The levels metaphor is ubiquitous in our descriptions of science and the world. So simple and elegant, the metaphor takes an apparently heterogeneous collection of objects and arranges them in space from bottom to top. The metaphor works in so many contexts because it leaves open just what kinds of object are to be arranged, what distinguishes top from bottom, and what it means to say that an object is at some levels and not others. This flexibility explains the metaphor’s fecundity, but it also helps to obscure the fact that it is used in many different ways in many different contexts.
A survey of kinds of levels drawn from science and philosophy would have to include levels of abstraction (Floridi 2008), aggregation (Wimsatt 1997), analysis (Shepherd 1994; Churchland & Sejnowski 1992), causation and explanation (Kim 1998), implementation (Marr 1982), organization (Churchland & Sejnowski 1992), processing (Craik & Lockhart 1972), realization (Gillett 2002), sizes (Wimsatt 1976), sciences, theories, and explanations (Oppenheim & Putnam 1958). Many of these familiar applications of the levels metaphor are distinct from but also clearly related to one another. And when they are related, they often have rather indirect and reticulate connections. The level metaphors thus takes subtly different forms when applied in neighboring contexts, and this obscures the extent to which features of one application of the metaphor do and do not transfer from one context to the next. My first thesis, then, is that our ways of describing science and the world contain many distinct, legitimate applications of the levels metaphor that are either unrelated or that have only indirect relations with one another.[1]
This descriptive pluralism about the levels metaphor is directly opposed to eliminativism about levels (Fehr 2004; Machamer & Sullivan 2001; Thalos 2013). The suggestion that we might be better off abandoning the levels metaphor is about as likely to win converts as the suggestion that we should abandon metaphors involving weight or spatial inclusion. These metaphors are too basic to how we organize the world to seriously recommend that they could or should be stricken from thought and expression. Yet, descriptive pluralism about the levels metaphor is consistent with the thought that some applications of the metaphor distort the structure of the world or represent it as having conceptually incoherent structures. I discuss some examples below. The central message of this paper is that there can be no single verdict concerning the utility or conceptual soundness of the levels metaphor simpliciter. The metaphor must be evaluated and used with caution, especially when it is called on to settle disputes about the character of science and the metaphysical structure of the world.
As some motivation for adopting this proposal, and as a step toward a more positive thesis, I show that we can avoid some simple confusion by separating the different applications of the metaphor. To make this case, I build slowly toward a particular application of the metaphor that, as I have argued elsewhere (Craver 2001, 2007), is central to explanatory practices in neuroscience and across the special sciences: levels of mechanisms. This application of the levels metaphor is metaphysically plausible and, so far as I can tell, more or less innocuous; that is part of its virtue. Yet this simple and useful application of the metaphor can begin to appear problematic when it is inappropriately assimilated to other applications that serve altogether different purposes in our thinking about science and the world.
My point is not to defend levels of mechanisms as the one true application of the levels metaphor (that would be as pointless as eliminativism). Rather, my first positive goal is to provide a reasonably clear account of levels of mechanisms and to show that this application is metaphysically benign yet exceptionally important for doing science. Levels of mechanisms are, as would be expected, richly but indirectly connected with many other applications of the metaphor. My second goal is to highlight and disentangle some of the confusions that arise from failing to keep levels of mechanisms distinct from other senses of levels. In particular, I show that commitment to the existence of levels of mechanisms entails no commitment to: a) monolithic levels in nature, b) the stratification of sciences by levels, or c) a tidy hierarchy of theories among the sciences. I will also show why levels of mechanisms are d) distinct from Marr’s views about levels of abstraction and e) distinct from levels of realization more generally. I argue that f) the idea of interlevel causation is conceptually awkward within levels of mechanisms (but not to levels of size, for example). Furthermore, g) the idea of levels of mechanisms nicely expresses the idea of emergence as a kind of non-aggregativity while providing no support to those who seek evidence in biology for a more robust kind of emergence. The failure to disambiguate altogether separate applications of the levels metaphor creates a conceptual malaise for which levels of mechanisms are at least a partial cure.