What differential susceptibility theory explains
Differential susceptibility theory proposes that people vary in how strongly their development and functioning respond to environmental conditions, and that some of the individuals most adversely affected by difficult circumstances can also benefit especially strongly from supportive circumstances. The defining prediction is responsiveness in both directions. A child whose behavior changes sharply in the context of inconsistent, coercive caregiving may also show unusually large gains when interactions become predictable, warm, and responsive. Another child may be relatively less affected by either environment. The theory describes differences in the slopes of environmental response, not a fixed ranking of people as strong or weak. Its canonical formulation was developed by Belsky and Pluess (2009) and extended within evolutionary developmental theory by Ellis and colleagues (2011).
The scientific question is why a comparable exposure does not have identical consequences for everyone. An average association between parenting and child adjustment can conceal differences in magnitude, timing, and sometimes direction across individuals. Differential susceptibility supplies a specific interaction hypothesis: a moderator, such as a temperament characteristic, changes the association between environmental quality and an outcome, with greater responsiveness evident at adverse and beneficial ends of a sufficiently measured environmental range. This hypothesis has gained support in particular literatures, especially temperament-by-parenting studies, while evidence varies substantially across markers, outcomes, designs, and developmental periods (Slagt et al., 2016).
Environmental sensitivity is a wider family of questions about individual differences in registering, processing, and reacting to external influences. Differential susceptibility is one proposition within that family. The environmental influences at issue include caregiving, educational practices, social relationships, intervention exposure, neighborhood conditions, and stressors; each must be defined and measured rather than treated as an interchangeable expression of a single environmental quality. An interaction involving classroom feedback need not generalize to family adversity, psychotherapy, or socioeconomic conditions (Pluess, 2015).
The conceptual origin: from vulnerability to developmental plasticity
Why vulnerability models were incomplete
Traditional diathesis-stress models ask why some individuals develop difficulties under strain while others remain comparatively resilient. They place particular attention on a vulnerability characteristic and an unfavorable exposure. The framework is important for explaining risk, but by itself does not predict what occurs when the environment becomes notably supportive. Differential susceptibility extends the range of outcomes and exposures to ask whether a putative risk marker might also signal greater capacity to benefit. The same temperamental reactivity that predicts difficulty under harsh parenting could, under alternative conditions, predict an unusually favorable trajectory (Belsky and Pluess, 2009).
The central intellectual shift is from treating a marker as inherently disadvantageous to examining the shape of its context-dependent association with development. A response profile may be costly in one set of circumstances and advantageous in another. This remains an empirical question about specified exposures and specified outcomes. A theoretical label does not establish that a particular child will respond in both directions, and sensitivity itself does not imply superior performance, psychopathology, or immutable identity.
Evolutionary reasoning and diversity of response
Evolutionary accounts suggest that variation in developmental responsiveness could be maintained when environments are heterogeneous and future conditions cannot be perfectly predicted. Strategies associated with greater malleability may provide benefits in changing environments while carrying costs when adverse exposure is prolonged. Relatively less responsive strategies may be advantageous under some conditions. Such arguments explain why variation might be plausible; they cannot independently confirm the proposed developmental interaction in human data. Ellis et al. (2011) describe convergences between differential susceptibility and biological sensitivity to context while also addressing their distinct emphases.
Evolutionary explanation operates at a different level from a longitudinal estimate of parenting effects or an experimental test of an intervention. It is possible to observe a for-better-and-for-worse interaction without identifying its evolutionary origin, and possible to develop an evolutionary model without demonstrating the predicted interaction in a particular sample. The strongest research retains these levels of explanation separately.
Four competing models of environmental responsiveness
Diathesis-stress
In a diathesis-stress pattern, a vulnerability group has worse outcomes primarily when environmental conditions are unfavorable. As support increases, the groups converge or become similar; the hypothesis does not require a larger advantage for the vulnerable group under especially supportive conditions. For example, a temperament marker may predict greater adjustment difficulties during coercive family interactions while offering no special advantage under warm parenting. This interaction is meaningful even when differential susceptibility is unsupported (Roisman et al., 2012).
Differential susceptibility
In a differential-susceptibility pattern, the more responsive group shows worse outcomes under adverse conditions and better outcomes under supportive conditions. The idealized interaction crosses within an adequately observed portion of the exposure distribution. Both the negative and positive portions of the environment must be represented by credible measurements, and the modeled benefits and harms should be substantive rather than artifacts of extrapolation. Stronger responsiveness is a relationship between a moderator, environment, and outcome; it does not imply a globally heightened sensitivity to everything (Belsky, Pluess, and Widaman, 2013).
