative outcomes are less informative about the reward probabilities in this scenario, inducing a lower learning rate.

Considering the association between exposure to stressors and diminished agency (Bryant and Trockel 1976; Maier and Seligman 1976; Chorpita and Barlow 1998), as well as our prior work showing that diminished agency beliefs directly reduce learning rates (Dorfman et al. 2019, 2021; Cohen et al. 2020), we preregistered three primary hypotheses. First, we expected that participants in the ELA group would have lower subjective and latent agency beliefs compared to no-ELA participants. Because we predicted an association between ELA group and agency beliefs, we also expected that there might be a relationship between ELA severity and agency beliefs. Specifically, our second hypothesis was that participants with more severe adversity exposure would have decreased subjective and latent agency beliefs. Lastly, because we expected adversity and agency to be negatively correlated and model-derived latent agency beliefs modulate learning rates, we hypothesized that the ELA group would have lower learning rates (averaged across conditions and feedback type) compared to the no-ELA group.

To test our first hypothesis that participants in the ELA group would have lower agency beliefs, we calculated both subjective and latent measures of agency. The subjective measure of agency was calculated by taking the mean of each participants’ binary responses (“Yes” - 1/“No” - 0) to the question, “Did the [bandit/tycoon/sheriff] cause this outcome?” across all trials of the behavioral task. To calculate a latent agency variable, we took the mean of the trial-by-trial model parameter, ψ, that encodes the posterior probability of beliefs about agency over outcomes. Contrary to our hypothesis, there was no significant difference in either the mean subjective or latent agency between ELA groups, subjective: t(168) = 1.0824, P = 0.280, CI = [−0.0228, 0.0781]; latent: t(168) = 1.607, P = 0.1098, CI = [−0.000, 0.007].

To test our second hypothesis that adversity severity is associated with decreased agency beliefs, we calculated a continuous measure of adversity severity and correlated this measure with participants’ average subjective agency beliefs. Contrary to our hypothesis, we did not find a significant correlation between average subjective agency beliefs in the behavioral task and severity, rs = 0.134, P = 0.080. However, we did find that the latent agency belief parameter was negatively correlated with adversity severity, r(168) = −0.21, P = 0.007, suggesting that participants with more severe adversity exposure have decreased latent agency beliefs that are implicitly derived from their choices (Fig. 2). A linear regression approach to this analysis incorporating additional predictors for age and gender showed no significant effects of age/gender, P = 0.496/0.292, but a significant effect of adversity severity, F(3, 166) = 2.97, P = 0.013.

Figure 2.

Figure 2.

Open in a new tab

The model-derived probability of agency (ψ) is negatively correlated with continuous ELA severity.

To investigate our third hypothesis about the relationship between learning rate magnitudes and adversity exposure, we averaged each participant's trial-by-trial learning rates from the winning model (Empirical Bayesian) across all trials. Because these learning rate means were nonnormally distributed, we conducted a Mann–Whitney Wilcoxon rank sum test. In line with our preregistered hypothesis, average learning rates for participants in the ELA group were significantly lower than those for the no-ELA group, W = 4375, P = 0.018 (Fig. 3). We also looked at the relationship between the continuous measure of adversity severity and average learning rates, finding no correlation between ELA severity and the mean total learning rates (averaged across all trials), rs(168) = −0.051, P = 0.512.

Figure 3.

Figure 3.

Open in a new tab

Participants in the ELA group had significantly lower average learning rates from the winning model compared to participants without ELA exposure.

Together, we find that adversity severity impacts latent agency beliefs derived from our computational model. However, neither severity nor ELA is related to subjective (explicit) agency beliefs in our task. In addition, we demonstrate that ELA exposure is related to overall rates of learning, with participants in the ELA group showing dampened learning rates.

Given prior evidence suggesting a relationship between early-life stressors, learning, and controllability beliefs, the current study sought to investigate whether there is an association between ELA and agency-modulated reinforcement learning. To date, only a few human studies have used reinforcement learning paradigms and computational models to explore this association (Hanson et al. 2017; Kennedy et al. 2021; Sacu et al. 2024). To our knowledge, the current study is the first to formally investigate the relationship between ELA and reinforcement learning while also manipulating agency beliefs.

