Social plasticity is a ubiquitous feature of animal behaviour. Animals must adjust the expression of their social behaviour to the nuances of daily social life and to the transitions between life-history stages, and the ability to do so affects their Darwinian fitness. Here, an integrative framework is proposed for understanding the proximate mechanisms and ultimate consequences of social plasticity. According to this framework, social plasticity is achieved by rewiring or by biochemically switching nodes of the neural network underlying social behaviour in response to perceived social information. Therefore, at the molecular level, it depends on the social regulation of gene expression, so that different brain genomic and epigenetic states correspond to different behavioural responses and the switches between states are orchestrated by signalling pathways that interface the social environment and the genotype. At the evolutionary scale, social plasticity can be seen as an adaptive trait that can be under positive selection when changes in the environment outpace the rate of genetic evolutionary change. In cases when social plasticity is too costly or incomplete, behavioural consistency can emerge by directional selection that recruits gene modules corresponding to favoured behavioural states in that environment. As a result of this integrative approach, how knowledge of the proximate mechanisms underlying social plasticity is crucial to understanding its costs, limits and evolutionary consequences is shown, thereby highlighting the fact that proximate mechanisms contribute to the dynamics of selection. The role of teleosts as a premier model to study social plasticity is also highlighted, given the diversity and plasticity that this group exhibits in terms of social behaviour. Finally, the proposed integrative framework to social plasticity also illustrates how reciprocal causation analysis of biological phenomena (i.e. considering the interaction between proximate factors and evolutionary explanations) can be a more useful approach than the traditional proximate–ultimate dichotomy, according to which evolutionary processes can be understood without knowledge on proximate causes, thereby black-boxing developmental and physiological mechanisms.