Vantage sensitivity
Vantage sensitivity focuses on differences in benefit from positive experiences. A group may respond particularly well to an enriching program or supportive relationship without showing greater vulnerability to adverse exposures. This may be the better description when data cover the benefits of an intervention but contain little information about adverse conditions. Pluess and Belsky (2013) developed this framework to separate variation in positive response from the stronger two-sided prediction of differential susceptibility.
General or weak responsiveness
A fourth observed pattern is a comparatively shallow response across the measured range, sometimes called lower environmental sensitivity. It must be understood relative to particular environmental indicators and outcomes. A child showing little response to one parenting questionnaire may respond strongly to peer rejection, discrimination, sleep disruption, or educational support. Statistical non-significance in a small study is not evidence that an individual is environmentally insensitive. The broader environmental sensitivity framework accommodates variation without requiring a binary split between sensitive and insensitive people (Pluess, 2015).
How an interaction is represented scientifically
Suppose researchers measure an environmental exposure E, a putative susceptibility marker S, and a developmental outcome Y. A simplified regression includes terms for E and S as well as their product E × S. The product term asks whether the association between E and Y differs across values of S. When E ranges from adverse to supportive conditions, a positive interaction may be consistent with stronger gains for the more susceptible group as conditions improve, but the meaning of its sign depends on coding. The interaction alone does not prove a crossover, a causal environmental effect, or the theoretical superiority of differential susceptibility over competing explanations.
A convincing test inspects predicted outcomes across the observed exposure distribution, the estimated crossover point, uncertainty around that point, and the regions where group differences are statistically distinguishable. Researchers also examine whether support for one response model depends on distant extrapolation, influential observations, measurement reliability, omitted confounders, or a selected analytic specification. Roisman and colleagues (2012) proposed explicit criteria for distinguishing differential susceptibility from diathesis-stress rather than relying on a visually appealing pair of crossing regression lines.
Competitive and confirmatory testing
Many early gene-by-environment analyses were exploratory: investigators tested an interaction and then interpreted its shape after seeing the result. A more informative approach preregisters plausible competing models, chooses the marker and environmental range in advance, specifies primary outcomes, and compares model fit. Belsky, Pluess, and Widaman (2013) described methods for directly comparing differential-susceptibility and diathesis-stress patterns. These techniques improve the clarity of hypothesis testing but cannot rescue weak exposure measures or nonrepresentative samples.
A hypothetical illustration
Imagine two groups of children measured across a validated range of caregiving quality. At the adverse end, children with high negative emotionality have relatively more behavior difficulties than children with lower negative emotionality. In moderately supportive conditions, the groups are similar. At the highly supportive end, the first group has fewer difficulties and greater competence than the second. This would match the idealized two-sided prediction, provided uncertainty intervals and the observed distribution support the crossover. If the groups merely converge at the supportive end, the result fits a vulnerability pattern more closely. If only the positive advantage appears, vantage sensitivity deserves consideration.
The illustration is conceptual rather than a claim about numerical effects in any particular study. Real outcomes rarely map onto a single perfectly ordered axis, and favorable development often includes multiple partly independent dimensions: mental health, academic performance, social competence, subjective well-being, and adaptive functioning. Evidence for one dimension does not automatically establish susceptibility across the others.
What counts as an environment?
Environmental exposure is multidimensional
The environment in developmental research includes the quality of repeated interactions, material resources, cultural expectations, chronic threats, and opportunities for learning. Studies vary in whether they assess supportive parenting, parental hostility, family conflict, attachment-related caregiving, school climate, intervention participation, or other exposures. A measure of conflict is not simply the negative pole of a measure of warmth; a family may experience both. Researchers therefore need to decide whether the theory concerns a single continuum or two related dimensions and must test the alternatives explicitly (Ellis et al., 2011).
Context is experienced as well as observed
An objectively similar event can be interpreted differently depending on developmental stage, prior learning, expectations, social position, and support. Nonetheless, defining environment only through the participant's subjective report risks overlap between the measured marker and the measured exposure. Someone who reports greater emotional distress may also describe everyday events as more adverse; common reporter effects can inflate statistical interaction patterns. Observations, repeated assessments, independent informants, and natural experiments help disentangle contextual exposure from the perception of context.
Developmental timing and dose
Exposure may differ in intensity, duration, predictability, and timing. A short experimentally administered feedback condition cannot be treated as equivalent to a decade of family relationships. Sensitive periods are theoretical and empirically testable possibilities, not assumptions established merely because participants were young. Longitudinal designs are useful for asking whether earlier temperament predicts differential responses to later environments, whether responsiveness itself changes, and whether accumulated adversity alters future response. Timing, duration, and cumulative exposure should be modeled where available.