Prior research has shown an association between ELA and locus of control (LOC)—the tendency to believe that life and world events are the result of either internal or external causes (Rotter 1966; Culpin et al. 2015). Here, we instead aimed to determine whether ELA was associated with agency beliefs that are generated and updated through learning and causal inference processes, allowing us to subsequently explain how these agency beliefs dynamically shape optimistic and pessimistic learning. We hypothesized that adults who retrospectively reported severe exposure to ELA would demonstrate lower subjective and latent agency beliefs compared to adults who reported no exposure to early-life adversity.

We analyzed participants’ subjective beliefs about agency as reported in the behavioral task and found no evidence that these beliefs differed between ELA groups, contrary to prior findings using LOC. However, our task-based, trial-by-trial agency judgments likely differ from long-term beliefs measured by self-report scales in important ways. For example, short-term beliefs in the AMFL task are easy to alter (as evidenced by our current and prior work), while LOC measures have been shown to be very stable over long time courses (Hovenkamp-Hermelink et al. 2019). Future work should explore additional measures of subjective long-term and short-term agency beliefs to try to uncover whether and when these are independent.

In contrast to subjective agency beliefs, we also explored whether the latent agency parameter is correlated with continuous ELA severity and found a negative correlation, demonstrating that participants with more severe ELA have decreased latent agency beliefs. These disparate results suggest that participants’ subjective agency judgments might not reflect the same dynamic changes that the latent agency parameter can capture based on choice behavior. This could be due to a purely quantitative difference between the two agency measurements. Our task-based subjective judgments are binary (“yes”/“no” responses), but the latent beliefs are continuous (probability between 0 and 1), meaning that participants’ judgments in the task are necessarily constrained. We speculate that ELA, which is often both aversive and uncontrollable, is leading to these alterations in agency beliefs.

To examine learning, we first replicated modeling results from prior work (Dorfman et al. 2019, 2021; Cohen et al. 2020), and found that a version of our Empirical Bayesian model fits choice behavior best. The model also shows the expected pattern of learning rates given the task structure, where participants learn more from positive outcomes in the adversarial condition and more from negative outcomes in the benevolent condition (Supplemental Fig. S3). In addition, our results demonstrate that adults who were exposed to childhood adversity have significantly lower average learning rates than those who were exposed to little or no childhood adversity. These findings are in line with prior work showing that adults with adverse childhood experiences have lower learning rates on a patch foraging task (Lloyd et al. 2022), and adversity-exposed adolescents have decreased learning on an associative learning task compared to adolescents without adversity exposure (Hanson et al. 2017).

In summary, the current study found initial evidence that both agency and reinforcement learning are influenced by exposure to early-life adversity. The novel approach of our behavioral task and model allowed us to investigate the direct influence of agency on reinforcement learning and its association to ELA, as well as to test a potential mechanism for this association using modeling. However, our results suggest that this relationship is nuanced. There could be additional aspects of adversity that could influence agency-modulated learning, such as type of adversity (i.e., threat vs. deprivation; Sheridan and McLaughlin 2014) or perceptions of or attitudes toward adversity. Future work should incorporate additional measures of adversity and consider longitudinal approaches to help further elucidate the relationship between childhood stress, agency, and learning.

Data access

This study was preregistered: https://osf.io/bt74x. Model and task code are available at the following GitHub link: https://github.com/hayleydorfman/valence-control.

Supplementary Material

Supplement 1
Supplemental_Material.docx (322.2KB, docx)

Acknowledgments

This research was supported by the National Science Foundation (NSF SBE Postdoctoral Fellowship awarded to H.M.D.), the Oak Ridge Institute for Science and Education (ORISE), the U.S. Department of Energy, and the Office of the Director of National Intelligence (ODNI) (Intelligence Community Postdoctoral Research Fellowship awarded to H.M.D.), and the National Institute on Drug Abuse (Grant No. R01 DA042855 awarded to E.A.P.).

