In 1970, R. C. Lewontin provided a formulation according to which the theory of natural selection is composed of a set of general principles that apply to entities at different levels in the biological hierarchy. Lewontin's principles of phenotypic variation, differential fitness, and heritability apply to any entities capable of evolving by natural selection.
Such formulation reveals the remarkable fact that the theory of natural selection does not privilege any particular level of biological organization. Indeed, we now have a theory of ‘multi-level selection’, part of which has been developed using the Price equation. Evolution and the Levels of Selection is a book on the conceptual, methodological, and metaphysical foundations of natural selection as a ‘multi-level’ theory, with the Price formalism as a foundational core.
In Chapter 1, “Natural Selection in the Abstract,” Okasha reviews the different general characterizations of natural selection, including the Hull-Dawkins replicator-interactor distinction. He opts for a Lewontin-type formulation, and makes a distinction, to be used in a later chapter, between ‘synchronic’ and ‘diachronic’ approaches to the unit levels of selection problem. Diachronically viewed, units of selection and their levels are themselves evolutionary products. The Price equation is introduced, both in terms of mathematical derivation and interpretation. It is stressed that no particular biological mechanisms enter in it, nor there is any fixed meaning in the character variable involved. It applies both to discrete and continuous settings. Evolution is thus defined as change in the relative frequency of types in a population, be it genes, individuals, or groups. How the statistical decomposition of evolutionary change embodied in Price's equation is to be related to the causal decomposition of evolutionary change is a central issue. Price's equation provides a decomposition of evolutionary change in terms of the change attributed to the statistical association of a given character value with fitness, and change attributed to transmission bias. Yet, statistical correlation does not lead to causation. Whether the causal stipulations turn out to be true is an empirical question. Okasha shows how the causal decomposition can include random drift broadly understood, and ends the chapter with the Lewontin-type formulation framed in terms of the Price equation.
In Chapter 2, “Selection at Multiple Levels,” a conceptual model for the analysis of multilevel selection is presented and the notion of a biological hierarchy is discussed. In particular, it is shown that “the problem of characterizing the biological hierarchy cannot be divorced from the problem from the problem of characterizing multilevel selection; they are interdependent” (45). How hierarchical organization arises in the course of evolutionary transitions and what sort of structural relations it generates is a major question explored. The conceptual model is of a collective with nested particles in it. The problem turns on whether one is tracking evolutionary change at the ‘particle’ or the ‘collective’ levels. In understanding multilevel selection, the “key issue is whether the particles and the collectives (or both) constitute the focal level” (56). The distinction made in the literature between two types of multilevel selection MLS1 (particle types are the focal units) and MLS2 (particles and collective types are both focal units of analysis) is reconsidered under the conceptual model presented and a multilevel formulation of the Price equation is presented.
In Chapter 3, “Causality and Multi-Level Selection,” the task is to understand how selection works at ‘crossed’ causal levels under an explicit distinction between causation and correlation. The levels of selection is thus posed: “when is a character-fitness covariance indicative of direct selection at the level in question, and when is it a byproduct of direct selection at another hierarchical level?” (78–79, italics in the original). Indeed, the production of cross-level by-products is a defining characteristic of the causal action of multilevel selection: whether a given evolutionary outcome is a by-product or not of selection acting at another level is to be discriminated. The patterns of MLS1 and MLS2 are reviewed under this light, and the scope of the Price equation is assessed and compared with the technique of contextual analysis. Cross-level by-products are analysed in the ‘upwards direction’, from lower to higher hierarchical levels, and also in the ‘downwards direction’, from higher to lower levels, as it would be the case of species selection.
The framework developed in Chapters 1–3 is applied in Chapters 4–6, “Philosophical Issues in the Levels of Selection Question,” “The Gene's Eye View and Its Discontents,” and “The Group Selection Controversy.” Concepts like emergence, causal screening off, causality and pluralism, and reductionism are revisited thoroughly all along.
Chapter 6 opens with a historical account of the ‘group selection controversy’. The notion of group selection is analyzed using the MLS1-MLS2 distinction. Okasha shows how some of the notions of kin selection, reciprocal altruism, and evolutionarily stable strategies, associate with the history of the group selection debates. An interesting upshot of the discussion is that even if the term group selection, as used by Wynne-Edwards, for example, referred to MLS2, models of group selection, such as the trait group model, really describe MLS1 scenarios. Although Okasha's emphasis is on selection rather than adaptation, his discussion shows how the notion of ‘group adaptation’ is a difficult one to adjudicate. A central methodological question in the whole chapter addresses how a group is to be characterized as a unit of analysis, and the assumptions of some of the main approaches are compared and assessed.
In Chapter 7, “Species Selection, Clade Selection, and Macroevolution,” the causal structure of species selection is re-examined, and the shortcomings of the concept of emergence as an notion to account for the higher-level character of species selection are highlighted. The study of character-fitness covariance as a by-product of selection at a lower or higher level is independent of whether the character is ‘emergent’ in any specified sense. The central problem is to determine whether lower-level selection produces higher-level character-fitness covariance, over any other processes that might be involved. Regarding clade selection, Okasha puts into question the existence of genuine clade-level effects on species fitness.
In Chapter 8, “Levels of Selection and the Major Evolutionary Transitions,” the diachronic approach to the level of selection question according to which the biological hierarchy has itself evolved takes center stage. Okasha reviews the works of Buss, Maynard Smith and Szathmary, and Michod, distinguishing between ‘genic’ and ‘hierarchical’ approaches in the explanation of evolutionary transitions, and puts his multilevel framework to work on how the latter take place.
In Okasha's account, the early stages of a transition can be interpreted as a MSL1 scenario: the particles begin to interact cooperatively. MSL2 is relevant when a process of collective reproduction evolves, and fitness can be consistently defined at that level. Okasha argues that during the transition, collective fitness ceases to be defined as average particle fitness but is proportional to it. The transition is completed when collective fitness is decoupled from particle fitness; it is not defined in terms of the latter nor is it proportional to it.
I hope this sampling of ideas has been sufficient to show how Evolution and the Levels of Selection has to become a must for philosophers and biologists working on the foundations of evolutionary theory.