Temperament as a candidate susceptibility marker
Temperament encompasses relatively early-emerging differences in emotional reactivity, activity, attention, and regulatory functioning. It is especially relevant because it can be measured before some subsequent environmental exposures or outcomes. Researchers have tested negative emotionality, difficult temperament, surgency, and effortful control as potential moderators. These constructs are overlapping but distinct, and each has its own measurement challenges and developmental interpretation (Slagt et al., 2016).
Difficult temperament and negative emotionality
A major meta-analysis by Slagt and colleagues (2016) synthesized 105 samples from 84 longitudinal studies on temperament-by-parenting interactions. Children with more difficult temperament displayed greater vulnerability in relation to negative parenting and greater benefit in relation to positive parenting, a pattern consistent with differential susceptibility. The findings were not uniform across temperament dimensions: support involving negative emotionality appeared when that characteristic was assessed in infancy, while surgency and effortful control did not consistently act as susceptibility moderators. The meta-analysis also found stronger moderation when parenting was observed rather than measured only through questionnaires.
These results give substantive support to one family of susceptibility claims while revealing why a general rule about highly emotional children would overstate the evidence. Difficult temperament instruments may combine multiple behavioral characteristics, and parent-rated temperament can be influenced by a caregiver's experiences with the child. Studies also vary in the outcomes and ages assessed. The relevant conclusion is conditional: some measured temperament-by-parenting interactions fit the predicted two-sided pattern, with effect variation tied to developmental period and measurement design.
Behavioral inhibition, approach, and regulatory control
Behavioral inhibition involves wariness and withdrawal in unfamiliar situations; effortful control concerns the regulation of attention and behavior. Neither characteristic is inherently a universal plasticity factor. Behavioral inhibition may be associated with anxiety-related outcomes under certain circumstances, but clinical risk and responsiveness to both adverse and beneficial environments are different questions. Similarly, self-regulation can predict adjustment directly without necessarily moderating environmental effects. Model-specific research must separate the moderator's main association with outcomes from an interaction with context.
For further treatment of foundational constructs and developmental assessment, see the published English scientific articles What Is Temperament in Psychology? and Infant Temperament: Reactivity, Self-Regulation and Early Development. Their conceptual roles complement the present article's focused test of environmental moderation.
Why genetics enter the theory
The original literature considered candidate genetic polymorphisms as possible markers of responsiveness, alongside temperament and physiological indices. In principle, genotypes are measured before later developmental outcomes and cannot be altered by postnatal parenting. This temporal feature can reduce some forms of reverse causation in the moderator. It does not eliminate population stratification, passive gene-environment correlation, complex ancestry effects, or measurement error in the environment. Nor does evidence of genetic association reveal a specific psychological or biological mechanism (Belsky and Pluess, 2009).
Early candidate-gene evidence
A meta-analysis of 22 randomized experiments involving 3,257 participants reported stronger intervention effects in groups classified by prespecified candidate genetic markers as more susceptible than in comparator groups (van IJzendoorn and Bakermans-Kranenburg, 2015). The authors treated those findings as evidence that response to experimentally manipulated environments can differ by genetic marker. Randomized intervention assignment strengthens inference concerning the intervention's effects and helps clarify environmental exposure. However, genetic subgroup analysis still depends on reliable markers, adequately powered interaction testing, appropriate multiple-testing control, and replication across distinct samples.
The candidate-gene replication problem
The broader candidate-gene-by-environment literature has a well-documented history of small samples, researcher flexibility, selective reporting, and results that did not consistently replicate. A systematic critical review by Duncan and Keller (2011) identified these problems in psychiatric research. Findings from a positive collection of early susceptibility experiments therefore do not establish that any one common variant reliably determines responsiveness across populations and environmental exposures. A named allele should never be presented as a clinical susceptibility test or an individual destiny.
Genome-wide and polygenic approaches avoid restricting investigation to one favored biological candidate, yet they introduce challenges of sample size, discovery-to-replication transfer, ancestry representation, outcome heterogeneity, and genetic effect interpretation. In one study of child anxiety treatment, Keers and colleagues (2016) reported a genome-wide approach to variation in treatment response. That research contributes a potential direction for understanding heterogeneity; its estimates and prediction procedures require independent evaluation before routine personalized use.
Heritability is a population statistic
A twin study of adolescents by Assary and colleagues (2021) suggested that variation in measured environmental sensitivity has multiple heritable components and genetic overlap with personality characteristics. A heritability estimate describes variation in a studied population under its conditions. It neither assigns a percentage of an individual's personality to genes nor sets a fixed upper bound on developmental change. Genetic influences can act through multiple biological and social pathways, and gene-environment correlations can complicate interpretation.
