The differential susceptibility theory proposed by Jay Belsky is another interpretation of psychological findings that are usually discussed according to the diathesis-stress model. Both models suggest that people's development and emotional affect are differentially affected by experiences or qualities of the environment. Where the Diathesis-stress model suggests a group that is sensitive to negative environments only, the differential susceptibility hypothesis suggests a group that is sensitive to both negative and positive environments. A third model, the vantage-sensitivity model, suggests a group that is sensitive to positive environments only. All three models may be considered complementary, and have been combined into a general environmental sensitivity framework.

The idea that individuals vary in their sensitivity to their environment was historically framed in diathesis-stress or dual-risk terms. These theories suggested that some "vulnerable" individuals, due to their biological, temperamental and/or physiological characteristics (i.e., "diathesis" or "risk 1"), are more vulnerable to the adverse effects of negative experiences (i.e., "stress" or "risk 2"), while other "resilient" individuals are not affected by these negative experiences (see Figure 1). The differential susceptibility hypothesis and the related notion of biological sensitivity to context suggested that individuals thought to be "vulnerable" are not only sensitive to negative environments, but also to positive environments (see Figure 2). Thus, according to the differential susceptibility hypothesis, some individuals are "susceptible" or "plastic", in that they are more influenced than others by environmental influences in a "for better and for worse" manner.

Belsky suggests that evolution might select for some children who are more plastic, and others who are more fixed in the face of, for example, parenting styles.

Belsky offers that ancestral parents, just like parents today, could not have known (consciously or unconsciously) which childrearing practices would prove most successful in promoting the reproductive fitness of offspring—and thus their own inclusive fitness. As a result, and as a fitness optimizing strategy involving bet hedging, natural selection might have shaped parents to bear children varying in plasticity. This way, if an effect of parenting had proven counterproductive in fitness terms, those children not affected by parenting would not have incurred the cost of developing in ways that ultimately proved "misguided".

Importantly, natural selection might favour genetic lines with both plastic and fixed developmental and affective patterns. In other words, there is value to having both kinds at once. In light of inclusive-fitness considerations, children who were less malleable (and more fixed) would have "resistance" to parental influence. This could be adaptable some times, and maladaptive other times. Their fixedness would not only have benefited themselves directly, but even their more malleable siblings indirectly. This is because siblings, like parents and children, have 50% of their genes in common. By the same token, had parenting influenced children in ways that enhanced fitness, then not only would more plastic offspring have benefited directly by following parental leads, but so, too, would their parents and even their less malleable siblings who did not benefit from the parenting they received, again for inclusive-fitness reasons. The overall effect may be to temper some of the variability in parenting. That is, to make more conservative bets.

This line of evolutionary argument leads to the prediction that children should vary in their susceptibility to parental rearing and perhaps to environmental influences more generally. As it turns out, a long line of developmental inquiry, informed by a "transactional" perspective, has more or less been based on this unstated assumption.

Belsky, Bakermans-Kranenburg, & Van IJzendoorn, (2007) delineated a series of empirical requirements—or steps—for evidencing the differential susceptibility hypothesis. Particularly they identify tests that distinguish differential susceptibility from other interaction effects including diathesis-stress/dual-risk.

While diathesis-stress/dual-risk arises when the most vulnerable are disproportionately affected in an adverse manner by a negative environment but do not also benefit disproportionately from positive environmental conditions, differential susceptibility is characterized by a cross-over interaction: the susceptible individuals are disproportionately affected by both negative and positive experiences. A further criterion that needs to be fulfilled to distinguish differential susceptibility from diathesis-stress/dual-risk is the independence of the outcome measure from the susceptibility factor: if the susceptibility factor and the outcome are related, diathesis-stress/dual-risk is suggested rather than differential susceptibility. Further, environment and susceptibility factor must also be unrelated to exclude the alternative explanation that susceptibility merely represents a function of the environment. The specificity of the differential-susceptibility effect is demonstrated if the model is not replicated when other susceptibility factors (i.e., moderators) and outcomes are used. Finally, the slope for the susceptible subgroup should be significantly different from zero and at the same time significantly steeper than the slope for the non- (or less-) susceptible subgroup.