For a science of layered mechanisms: beyond laws, statistics, and correlations

Abstract

Two general claims are made in this work. First, we need several different layers of “theory,” in particular for understanding human behavior. These layers should concern: the cognitive (mental) representations and mechanisms; the neural underlying processes; the evolutionary history and adaptive functions of our cognition and behaviors; the emergent and complex social structures and dynamics, their relation and feedbacks on individual minds and behaviors, and the relationship between internal regulating goals and the external functions/roles of our conduct; the historical and cultural mechanisms shaping our minds and behaviors; the developmental paths. Second, we do not just need “predictions” and “laws” but also “explanations”; that is, we need to identify the mechanisms producing (here-and-now, or diachronically) a given phenomenon. “Laws” are not enough; they are simply descriptive and predictive; we need the “why” and “how.” Correlations are not enough (and they are frequently misleading). We need computational models of the processes postulated in our theories1.