Biological sensitivity and stress-system hypotheses
Individual variation in stress-system responsiveness has been proposed as one possible mechanism linking environmental exposure with behavioral outcomes. Candidate measures include autonomic responses, hypothalamic-pituitary-adrenal activity, and physiological reactivity in defined tasks. These measures vary by task, time of day, previous exposure, developmental stage, and measurement protocol. One elevated physiological response cannot identify a generally susceptible person, and low reactivity does not establish resistance to environmental influences.
Ellis and colleagues (2011) compared differential susceptibility with biological sensitivity to context, identifying shared emphasis on two-sided context sensitivity and conceptual differences regarding developmental calibration. The adaptive calibration model of stress responsivity further proposes that stress-response patterns can develop in relation to environmental conditions. These are explanatory frameworks whose specific mechanisms and predictions require tests capable of distinguishing them from temperament and experience-based alternatives.
Mechanistic accounts must also separate biological reactivity from behavioral regulation. A person may detect environmental cues intensely while regulating their responses effectively. Another may show heightened physiological arousal in a laboratory but only modest differences in everyday social functioning. Multi-method repeated measurement is required before these components can be treated as parts of a verified causal pathway.
Sensory processing sensitivity and the limits of trait labels
Sensory processing sensitivity is a proposed individual-difference construct concerning deep processing, sensitivity to subtle stimuli, and emotional or sensory responsiveness. It overlaps with the broader discussion of environmental sensitivity, but scales associated with a highly sensitive person identity do not directly demonstrate for-better-and-for-worse developmental interactions. To validate a measure as a susceptibility marker, researchers must show that it predicts differential outcomes under independently measured beneficial and adverse environments, beyond direct trait-outcome correlations and reporting overlap (Pluess, 2015).
A person identifying as highly sensitive may experience important everyday differences in attention, sensory load, or emotional response. Such experiences can be examined seriously without converting the identity into a psychiatric diagnosis or assuming a universal biological subtype. For the construct's own evidence base and limitations, see Sensory Processing Sensitivity: Highly Sensitive Person Research and Limitations. Here its relevance is narrower: whether a marker predicts differential response to measured environmental conditions.
Evidence from parenting and family relationships
Parenting studies offer one of the most developed tests of differential susceptibility because they examine variation in family experience alongside prospective developmental outcomes. In the Slagt et al. (2016) meta-analysis, the consistent signal for difficult temperament was visible across several classes of adjustment, including internalizing and externalizing outcomes and dimensions of social or cognitive competence. The results also showed that the evidential pattern changes when a child attribute is measured at different developmental points and when parenting is assessed observationally.
Association should be interpreted carefully. A child can elicit responses from caregivers, parents and children share genetic and environmental characteristics, and family relationships change reciprocally. Repeated observational reports, covariate control, sibling comparisons, and interventions can clarify competing explanations, although each design has limitations. A moderating interaction between parent-rated temperament and parent-rated adjustment may partly reflect common measurement variance rather than a developmental response mechanism.
What controlled feedback experiments add
A within-subject experiment by Slagt and colleagues (2017) manipulated positive and negative evaluative feedback in a parenting-relevant context. Negative feedback influenced affect, but positive feedback did not produce robust effects on the investigated responses, and negative emotionality did not moderate the feedback effects as predicted. The study is informative precisely because it challenges an easy assumption that every reactive child should show an experimentally observable for-better-and-for-worse response.
A controlled feedback task permits a tighter causal interpretation of the feedback manipulation in that setting, while the short exposure differs from the sustained relational environment examined in longitudinal studies. Divergent findings can reflect differences in exposure scale, outcomes, developmental context, or a failure of a proposed universal moderator. Good synthesis therefore compares designs rather than assuming that positive longitudinal evidence invalidates experimental null findings or vice versa.
A transactional view of family life
Parents' behaviors change with children, and children's emerging characteristics can influence later parenting. That reciprocity creates complex pathways: temperament may moderate an environmental effect, predict exposure to certain interactions, or itself change through repeated experiences. The same dataset can be consistent with several mechanisms. Separating selection, evocation, direct effects, and true moderation requires explicit longitudinal models and, when possible, exogenous environmental variation. The independent topic of goodness of fit between child temperament and context is examined in Goodness of Fit: How Temperament Interacts with Parenting and Context.
Schools, peers, and broader developmental settings
Educational support, teacher responsiveness, peer inclusion, bullying, and classroom demands are plausible domains for differential response. A child who is especially affected by criticism might also be especially responsive to encouragement or stable classroom expectations. Yet evidence from parenting cannot be transferred to these settings without evaluating a specific marker-environment-outcome combination. Teacher reports and student self-reports can capture different aspects of classroom functioning; repeated observations and school-level measures help distinguish individual sensitivity from differences in actual exposure.
