Zebras, Bacteria and Asteroids

Toby Friend on what’s real and whether whatever it is can be part of a unified science

Two tenets are of significant concern to today’s philosophers of science: one continues to be that age-old idea of Scientific Realism, the other is a more contemporary assertion of the Metaphysical Unity toscience. Although the motivations for or against them are very different, there seems to be a pay-off with the degree to which anyone has so-far been able to accept one given their acceptance of the other. Or at least, that is what a survey of recent debate would seem to suggest. Why is this? I’ll hazard a guess after laying out what exactly the tenets claim and how philosophers have tried to orient themselves between them.

Roughly speaking, Scientific Realism is the view that the posits of science are, by and large, real. There is a nauseatingly lengthy literature trying to precisify this basic thought, but we need not go too deep. Zoologists posit zebras, biologists posit bacteria, and astrophysicists posit asteroids. For that reason, at least, scientific realists believe that there really exist zebras, bacteria and asteroids. Scientific Realism is controversial for wholesale reasons, such as the underdetermination of theories by evidence and their inevitable replacement by theories with contrasting posits. It is also controversial in particular cases. Sure, zebras are real, just look with your naked eyes onto the Serengeti plains! But bacteria and asteroids require instruments to see; and some of science’s posits don’t seem observable at all, such as muons and economic markets. Should we be realist about all of these too? All the same, a realist attitude to the posits of sophisticated sciences remains popular, if not the default.

The status of science’s Metaphysical Unity is a more inchoate debate. General questions of scientific unity have been in consideration for a long time, such as whether all scientific theories can be translated or “reduced” to the terminology and laws of physics, or whether there is a unique scientific method. Once upon a time it was seen to be crucial for there to be positive answers to these questions in order to show that science is a legitimate enterprise for gaining knowledge about the world. If we can’t say what makes the various disciplines within scientific inquiry theoretically or methodologically unified when compared with, say, pseudosciences like astrology and homeopathy, then we can’t say why the output of the former are to be trusted. Or so the thought went.

Nevertheless, the hope for theoretical and methodological unity has been largely put to rest. While there have been celebrated cases of theoretical reduction in science (e.g. of thermodynamics to statistical mechanics), the prospects (let alone evidence!) for anticipating the translation of, say, evolutionary biology or macroeconomics to quantum field theory, are slim. Similarly, philosophers have largely given up the attempt to determine any unifying method across scientific practice. Scientists will find themselves engaged in confirmation, falsification, derivation, bayesian inference, abduction, informal and formal proofs and all sorts of other types of reasoning for which it is hard to determine any non-trivial commonality. And that’s not even to begin comparing the difference in experimental instruments used across the sciences.

The search for Metaphysical Unity is arguably born out of the fading light of these former dreams. Science may not be methodologically or theoretically unifiable, but the abundance of impressive technological applications it has been put to, and its novel predictions strongly suggest it’s a legitimate way to gain knowledge of the world. Moreover, there is, in the end, only one actual world, and all the posits of our sciences must find a way to coexist without infringing on each other’s credibility. This coexistence isn’t made possible because the posits are separated from each other -- there is no “ecology land” or “physics planet”. Indeed, the posits of science are often to be found in exactly the same place for the simple reason that they are compositionally related. Populations are composed of organisms, organisms composed of cells, cells of chemicals, chemicals of atoms, atoms of elementary particles.

Moreover, these compositional relations can appear to indicate an explanatory dependence of what’s being composed on what’s doing the composing. For instance, it can seem reasonable to think that it’s because elementary particles are as they are that atoms are as they are, and because organisms are as they are that populations are as they are, etc. Those who defend the metaphysical unity of science believe that some crucial sense of science’s unity can be salvaged by careful scrutiny of the dependencies conferred by these compositional relations.

Yet it is exactly this scrutiny of composition which has led to a compromise in Scientific Realism. The issue which reveals itself repeatedly is that the tighter one tries to make the compositional relationship between composed and composing entities in the interest of Metaphysical Unity, the more frustrated will be one's ability to take the composed entity as being genuinely real. To show this, I’ll describe three broad positions one might hold.

The first, and perhaps starkest of examples of this frustration, comes with endorsement of Compositional Eliminativism. This view takes the sciences to be metaphysically unified for the trivial reason that, despite initial appearances, only the most basic constituents of reality exist. Under this view, then, composition is not really a relation at all, since nothing that is composed really exists. Ecologists may talk of populations, zoologists of organisms, biologists of cells and chemists of chemicals, etc., but ultimately everything that exists -- everything that’s real -- is the stuff that is said to compose these things. That, we may suppose, is just what it is the prerogative of fundamental physics to give labels and laws for. Since it’s only that domain which exists for the eliminativist, the metaphysical unity of science is therefore trivial, since nothing can be disunified from that which doesn’t exist.

