For more than a decade, the selected effects theory of function has been the favorite theory of function among philosophers of mind and science. According to this theory, it is the function of a certain trait of a certain type (e.g. a heart) to do what traits of that type (hearts) were selected for in the past. One of the main reasons why this theory is attractive is its promise to provide a naturalistic account of the normative aspects of functions. Intuitively, an item can have a function without being able to perform that function. A function seems something an item is supposed to do; and the item malfunctions if it does not do what it is supposed to do. According to the selected effects theory of function, these norms are due to the selective character of the process by means of which these items are generated and maintained. As it is the past success of a type that determines a token's function, an item can have a function without being able to perform it.

Paul Davies' Norms of Nature aims to hit the selected effects theory right at its heart. Davies argues that the idea that there are norms of nature does not fit well with a naturalistic point of view and that a naturalist should reject the intuition that natural traits can malfunction. As an alternative, Davies proposes a version of Cummins' (Reference Cummins1975) systemic capacity theory of function. According to Cummins, function attributions are tied to the analytic strategy of explanation. Analytic explanations explain a capacity of a complex system as the result of the coordinated operation of parts. The functions of a part are the capacities of that part that figure in an analytic explanation. A common objection against Cummins' theory maintains that the theory allows the attribution of functions where in fact no functions exists. Davies rightly argues (as did Amundson and Lauder (Reference Amundson and Lauder1994) and Wouters (Reference Wouters1999)) that this objection is often stated but never argued for, and that Cummins' constraints on the application of analytic explanations contains enough resources to answer the worry. Cummins framed his theory entirely in explanatory terms. Davies argues that the restrictions on the application of the theory are better framed in terms of the kinds of systems to which the theory applies, a move akin to that of Craver (Reference Craver2001).

Another reason why this book deserves attention is Davies' discussion of what to expect from a naturalistic theory. Davies distinguishes two levels. The first is the level of the individual function attributions. A naturalistic theory should make the attribution of functions an empirical matter; it should not render the attribution of functions immune to test or difficult to substantiate. As several authors before him (Amundson and Lauder (Reference Amundson and Lauder1994), Wouters (Reference Wouters1999)) have argued, the selected effects theory scores bad on this point.

The second level concerns the methodology and the ontology of the theory as a whole. Both should be in agreement with scientific practice. Davies takes issue with the prevailing view that a naturalistic theory of function should explicate criteria of application employed by working biologists. According to him, the aim of a naturalistic theory is to explain how an apparent powerful concept is best understood within a broad domain of inquiry. Biologists are not especially qualified for this task. For that reason it makes no sense to look at what they have in mind when they attribute functions. Instead, philosophers should study how researchers use function attributions to further inquiry. Davies argues that function attributions have a role both in historical and mechanistic explanation. Another important role, emphasized by Davies, is their role in guiding inquiry. Finally, whereas the intuition that functions have normative properties should be rejected, function attributions do have important normative and forward looking aspects: they express expectations about what an item will do in the future.

With regard to ontology, Davies argues that the abstract, causally inefficacious norms postulated by the selected effects theory are explanatory superfluous and beyond scientific test, and should, therefore, be rejected by naturalistic philosophers.

What remains is the illusion that functions are normative. Davies speculates that this illusion arises out of our understanding of the functional interdependencies between the parts of a self-reproducing system. He also argues that historical theories of function do not succeed in accounting for malfunction. The latter argument assumes that types are individuated in functional terms. This way of individuating types is advocated by several prominent selected effect theorists (most notable Millikan and Neander). However, if types are individuated in terms of homology rather than function, Davies' argument does not work.

In summary, I found this book particularly interesting because of its discussion of naturalistic methodology and its rejection of the intuition that natural traits can malfunction. Hence, I recommend this book to anyone interested in a naturalistic theory of function and to anyone interested in the question how to evaluate a naturalistic theory of a certain concept.