Cultural context further changes what counts as support, autonomy, predictability, and social success. A behavior regarded as adaptive in one social setting may have different meaning in another. Socioeconomic circumstances can affect both the range of experienced environments and the feasibility of supportive interventions. Samples that include only a narrow range of favorable classrooms may be unable to detect the adverse part of the hypothesized crossover. Likewise, a study conducted mainly under chronic adversity cannot test the predicted heightened benefit from genuinely enriching circumstances.
Psychological interventions and differential response
The intervention question is whether a specific participant characteristic predicts who benefits more from a given randomized program. Such moderation matters because mean treatment effects can mask heterogeneity in benefit. Random allocation of the intervention helps identify its causal effect, although a genetically or temperamentally defined susceptibility subgroup is rarely randomized. Large prespecified moderator analyses remain necessary, and apparent subgroup gains should be interpreted alongside confidence intervals, absolute changes, potential harms, adherence, and baseline severity.
The review by Bakermans-Kranenburg and van IJzendoorn (2015) and the related experimental meta-analysis by van IJzendoorn and Bakermans-Kranenburg (2015) make a substantive case for testing individual differences in intervention response. An intervention setting, however, generally investigates movement toward a beneficial environment. Without an independently measured adverse range, greater improvement among one subgroup may be evidence of vantage sensitivity or treatment-effect moderation rather than complete two-sided differential susceptibility.
Treatment prediction is an especially demanding application
In clinical trials, symptom change may depend on treatment type, therapist factors, participant preferences, previous experiences, expectancy, engagement, and the natural course of symptoms. A putative susceptibility marker must add reliable prospective prediction beyond readily available clinical features before it is useful for assigning care. A genetic interaction found in one trial does not justify withholding an evidence-based intervention from people lacking a marker. As a related illustration of interpretive complexity, large genomic research on childhood anxiety treatment outcomes has produced challenges for stable molecular prediction (Coleman et al., 2016).
Development across childhood, adolescence, and adulthood
Developmental change matters for both susceptibility markers and relevant environmental exposures. Infant negative emotionality may be assessed through repeated crying, distress, or wariness, whereas adolescent sensitivity may involve peer feedback, autonomy needs, identity development, and perceived social evaluation. Adult outcomes extend to workplace relationships, intimate partnerships, caregiving roles, and chronic stress. A measure validated for one age group may lose construct equivalence in another.
Temperament is an early developmental foundation for some aspects of personality; adult personality traits incorporate extended experiences, characteristic adaptations, and more elaborated self-regulation. A stable rank ordering on a trait does not imply that an individual's average level or context-response slope remains unchanged. Conversely, observed shifts in emotion across daily situations are personality states and do not by themselves establish durable trait change. These distinctions matter when interpreting long-term susceptibility claims and longitudinal research on personality development.
Current evidence is strongest in some childhood family settings. Generalization to adult occupational success, romantic relationship satisfaction, or global life outcomes remains an empirical extension that requires direct tests. The published Personality Development Across the Lifespan: Continuity and Change provides broader context for stability and change without treating differential susceptibility as the sole mechanism.
Between-person patterns and within-person change
Most moderation studies compare different people: those higher and lower on a marker show different associations between measured environment and outcomes. This between-person result cannot automatically establish that a particular person's behavior changes at the same rate when that person's own environment improves or deteriorates. Distinct mechanisms can generate the same between-person graph, including selection into environments and stable unmeasured differences.
Within-person designs repeatedly measure the same individual across situations or use experimental manipulation to estimate individual-specific response. Experience sampling, ecological momentary assessment, multiwave panel designs, and intensive time-series studies can help ask whether environmental responsiveness is stable, domain-specific, or contingent on changing internal states. Measurement schedules must match the duration of the process, and temporal precedence must be established before interpreting a response as causal. Slagt et al. (2017) illustrate both the promise and limitations of controlled within-person environmental variation.
Measuring susceptibility and its markers
Temperament and behavioral observation
Caregiver questionnaires provide access to repeated behavior across ordinary situations but may be affected by observer expectations, stress, and shared-method variance. Laboratory observation can assess responses under standardized conditions, yet a brief laboratory task may not reflect the environment a participant usually inhabits. Teacher, parent, clinician, and participant reports should be interpreted in relation to the construct and developmental stage rather than averaged indiscriminately. The quality of environmental exposure measures is equally consequential.
Physiological and genetic assessments
Measures of autonomic and endocrine activity are potential proxies for mechanisms, and genetic information may contribute to explanatory research. Their scientific value is evaluated through reliability, replicated interactions, population comparability, and incremental predictive validity. A biomarker that correlates with anxiety or arousal is not automatically a moderator of both beneficial and adverse exposures. The Assary et al. (2021) twin study supports investigating multiple heritable components while leaving the identification of practical individual response indicators as a distinct task.