Author contributions: H.M.D., K.A.M., and E.A.P. contributed equally to conceptualization. H.M.D. and E.A.P. contributed to funding acquisition. H.M.D. served as lead for investigation, formal analysis, methodology, and writing—original draft. B.J.W.D. served in a supporting role for investigation and writing—review and editing. K.A.M. and E.A.P. served as lead for writing—review and editing. E.A.P. served as lead for project administration.

Footnotes

[Supplemental material is available for this article.]

References

  1. Abramson LY, Seligman ME, Teasdale JD. 1978. Learned helplessness in humans: critique and reformulation. J Abnorm Psychol 87: 49–74. 10.1037/0021-843X.87.1.49 [DOI] [PubMed] [Google Scholar]
  2. Bernstein DP, Ahluvalia T, Pogge D, Handelsman L. 1997. Validity of the Childhood Trauma Questionnaire in an adolescent psychiatric population. J Am Acad Child Adolesc Psychiatry 36: 340–348. 10.1097/00004583-199703000-00012 [DOI] [PubMed] [Google Scholar]
  3. Bifulco A, Brown GW, Harris TO. 1994. Childhood Experience of Care and Abuse (CECA): a retrospective interview measure. J Child Psychol Psychiatry Allied Discip 35: 1419–1435. 10.1111/j.1469-7610.1994.tb01284.x [DOI] [PubMed] [Google Scholar]
  4. Bifulco A, Bernazzani O, Moran PM, Jacobs C. 2005. The childhood experience of care and abuse questionnaire (CECA.Q): validation in a community series. Br J Clin Psychol 44: 563–581. 10.1348/014466505X35344 [DOI] [PubMed] [Google Scholar]
  5. Bryant BK, Trockel JF. 1976. Personal history for psychological stress related to locus of control orientation among college women. J Consult Clin Psychol 44: 266–271. 10.1037/0022-006X.44.2.266 [DOI] [PubMed] [Google Scholar]
  6. Chorpita BF, Barlow DH. 1998. The development of anxiety: the role of control in the early environment. Psychol Bull 124: 3–21. 10.1037/0033-2909.124.1.3 [DOI] [PubMed] [Google Scholar]
  7. Cohen AO, Nussenbaum K, Dorfman HM, Gershman SJ, Hartley CA. 2020. The rational use of causal inference to guide reinforcement learning strengthens with age. NPJ Sci Learn 5: 16. 10.1038/s41539-020-00075-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Culpin I, Stapinski L, Miles ÖB, Araya R, Joinson C. 2015. Exposure to socioeconomic adversity in early life and risk of depression at 18 years: the mediating role of locus of control. J Affect Disord 183: 269–278. 10.1016/j.jad.2015.05.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Dorfman HM, Bhui R, Hughes BL, Gershman SJ. 2019. Causal inference about good and bad outcomes. Psychol Sci 30: 516–525. 10.1177/0956797619828724 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dorfman HM, Tomov MS, Cheung B, Clarke D, Gershman SJ, Hughes BL. 2021. Causal inference gates corticostriatal learning. J Neurosci 41: 6892–6904. 10.1523/JNEUROSCI.2796-20.2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Finkelhor D, Hamby SL, Ormrod R, Turner H. 2005. The Juvenile Victimization Questionnaire: reliability, validity, and national norms. Child Abuse Negl 29: 383–412. 10.1016/j.chiabu.2004.11.001 [DOI] [PubMed] [Google Scholar]
  12. Glynn LM, Stern HS, Howland MA, Risbrough VB, Baker DG, Nievergelt CM, Baram TZ, Davis EP. 2019. Measuring novel antecedents of mental illness: the Questionnaire of Unpredictability in Childhood. Neuropsychopharmacology 44: 876–882. 10.1038/s41386-018-0280-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hanson JL, van den Bos W, Roeber BJ, Rudolph KD, Davidson RJ, Pollak SD. 2017. Early adversity and learning: implications for typical and atypical behavioral development. J Child Psychol Psychiatry Allied Discip 58: 770–778. 10.1111/jcpp.12694 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hovenkamp-Hermelink JHM, Jeronimus BF, van der Veen DC, Spinhoven P, Penninx BWJH, Schoevers RA, Riese H. 2019. Differential associations of locus of control with anxiety, depression and life-events: a five-wave, nine-year study to test stability and change. J Affect Disord 253: 26–34. 10.1016/j.jad.2019.04.005 [DOI] [PubMed] [Google Scholar]
