Does the Multiple Realizability Argument Generalize?

Abstract The multiple realizability argument consists in inferring that mental properties are not physical properties from the premise that they admit a highly heterogeneous range of possible physical realizations. Depending on the degree of heterogeneity that is considered sufficient for genuine multiple realizability, this inference may extend to paradigmatically physical properties. In this paper, I will argue that a strong version of the argument, requiring an extreme degree of heterogeneity, succeeds in avoiding this generalization, whereas a weaker one extends its conclusion to paradigmatically physical properties such as temperature. Section 1 contains an analysis of the concept of multiple realizability, which is carried out through the presentation of the strong and the weak versions. The former requires different possible realizers of multiply realizable properties to possess highly heterogeneous causal powers, the latter posits the weaker requirement that they perform the functions associated with the property they realize in relevantly different ways. Instances of temperature share enough physical features that they do not count as multiply realizable according to the requirement associated with the strong version. Hence, if this requirement was met by mental properties, then the multiple realizability argument would be safe from generalizing to temperature. However, as I will argue in Section 2, it is implausible that this is the case, since it is doubtful that even properties significantly more physically heterogeneous than mental properties meet it. At the same time, as shown in Section 3, the weaker characterization of multiple realizability applies to temperature. Hence, such weaker version of the argument is subjected to the charge of generalization to temperature. In Section 4, I will discuss a potential objection to the assessment provided in the previous three sections, consisting in conceding that the argument generalizes, while maintaining that temperature is a physical property in a different sense from that involved in the argument. In Section 5, I conclude that the multiple realizability argument, as it stands, cannot be employed to maintain a nonreductivist account of mental properties without also committing to a nonreductivist account of properties like temperature, unless a suitably intermediate version of the argument is developed and the adoption of the constraint on which it is based independently justified.