Environmental sensitivity self-report scales
Questionnaires can measure how intensely a respondent describes noticing, interpreting, or reacting to environmental events. They do not directly measure the counterfactual question central to differential susceptibility: how would the same person function under substantially different circumstances? Validation therefore depends on prospective or experimental moderation studies, measurement invariance across populations, and observations of actual environmental variation. Self-report sensitivity scores should be treated as psychological measures with specified reliability and domain limits, not categorical diagnoses.
How to evaluate the strength of a study
A strong analysis begins with a theory-defined marker and a specified environmental exposure, followed by a priori predictions about outcomes. It samples the meaningful exposure range and uses measures with evidence of validity in the studied group. The analytic approach compares plausible alternative shapes, reports interaction estimates with uncertainty, tests sensitivity to assumptions, handles missing data and multiple comparisons, and makes a replication plan. A clear, large main effect of environment cannot substitute for an interaction test, because average effects and heterogeneity are different hypotheses (Roisman et al., 2012).
Why a crossover graph can mislead
An apparent crossover may lie outside the observed data; its location may depend on a few influential participants or on regression to the mean; differences at the supposedly favorable end may be statistically uncertain. Outcomes may have ceilings or floors, and subgroup definitions selected after inspection exaggerate contrasts. Testing multiple temperament facets, environments, and outcomes increases the chance of a nominally significant interaction. Confirmatory model comparisons and transparent sensitivity analyses are therefore important (Belsky, Pluess, and Widaman, 2013).
Statistical association, causation, and mediation
A prospective interaction establishes prediction under the measured conditions, not necessarily that changing parenting would reproduce the observed contrast. Randomized environmental interventions support causal inference about the assigned program, but mediation through stress biology or emotion regulation demands further evidence. A marker might be an indicator of a causal pathway, correlated with another actual moderator, or partly shaped by prior environments. Researchers must demonstrate that a proposed process changes before treating it as a mechanism.
Replication, contradictory evidence, and current scientific status
Established findings
Individual differences in temperament and responsiveness to context are substantial topics of developmental psychology. The theoretical possibility of environmental moderation is well grounded, and there is meta-analytic evidence that difficult temperament moderates associations between parenting and childhood adjustment in a pattern compatible with two-sided susceptibility (Slagt et al., 2016). Methodological research has clarified how to distinguish competing interaction shapes (Roisman et al., 2012). These findings support serious use of the theory as a research framework.
Findings that remain limited or contested
The precise set of biological markers that reliably identify broadly susceptible individuals remains uncertain. Single candidate gene interpretations confront the wider replication problems identified by Duncan and Keller (2011). Experimental evidence does not uniformly reproduce two-sided moderation: Slagt et al. (2017) found no predicted moderation by negative emotionality in a feedback experiment. One cannot infer identical sensitivity across all environments, outcomes, cultures, or ages from the strongest positive studies.
Some investigations identify greater benefit from supportive experiences without showing vulnerability to adverse experiences. Such findings are important, but vantage sensitivity may fit them better. Other investigations document risk-specific interactions resembling diathesis-stress. Science advances when a competing framework fits better, because the correct object of inquiry is the empirical response pattern. Reporting such alternatives strengthens rather than dilutes the theoretical test.
What newer genetics does and does not settle
Modern quantitative genetic work supports a multicomponent view of measured environmental sensitivity (Assary et al., 2021). It does not identify one susceptibility gene or prove that the observed temperament-by-parenting associations are caused by genetics. Genome-wide prediction requires sufficient power, independent validation, cross-ancestry evaluation, and demonstration of clinically or educationally meaningful value. A theoretical predisposition toward environmental responsiveness is best investigated as probabilistic, multidimensional, and potentially context-specific.
Culture, inequality, and sampling boundaries
Cross-cultural comparisons require more than translating a questionnaire. Parenting practices, emotional expression, school expectations, the meanings of interpersonal support, and the distribution of stressors vary. Measurement invariance should be evaluated before attributing score differences to different levels of susceptibility. Researchers should also examine whether social conditions produce unequal access to beneficial environments, because absence of observed benefit may reflect restricted opportunity rather than low capacity to respond.
Economic adversity, discrimination, displacement, and chronic insecurity can constrain environmental variation and compound outcomes. The theory cannot be used to imply that vulnerable individuals are responsible for overcoming structural conditions through sensitivity management. Environmental supports are material and relational as well as psychological. Group-level interactions should guide better hypotheses and inclusive program design, with careful evaluation in the actual populations served.
Clinical boundaries and ethics
Differential susceptibility is a research theory about heterogeneous context responsiveness, not a psychiatric disorder or diagnostic category. Negative emotionality, shyness, physiological reactivity, sensory processing sensitivity, and scores on susceptibility questionnaires are not interchangeable with symptoms or clinical diagnoses. Clinical assessment depends on established criteria for specific disorders, evidence of distress or impairment where relevant, developmental context, and appropriate professional evaluation. A child who reacts strongly to stressful interactions should not be labeled clinically disordered because of that reaction alone.
