Levels-eliminativists, including this symposium’s participants, have raised a variety of criticisms about specific conceptions of levels as well as the broad use of levels as a metaphor or heuristic in science. The general question inspired by such criticisms and explored in this symposium is: then what? Craver (2014) has suggested it’s impractical to legislate metaphor use in science, including invocations of levels. And, in their talks, Craver and Havstad each outline visions for what should be preserved of levels concepts. In this talk, I explore the idea that scientists and philosophers should simply jettison appeals to levels in all but the most specific, narrow uses when appropriate, such as levels of buildings. An interesting fact about our world is that opportunities for such a use of the levels concept are strikingly infrequent. I begin by outlining some of the roles the levels concept plays in scientific discourse, focusing especially on biology. I then briefly survey criticisms of levels and outline how these criticisms are relevant to these uses of levels. This discussion is in part informed by Ylikoski’s preceding talk. I then motivate alternative concepts in place of levels in each of the uses I’d surveyed. The approach I take in this talk is to use the criticisms raised of levels to help inspire better replacements. ‘Levels’ is used variously to describe the world and to describe our investigations and representations of the world; any replacement must apply to phenomena or to scientific methods (not both). ‘Levels’ assumes regularity and well-orderedness of our world; any replacement should take seriously interconnection and variability. One primary conclusion of my talk is that no single concept is apt for all the distinct roles ‘levels’ plays; this is one source of the difficulties plaguing the concept. Thus, I cannot hope to motivate a unitary replacement concept in this talk. Instead, I conclude by surveying some alternative ways to structure biology textbooks—instead of the now-ubiquitous framing of levels of organization—to illustrate the opportunities for theoretical improvements created by letting go of levels. Surely the levels concept will always be an option for various theoretical uses in science and philosophy. My point here is that turning to levels eclipse other possibilities, some of which may well lead to theoretical or methodological improvements.

The intuitive notion of level is often employed by social scientists and philosophers of social science to conceptualize tricky theoretical challenges. While it serves as an organizing metaphor for thinking, its assumptions and implications are never fully articulated. Consequently, unacknowledged and unhelpful conceptual commitments may be introduced to the debate. In this paper, I will show how this happens in the case of the sociological micro-macro problem. Sociology deals with phenomena in a wide variety of temporal and spatial scales, from an individual's cognitive and emotional processes to long-term changes in territorial societies. Connecting the data and theories of phenomena at these various scales is an important but thorny theoretical challenge. When combined with competing explanatory ambitions of different research traditions, fears (or fantasies) of reductionism, and general conceptual ambiguity, it is easy to understand why thinking in terms of levels has felt tempting to many. However, we should avoid the temptation for three sorts of reasons. First, conceptualizing the micro-macro problem in terms of levels misses crucial features of the problem. The levels mindset often abstracts away from the heterogeneity of micro and macro properties, making the discussion sterile and difficult to challenge. Similarly, it misses the contrastive nature of the micro-macro distinction: the social scientists use the distinction flexibly, and the same phenomenon can be either micro or macro depending on its contrasts. Finally, the focus on levels perspective makes it difficult to see that the relevant sociological questions are more about causation, dynamics, and history than about constitution or realization. Second, the levels conceptualization introduces assumptions that are both unnecessary and unhelpful. For example, philosophers often automatically assume that the levels in the social sciences are both comprehensive and unique. However, neither of these assumptions has ever been demonstrated. On the contrary, there exist substantial challenges to such presumptions. Even the less assuming question: "how many levels are there in the social sciences?" might not have a meaningful answer. Furthermore, the levels mindset invites poorly justified causal assumptions. For example, the beliefs that there is some explanatorily privileged level or that causes and effects must be at the "same level" or of the same granularity lack independent justification. Third, thinking about the micro-macro problem in terms of levels suggests solutions that are distractions from the point of view of the development of substantial social scientific theories. The conceptualization of the micro-macro problem has invited philosophers of social science to import conceptual tools from philosophy of mind. There has been a hope that concepts like supervenience, realization, and downward causation could help make sense of the micro-macro problem or the issues related to methodological individualism. However, this has been entirely unhelpful. The relation between micro and macro is not analogical to the relation between mind and brain. Furthermore, the debate has turned a substantial theoretical challenge into a philosophical puzzle that does not need any concrete social scientific concepts.

Eliminativism about levels is an over-reaction to a real problem that demands instead principled pluralism. Levels eliminativism is motivated in part by the recognition of systematic failures of entailment between seemingly related but distinct ways of talking about levels. Levels pluralism, in contrast to eliminativism and to a facile, “anything goes” attitude, recognizes the distinct roles that, as Havstad calls it “leveling,” plays in the context of different intellectual practices and pursuits. After clarifying what I mean by saying that level talk is metaphorical, I suggest some basic questions (the relata, relations, and placement questions) that can help to diagnose the sense of level at play in the context of distinct scientific and philosophical. As incremental progress toward that end, I distinguish contexts of practice involved in explaining natural phenomena (levels of organization), in describing a system from different vantage points (e.g., Marr’s levels; the personal-sub-personal distinction), and in describing clusters of scientific activity directed at items in a given size scale (call these Feynman levels). I show that the sense of level in each of these distinct contexts a) plays a useful scientific role (with the possible exception of Feynman levels, as I’ll explain), and b) answers the relata, relations, and placement questions differently. This not only stands as a display of the sort of pluralism in the levels metaphor we find operating in science and philosophy (itself a curious social/psychological phenomenon) but recommends significant caution against running these constructs together.

Philosophical critique of scientific levels or candidate level-systems often blends into critique of the activity of leveling. Here I defend leveling, the activity, with no commitment to any given scientific system of levels or the notion that within a given scientific system there is necessarily one best way to level that system. After looking at instances of leveling across the life sciences, I conclude that to level something is to either conceive of that thing in a more inclusive context than that of just itself, or conceive of that thing as providing, for certain other things, a more inclusive context than just what is provided by each of those things by themselves. I also notice there are different kinds of context in which one can level: when leveling, the varying context might be spatial, temporal, compositional, conceptual, detail- or information-oriented, functional, or more. We like to level in lots of different ways, and to level lots of different things. Yet I still think it possible to defend a concept of leveling which respects the different and diverse outcomes of this conceptually coherent leveling activity. My primary aim is to make sense of different kinds of level talk—all of which rely on a concept of leveling, and seem worth saving (to me): for instance, talk of wholes being on a different compositional level than their parts; talk of first-order and second-order sentences being on different referential levels from one another; talk of cellular organelles, cells, and organs being on different spatial levels; talk of ages, epochs, periods, and eras being at different temporal levels; talk of tokens and types, instantiations and abstractions, thick and thin descriptions all being on different conceptual levels; and more. Secondarily, I hope to successfully relate leveling in the context of abstraction to leveling in these other contexts. When we conceive of a thing in a more inclusive context than just itself, we generally talk of having gone up a level. When we conceive of a thing as providing, for certain other things, a more inclusive context, we generally talk of having gone down a level. But “up” and “down” are just relational signs that indicate movement in opposing directions from one another. Moving up to a higher level in the sense of abstraction is often more inclusive in the sense of number of things included, but less inclusive in the sense of amount known about or ascribable to each thing included. Moving down to a lower level in the sense of abstraction is often less inclusive in terms of the number of things rightfully included in that new context but more inclusive in the sense of details attributable to each member of the set. This is a case where the relevant signs might flip, depending on what context or dimension is being emphasized, because there is often a negative correlation between amount of detail and extent of abstraction.