Although they might struggle to admit it themselves, there do exist compositional eliminativists. Aside from the straightforward, if extreme, metaphysical unity they can afford the sciences, their view has powerful arguments in its favour. One is the so-called “causal exclusion argument”. This points out that it is a methodological principle not to endorse the reality of more than one satisfactory causal explanation for things; e.g. if long-standing natural fluctuations in global temperatures did satisfactorily explain global warming, then we shouldn’t posit human activity as well (we do so exactly because the natural-fluctuations aren’t sufficient). The causal exclusion argument then claims that the posits of physics supply all the satisfactory causal explanations we could ever want. Hence, positing anything beyond physics (including zebras, zygotes and zephyrs) is unjustified by the methodological principle.

A different argument observes that if we were to take the explicit posits of sciences other than physics to be real we face unreasonable commitments concerning the objectivity of vagueness. Since there can be no plausible answer to when a loose hair on a zebra is no longer a constituent of it, then a zebra would have to be compositionally vague. But many philosophers today think that vagueness cannot be a feature of the world and must instead only be a feature of our ability to describe it. Hence, zebras (and its fellow composed entities) cannot be objective features of reality.

Let’s not pause to assess these arguments. For whatever their status, Compositional Eliminativism is not massively popular. It just sounds bizarre (if not a little hurtful) to say, for example, that biologists can’t be concerned with organisms nor ecologists with populations since, in fact, there just aren’t any organisms or populations. If that were right, then what are these scientists talking about? And how can we understand their enquiry as legitimate way of gaining knowledge about the world? Moreover, there seems something equally methodologically problematic in asserting the reality of things which are very difficult to “observe” in any traditional sense at the expense of things like zebras or colonies of bacteria, which it seems we can see relatively easily. As a consequence, philosophers have sought ways to wriggle out of the eliminativist’s arguments in the quest for scientific realism. The central point for the present discussion is that in doing so, they sabotage the clean-cut metaphysical unity supplied by eliminativism.

Consider, then, a second cluster of views which come under the umbrella term “reductionist”. The reductionist endorses zebras and zygotes by claiming that their composition is one of identity with ensembles of basic physical stuff. Traditional reductionists will claim, for example, that what it is to be the type of thing a zebra or zygote is just is the way it is to be the type of thing a certain highly specified ensemble of the basic building blocks supplied by our best physics. These reductionists therefore hope to avoid eliminativism by suggesting that the posits of all sciences just are the posits of physics at different levels of description.

There is surely a sense in which reductionists are able to proclaim the reality of all sciences’ posits whilst retaining something of the elegance of the eliminativist’s metaphysical unity. The posits are real because they just are pluralities of the posits of fundamental physics. But is that a reality worth its name? For the reductionist, the building blocks of such a world are not zebras and zygotes, they are leptons, quarks and gauge bosons (or perhaps their respective fields). The categorisation of huge zebra-shaped clumps of subatomic stuff seems to be an entirely pragmatic and human inclination affirming no objective divisions in nature.

Contemporary reductionists have found smart ways to eke out an explanatory role for composed individuals which are not in the basic inventory of fundamental physics. Some will say, for example, that composite entities may exhibit a degree of “robustness” and “novelty” in their behaviour which is epistemically salient enough to warrant a realist attitude towards them, despite being derived from fundamental physical processes. A related approach is to identify the composite things with what Daniel Dennett called “Real Patterns”. That is, we treat the buzzing maelstrom of fundamental physical processes like a bitmap of data and the composed entities as patterns in the bitmap, reference to which compresses the information required to convey what’s going on. Defenders of the idea maintain that since compressibility is a perfectly objective matter, there is more to the reality of composed entities than mere human inclination.

By advancing objective criteria for composed entities, reductionists inevitably pull away from the simple solution to metaphysical unity endorsed by the compositional eliminativist. Composed objects may achieve an independent identity, whether it be due to their behavioural novelty or the fact they can exist with a variety of alternative underlying compositional forms. But in so doing, this manoeuvre also breeds disunity due to the dichotomy between the qualified reality of composite entities and the full-blown reality of physics’ posits, which alone have the status of being constitutively basic.