Risk interpretation also requires care. The framework should never be used to declare that particular children are biologically destined to fail under adversity, that families determine outcomes with certainty, or that an unreplicated genetic marker predicts treatment success. Ethical applications emphasize adaptable support, choice, protection from harmful exposures, respect for diversity, privacy in genetic research, and avoidance of labeling based on preliminary individual prediction.
Practical meaning for caregiving, education, and interventions
The most useful practical implication is a research-based commitment to examine variation in response while maintaining supportive conditions as a general goal. Warmth, predictability, respectful communication, and accessible learning opportunities can be evaluated for differential benefit without withholding them from people presumed less sensitive. An intervention can be effective on average while also varying in effects across individuals; detecting that heterogeneity requires rigorous outcome measurement and replication. Bakermans-Kranenburg and van IJzendoorn (2015) explain why some subgroup-specific gains can be hidden by averages.
For parents and educators, the theory encourages observation of how a particular person reacts to different circumstances over time. It also encourages adjustment of demands, support, and feedback according to demonstrated needs rather than fixed temperament labels. For researchers and policymakers, it motivates testing treatment moderators, evaluating equitable access to enrichment, and studying how exposure duration and developmental stage affect response. These implications are hypotheses and principles for careful application, not guarantees of individualized outcomes.
Frequently asked questions
Is differential susceptibility the same as being a highly sensitive person?
No single self-description establishes the pattern. Differential susceptibility predicts greater response to both unfavorable and favorable environments for a specified outcome. A sensory sensitivity questionnaire describes reported sensory or emotional tendencies; it requires external validation as a moderator of actual experiences. The constructs may overlap in some populations, but their measures and evidential claims differ (Pluess, 2015).
Are highly susceptible children more likely to develop mental health problems?
Their risk may be elevated under particular adverse exposures, depending on the marker and outcome. The theory simultaneously predicts a possibility of greater benefit under supportive circumstances. A sensitivity characteristic is not itself a diagnosis, and an association at the group level cannot reliably predict an individual child's course. Meta-analytic support is strongest for particular difficult-temperament and parenting combinations rather than every temperament characteristic (Slagt et al., 2016).
Can a person become more or less environmentally susceptible?
Responsiveness can depend on developmental stage, prior experiences, physiological states, and the environment under study. Some markers show relative stability, yet the specific stability of a person's environmental-response slope requires repeated measures across meaningful exposure variation. Heritability estimates do not rule out change, and current evidence does not support treating a single susceptibility score as a lifelong invariant property (Assary et al., 2021).
Is differential susceptibility proven by a significant interaction?
A significant interaction says that an estimated association differs across values of a moderator. Researchers must additionally test whether the observed shape fits a two-sided crossover, a disadvantage-only effect, an advantage-only effect, or some other pattern. Confidence intervals, data-range coverage, alternative specifications, and replication are central to that judgment (Roisman et al., 2012).
Does differential susceptibility mean some people do not need supportive environments?
No. Lower estimated moderation in a particular study does not imply that supportive conditions are unimportant for a person. A small differential effect, a weak measurement, and an unavailable or unmeasured outcome are all possible explanations. Ethical practice supports universally respectful, safe conditions while studying whether specific additional supports produce different gains.
Can a genetic test identify who will benefit from therapy?
Research on genetic moderation of interventions remains methodologically demanding and has not established a general-purpose individual test for treatment assignment. Early positive experimental syntheses coexist with serious replication problems in candidate-gene research. Clinical decisions require validated predictive performance for the relevant treatment, population, and outcome; an unreplicated polymorphism or polygenic indicator does not meet that standard (van IJzendoorn and Bakermans-Kranenburg, 2015; Duncan and Keller, 2011).
Integrative conclusion
Differential susceptibility theory establishes a precise research question: under which conditions does a measurable individual characteristic predict larger effects of both adversity and support? Its strongest intellectual contribution is the requirement to examine developmental plasticity across environmental quality, rather than infer an inherent disadvantage from a characteristic associated with poor outcomes under adversity. Longitudinal temperament-by-parenting syntheses give meaningful, differentiated support, and controlled interventions provide important opportunities for stronger causal tests. Experimental counterfindings, genetic replication failures, and methodological debates show where the evidence remains conditional. The scientific task is to specify the people, environments, outcomes, developmental periods, and causal assumptions for every claim (Belsky and Pluess, 2009; Slagt et al., 2016; Roisman et al., 2012).
Related Articles
Cluster Personality Traits, Temperament and Individual Differences — The research chapter connecting temperament, context, personality development, and individual-difference methods.
