Does evolutionary theory contradict Platonic realism?

Evolutionary biologists don't tend to like Plato very much. Consider the following quote from Richard Dawkins:

" If, like Aristotle, you treat all flesh-and-blood rabbits as imperfect approximations to an ideal Platonic rabbit, it won’t occur to you that rabbits might have evolved from a non-rabbit ancestor, and might evolve into a non-rabbit descendant."

Dawkins explicitly draws on the renowned evolutionary biologist Ernst Mayr, who argued that the "dark hand of Plato" was the reason why evolutionary theory took so long to be (almost) fully accepted in the West. I will briefly address the above criticism and some others in this entry.

The thought here is that if one believes that members of a species all reflect the same unchangeable form, one is unlikely to accept that the ancestors of the members of this species belonged to a different species. Dawkins and Mayr may be right that Plato's influence has contributed to the great resistance to evolutionary theory, and even that platonic realism itself makes evolutionary ideas seem unintuitive. Nevertheless, it is important to emphasize that nothing in the theory of platonic realism contradicts the notion that a given species can evolve from another species. This is because the claim that all members of a species participate in the same form is perfectly compatible with the claim that said members have a disposition to bring about (however indirectly) members of a different species, who participate in a different form.

A more worrying objection is voiced by Prof Volker Sommer, a primatologist at UCL. Sommer argues that the categories that species constitute have "blurred edges". The evolutionary account of biological diversity entails that, rather than being discrete, sharply distinguished 'boxes', species are continuous with one another, in much the same way that colors are. Thus, just as there is no sharp boundary between orange and red, there may be no sharp boundaries separating two species, which means that there will be 'borderline' cases where there is no fact of the matter as to whether a given organism belongs to one species or another.

Sommer suggests that, because of this, there is a sense in which, far from merely defining preexisting "borders of the world", zoologists "create worlds". That is, the species to which an organism may or may not belong is a convenient artificial grouping, rather than a real or natural one. We create such artificial groupings all the time: for example, I might conceptually group some human beings under the name 'friends', and others under the name 'acquaintances'. Groupings like this are practically useful, because they help me navigate the complex world of human relationships (for example, to decide whose email to answer first, and so on). But they aren't real, in the way that we'd expect species to be real.

Interestingly, when I took his module back at UCL, I once heard Sommer say in a lecture that it was ironic for Darwin to have named his book The Origin of Species, since his theory implied that there really are no such things as species.

But if species aren't real /natural groupings, they can't be said to reflect forms -- as we saw previously, part of the point of positing a realm of forms is to explain why such real groupings exist.

This problem is well-known to philosophers of biology, and there is no way a single blog entry could do it justice. Hence I will content myself with a few points:

1. The objection, at most, shows that members of the same biological species don't participate in the same form. This is much more modest than saying that no grouping is made up of things participating in the same form. Realists don't need to believe that all the names and categories we use are real and not artificial, and hence linked to their respective forms. Few, for instance, would say that there is a form of the 'democrat' or of 'being an acquaintance'.

2. The objection relies on an inference from 'this grouping has vague boundaries' to 'this grouping is artificial, rather than natural'. But it really isn't obvious that a grouping doesn't have anything to do with the real structure of the world if it has borderline cases. Take two kinds, 'bird' and 'zird'. Something is a member of 'zird' if it is either a bird or a mobile phone. 'Zird' is an artificial grouping if there ever was one. The evolutionary story tells us that members of the kind 'bird' are descendants of the kind 'non-avian dinosaur'. This means that there almost certainly have existed organisms for which it is vague whether they belonged to 'bird' or 'non-avian dinosaur'. So 'bird' has vague boundaries. But it just seems wrong to say that 'bird' is on par with 'zird', both being equally artificial groupings. For one thing, that an animal exemplifies 'bird' explains much more than that it exemplifies 'zird'. The fact that Pete the parakeet is a member of 'bird' explains quite a lot about Pete, e.g. the fact that he has feathers and a beak. Indeed, these traits are what we would expect of a member of 'bird'. In contrast, the fact that he is also a member of 'zird' explains very little about him. It may perhaps explain the fact that he is a material thing, since all members of 'zird' are material things. But this explanation in terms of being a zird is totally unnecessary, as the fact that he is a bird can explain this just as well (since all birds are material things).

Of course, those suspicious of allegedly 'natural' groupings could retort that this just makes 'bird' more useful than 'zird', not any more real (see Nanay 2011 for a view along these lines), since it helps us make more predictions about the features of members of 'bird'. I would tentatively submit that high explanatory and predictive power is just part of what it means for a grouping to be 'natural'. In fact, I can't really see what more one could ask of a given grouping for it to genuinely count as 'natural', i.e. somehow reflective of the structure of nature. In that case, saying, 'I see that the grouping 'bird' helps you explain and predict many features of the members of 'bird', but is it a natural grouping?' would be conceptually confused, like saying 'I can see that you are single and male, but are you a bachelor?'.