In response, some reductionists have suggested that there may in fact be no basic level, thereby demoting even the posits of fundamental physics to the same qualified reality as everything else. However, this only serves to exacerbate the problem that the reductionist must treat some things’ existence as less real than one might hope. This is why Dennett admits his view is one “perched between instrumentalism and industrial-strength realism”.

We’d have more claim to the industrial-strength reality of composed entities if they were governed by different laws. If it so happened that when physical stuff takes the form of a zebra, the behaviour of the entire ensemble is no longer predictable with the laws of physics and requires some new, zoological laws. That’s what philosophers in the business would call “strong emergence”: the zebra emerges from the operations of its composites because it behaves in a way unpredictable, even in principle, from the laws governing them despite being somehow compositionally dependent on them. That would seem to supply composite entities with industrial-strength reality, but it’s a super-controversial idea.

To raise but one issue, when something strongly emergent exists (a zebra, perhaps), the parts (e.g. an atom) must operate very differently from how they do when they are not parts. Instead of being autonomous agents, as it were, now they get their instructions from a larger corporate body. But what could be the cause of such a difference in operation? Not the individual parts, since that would undermine the strong emergence. But neither could the cause be the composed entity, since that thing’s existence is exactly dependent on the change in operation of the parts.

All the same, what strong emergence gains in realism it loses in unity. At the location of any instance of strong emergence is a fault line between the science of the constituent parts and the science of what’s being composed. Metaphysical unity is lost because the behaviour of the posits described at the lower level science don’t fully explain the behaviour of the posits at the higher level. This disunity is writ large for defenders of a third kind of view: metaphysical pluralism. These pluralists effectively see strong emergence everywhere in science. For instance, Nancy Cartwright talks of the laws of science working like a ‘patchwork’ over a dappled world, applying here and there in certain sanitised circumstances and affording little unification. In their defence, pluralists stress the dearth of empirical evidence for thinking that the laws of physics apply everywhere without exception: what reason do we have for thinking the laws of physics can predict where a thousand dollar bill being blown about in St Stephen’s Square will land? Has anyone actually tried to model the behaviour of the electrons in someone’s a fingernail?

Pluralism is often associated with a less-then-fully realist view. Presumably the thought is that the only way to digest the reality of so many things posited by different sciences is to take their objective existence with a pinch of salt. By contrast, however, metaphysical pluralist John Dupré has claimed to see no reason why many overlapping and intersecting kinds might not be equally and genuinely real. What I want to emphasise here goes even further: it is the very invocation of pluralism and its wholesale endorsement of strong emergence that has allowed them to make space for a genuinely realist position which is not available to reductionists and eliminativists. That is, it is precisely by endorsing the pluralists’ disunified world that a thoroughgoing realist attitude to all the posits of science seems to be made possible.

We’ve charted the full extent of the spectrum from Compositional Eliminativism to Metaphysical Pluralism. At one end, the tenet of Metaphysical Unity is preserved at the expense of Scientific Realism, at the other end things are the other way round. We’ve also considered reductionist views perched between these two extremes. Reductionists have a twofold problem of retaining the sense of unity afforded by endorsing the posits of fundamental physics as what is explanatorily basic while simultaneously trying to make plausible the qualified reality of composed objects. At least, that’s the current state of play. So, what’s the reason for this pay-off in commitment to the two tenets and can it be avoided?

For better or worse, the issue seems to arise from the fact that all parties assume that metaphysical unity can only be achieved if compositional dependence is one which renders the composing and composed entities somehow indistinct. If a zebra is fully compositionally dependent on its subatomic constituents, so the thought seems to go, then it is nothing “over and above” those constituents, and so cannot have a distinct existence. This is why the failure of compositional dependence, by way of strong emergence, has seemed to be the only way to avoid the indistinctness.

Must we think of compositional dependence this way though? Perhaps not. We don’t, for instance, think of causal dependence this way. Most philosophers take it that what happens in the future (e.g. a storm) fully causally depends (though may not be determined) by what happens in its past (e.g. a change in pressure). Yet at least since Hume many also think that causes and effects are distinct existences and that their realities can be as “industrial-strength” as each other. So, if you want to have your realist cake and eat it in a unified way, then one tantalising, if unexplored, suggestion would be that compositional dependence just is causal dependence.


Toby Friend works on the Metaphysical Unity of Science (MetaScience) project, funded by the ERC (grant no 771509) at the University of Bristol. His research concerns the nature of causation, composition and scientific laws. He is especially interested in how these phenomena relate to each other.