What Is Temperament in Psychology? Definition, Models and Development — Early-emerging reactivity and regulation as developmental characteristics.
Goodness of Fit: How Temperament Interacts with Parenting and Context — The compatibility of child characteristics and the demands or supports of an environment.
Infant Temperament: Reactivity, Self-Regulation and Early Development — Age-specific temperament indicators and developmental measurement.
Sensory Processing Sensitivity: Highly Sensitive Person Research and Limitations — A related sensitivity construct and the evidence surrounding it.
Personality Development Across the Lifespan: Continuity and Change — Longitudinal continuity and developmental change in personality.
References
Assary, E., Zavos, H. M. S., Krapohl, E., Keers, R., and Pluess, M. (2021). Genetic architecture of environmental sensitivity reflects multiple heritable components: A twin study with adolescents. Molecular Psychiatry, 26, 4896–4904. https://doi.org/10.1038/s41380-020-0783-8
Bakermans-Kranenburg, M. J., and van IJzendoorn, M. H. (2015). The hidden efficacy of interventions: Gene × environment experiments from a differential susceptibility perspective. Annual Review of Psychology, 66, 381–409. https://doi.org/10.1146/annurev-psych-010814-015407
Belsky, J., and Pluess, M. (2009). Beyond diathesis-stress: Differential susceptibility to environmental influences. Psychological Bulletin, 135, 885–908. https://doi.org/10.1037/a0017376
Belsky, J., Pluess, M., and Widaman, K. F. (2013). Confirmatory and competitive evaluation of alternative gene-environment interaction hypotheses. Journal of Child Psychology and Psychiatry, 54, 1135–1143. https://doi.org/10.1111/jcpp.12075
Coleman, J. R. I., et al. (2016). Genome-wide association study of response to cognitive–behavioural therapy in children with anxiety disorders. British Journal of Psychiatry, 209, 236–243. https://doi.org/10.1192/bjp.bp.115.168229
Del Giudice, M., Ellis, B. J., and Shirtcliff, E. A. (2011). The adaptive calibration model of stress responsivity. Neuroscience and Biobehavioral Reviews, 35, 1562–1592. https://doi.org/10.1016/j.neubiorev.2010.11.007
Duncan, L. E., and Keller, M. C. (2011). A critical review of the first 10 years of candidate gene-by-environment interaction research in psychiatry. American Journal of Psychiatry, 168, 1041–1049. https://doi.org/10.1176/appi.ajp.2011.11020191
Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J., and van IJzendoorn, M. H. (2011). Differential susceptibility to the environment: An evolutionary-neurodevelopmental theory. Development and Psychopathology, 23, 7–28. https://doi.org/10.1017/S0954579410000611
Hartman, S., and Belsky, J. (2016). An evolutionary perspective on family studies: Differential susceptibility to environmental influences. Family Process, 55, 700–712. https://doi.org/10.1111/famp.12161
Keers, R., et al. (2016). A genome-wide test of the differential susceptibility hypothesis reveals a genetic predictor of differential response to psychological treatments for child anxiety disorders. Psychotherapy and Psychosomatics, 85, 146–158. https://doi.org/10.1159/000444023
Pluess, M. (2015). Individual differences in environmental sensitivity. Child Development Perspectives, 9, 138–143. https://doi.org/10.1111/cdep.12120
Pluess, M., and Belsky, J. (2013). Vantage sensitivity: Individual differences in response to positive experiences. Psychological Bulletin, 139, 901–916. https://doi.org/10.1037/a0030196
Roisman, G. I., Newman, D. A., Fraley, R. C., Haltigan, J. D., Groh, A. M., and Haydon, K. C. (2012). Distinguishing differential susceptibility from diathesis-stress: Recommendations for evaluating interaction effects. Development and Psychopathology, 24, 389–409. https://doi.org/10.1017/S0954579412000065
Slagt, M., Dubas, J. S., Deković, M., and van Aken, M. A. G. (2016). Differences in sensitivity to parenting depending on child temperament: A meta-analysis. Psychological Bulletin, 142, 1068–1110. https://doi.org/10.1037/bul0000061
Slagt, M., Dubas, J. S., van Aken, M. A. G., Ellis, B. J., and Deković, M. (2017). Children's differential susceptibility to parenting: An experimental test of for better and for worse. Journal of Experimental Child Psychology, 154, 78–97. https://doi.org/10.1016/j.jecp.2016.10.004
van IJzendoorn, M. H., and Bakermans-Kranenburg, M. J. (2015). Genetic differential susceptibility on trial: Meta-analytic support from randomized controlled experiments. Development and Psychopathology, 27, 151–162. https://doi.org/10.1017/S0954579414001369