“If I am right, one of the important roles of theory-constitutive metaphors is to accomplish nondefinitional reference fixing…” (493).

Richard Boyd wants to apply Black’s interaction view of metaphor to what he calls ‘theory-constitutive metaphors’ within the realm of science. Black argues that when we use metaphors what we are really doing is applying a sort of filter. Take the trite metaphor, “Time flies” as an example. Black’s view entails that the connotations (what Black calls a complex of associated commonplaces) of ‘flying’ and ‘time’ filter through each other. So, we begin to think of time as a phenomenon of movement, etc, etc. Similarly, Black argues that our conception of flying is also subtly altered: the process can (but needn’t) go both ways.

Boyd applies this to the process of theory construction in scientific practice. Take the example of the metaphor ‘a computer virus’ from computer science. Rather than simply a ‘helpful’ metaphor that can be explained in other, literal terms, in the early stages of development, the term ‘computer virus’ (with its associated commonplaces taken from the realm of medicine and virology) actually allows us to think in certain ways conducive to the further development of the metaphor, and, consequently, the metaphor’s ‘paraphrase’ (its literal definition). Boyd notes that, “the use of theory-constitutive metaphors encourages the discovery of new features of the primary and secondary subjects, and new understanding of theoretically relevant respects of similarity, or analogy, between them” (489).

Boyd wants to add that within the scientific realm, metaphors do not lose their potency after continual use. Unlike literary metaphors which become tired and trite through repetition, theory constitutive metaphors become more useful the more they are used: “Theory-constitutive metaphors …become, when they are successful, the property of the entire scientific community, and variations on them are explored by hundreds of scientific authors without their interactive quality being lost” (Boyd 487). So, Boyd argues, rather than simply providing a sort of one-time insight that becomes obvious after use (“Juliet is the sun”, etc.) scientific metaphors provide us with heuristics; successful metaphors guide our research programs.

This leads us to what Boyd calls “nondefinitional reference fixing”. I take this to mean a process through which one term, is made clearly distinct from another, similar, term without recourse to a definition. So rather than saying A is the following, and making a comprehensive list of what A ‘is’ to differentiate it from B, nondefinitional reference fixing implies that we go about it in a quite different way. Boyd suggest that theory-constitutive metaphors act in this way. He writes that

The rationale for ostensive introduction of general terms is to permit reference to kinds whose essential properties may not yet be known—and thus to accommodate linguistic categories to as yet only partially understood features of the world. It is thus typically impossible that the differences between the essential properties of such co-occurring kinds should be marked, either in actual reference fixing or in idealized models of dubbing, by entirely accurate and complete descriptions of their respective essential properties (493).
Thus, the computer virus metaphor allows us to distinguish a ‘virus’ from a ‘worm’ without recourse to a clear, complete definition (one that is not even available). Because both these phenomenon, especially in earlier stages, were not fully understood, such a complete definition was impossible. Yet, Boyd might argue, due to the theory-constitutive metaphors (‘virus’, ‘worm’) and their concomitant sets of associated commonplaces, we are able to disambiguate terms which may definitionally have appeared co-extensive or co-occurring. Thus, metaphors aren’t simply exterior to the ‘real’ work of science; they aren’t what get us on the road to science. Rather than simply setting us on the right path (metaphor as heuristic) they help us cut that path when our other, more definitional, tools are of no use.


  • Richard Boyd, “Metaphor and theory change: What is “metaphor” a metaphor for?” 481-532 in Metaphor and Thought, 2nd edition, edited by Andrew Ortony (Cambridge, Cambridge University Press, 1993).

I knew that exciting first quote would snag your attention. O Blandest of Readers!