  15. Kennedy BV, Hanson JL, Buser NJ, van den Bos W, Rudolph KD, Davidson RJ, Pollak SD. 2021. Accumbofrontal tract integrity is related to early life adversity and feedback learning. Neuropsychopharmacology 46: 2288–2294. 10.1038/s41386-021-01129-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Lambert HK, King KM, Monahan KC, McLaughlin KA. 2017. Differential associations of threat and deprivation with emotion regulation and cognitive control in adolescence. Dev Psychopathol 29: 929–940. 10.1017/S0954579416000584 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Lloyd A, McKay RT, Furl N. 2022. Individuals with adverse childhood experiences explore less and underweight reward feedback. Proc Natl Acad Sci 119: e2109373119. 10.1073/pnas.2109373119 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Maier SF, Seligman ME. 1976. Learned helplessness: theory and evidence. J Exp Psychol Gen 105: 3–46. 10.1037/0096-3445.105.1.3 [DOI] [Google Scholar]
  19. McLaughlin KA, DeCross SN, Jovanovic T, Tottenham N. 2019. Mechanisms linking childhood adversity with psychopathology: learning as an intervention target. Behav Res Ther 118: 101–109. 10.1016/j.brat.2019.04.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Raio CM, Lu BB, Grubb M, Shields GS, Slavich GM, Glimcher P. 2022. Cumulative lifetime stressor exposure assessed by the STRAIN predicts economic ambiguity aversion. Nat Commun 13: 1686. 10.1038/s41467-022-28530-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Rotter JB. 1966. Generalized expectancies for internal versus external control of reinforcement. Psychol Monogr 80: 1–28. 10.1037/h0092976 [DOI] [PubMed] [Google Scholar]
  22. Sacu S, Dubois M, Hezemans FH, Aggensteiner P-M, Monninger M, Brandeis D, Banaschewski T, Hauser TU, Holz NE. 2024. Early life adversities are associated with lower expected value signaling in the adult brain. Biol Psychiatry 10.1016/j.biopsych.2024.04.005 [DOI] [PubMed] [Google Scholar]
  23. Sheridan MA, McLaughlin KA. 2014. Dimensions of early experience and neural development: deprivation and threat. Trends Cogn Sci 18: 580–585. 10.1016/j.tics.2014.09.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Skinner EA, Zimmer-Gembeck MJ, Connell JP, Eccles JS, Wellborn JG. 1998. Individual differences and the development of perceived control. Monogr Soc Res Child Dev 63: i–vi. 10.2307/1166220 [DOI] [PubMed] [Google Scholar]
  25. Sumner JA, Colich NL, Uddin M, Armstrong D, McLaughlin KA. 2019. Early experiences of threat, but not deprivation, are associated with accelerated biological aging in children and adolescents. Biol Psychiatry 85: 268–278. 10.1016/j.biopsych.2018.09.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Tennen H, Drum PE, Gillen R, Stanton A. 1982. Learned helplessness and the detection of contingency: a direct test. J Pers 50: 426–442. 10.1111/j.1467-6494.1982.tb00228.x [DOI] [PubMed] [Google Scholar]
  27. Weathers FW, Blake DD, Schnurr PP, Kaloupek DG, Marx BP, Keane TM. 2013. The life events checklist for DSM-5 (LEC-5). Instrument available from the National Center for PTSD. www.ptsd.va.gov, 31. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1
Supplemental_Material.docx (322.2KB, docx)

Articles from Learning & Memory are provided here courtesy of Cold Spring Harbor Laboratory Press

RESOURCES