If I'm right, the fact that species (like Homo sapiens) have vague boundaries does not imply that they are purely artificial groupings, and hence that they cannot have a corresponding platonic form, which their members participate in.

These responses to Sommer's objection may not be decisive (at least in their current form), but they should at least give evolutionary biologists an opportunity to reconsider their natural aversion to platonism.

No demiurge, no forms

The last entry discussed realism, the view that there exist universals (traditionally known as 'forms'). As we saw, on realism, objects that share properties, e.g. two red objects, are said to exemplify or participate in the same universal, in this case, the universal 'redness'. Realists generally regard universals as abstract, existing in a realm causally disconnected from the realm of concrete things that we exist in.

I argued that a benefit of realism is its potential to explain why the world contains what might be called 'real resemblance groupings'. Things in the world, as it were, 'come in boxes', i.e. in natural categories that our concepts try to demarcate. Trope nominalists explain this by saying that things belong to the same real resemblance groupings because the individual properties ('tropes') they have resemble each other (e.g. the redness of object 1 resembles the redness of object 2), and leave it at that, leaving resemblance as an unexplained brute fact. The realist explanation goes further, as it explains why resemblance exists at all. Resemblance between concrete things in the world is ultimately explained by their participation in abstract universals.

However, I noted that realism faced various troubles. I will mention two:

1. Unintelligibility: the reader may have noticed that, frustratingly, I have used phrases like 'participate in' and 'exemplify' without ever explaining their meaning. Unfortunately, such sloppiness on my part reflects the very real problem that we are rarely, if ever, told by realists what it means for a thing to 'exemplify' or 'participate in' a universal. This is a serious weakness. Recall, the exemplification of / participation in universals is supposed to explain the existence of real resemblance groupings. But a purported 'explanation' can't do any explanatory work if it is unintelligible. Suppose my 'explanation' for the existence of real resemblance groupings is that they are 'troxed by gromphles'. It wouldn't do for me to criticize trope nominalists for offering an incomplete explanation of real resemblance groupings, as the 'explanation' I put forward is no explanation at all, since its meaning (if it has one) is impossible to understand. Of course, realists may answer that terms like 'exemplify' and 'participate in' are primitive, meaningful but impossible to analyze (like, perhaps, 'exists'). But in doing so they will be hard-pressed to convince their audience of the reality of what they are talking about. If I don't understand what the terms of the alleged explanation mean, I will never be able to accept the explanation itself.

2. Causal isolation: realists typically hold that universals are abstract, by which they mean that they are outside of space and time, and therefore causally isolated from the world of concrete things. That is, they cannot causally interact with such things. Two subproblems arise from this:

2a. if universals can't causally interact with concrete things, how can they be responsible for the real resemblance groupings in which concrete things find themselves?

2b. since you and I are concrete things, universals aren't causally related to us either. But this would seemingly make it impossible for us to know about them. Philosophers tend to accept the so-called 'causal theory of knowledge', which says that, for a knower to know a fact, there must be some sort of causal relationship between the knower and that which he knows. For example, I know that there is a cup on my desk because the cup's being on my desk causes me to have a perceptual experience of the cup.

It should be clear that the problems raised here are more than just unfortunate consequences of realism, that we just have to accept because we need realism to explain real resemblance groupings. Indeed, if these problems aren't dealt with, it turns out that realism can't explain anything at all.

I will now argue that these problems don't arise if the realist also believes in a demiurge. The demiurge (demiorgos = 'craftsman'), for Plato, was a God of some sort, who creates the concrete things of this world out of preexisting raw materials by modelling them on the perfect forms. Thus, things that belong to same real resemblance grouping do so because they were ultimately modeled according to the same form. Later platonists often interpreted the forms as eternal ideas in the demiurge's mind.

How does this more ancient variety of realism fare with respect to the foregoing problems?

1. Unintelligibility: to 'participate in' a universal is to be understood as having been modeled on this universal. This is much more intelligible than in the demiurge-less version of realism, because we know what it means for something to be modeled on something else. Moreover, if we take the later platonist view that sees universals as divine ideas, realism becomes even more intelligible, since we know what ideas are and we know what it means to model things on them (this current blog article has itself been modeled on my idea of a blog article on the importance of the demiurge to realism).

NB: this does not commit the realist to believing that the demiurge directly creates the members of real resemblance groupings, through spontaneous acts of creation. The demiurge could instead indirectly create them, by setting up the initial conditions and laws that would bring about the desired groupings. Hence, the theistic realism described here is compatible with scientific views of the development of the universe (more on this in another entry).

2. Causal isolation: in Plato's realism, the demiurge acts as a kind of causal bridge between the forms and the realm of concrete things. The demiurge contemplates the forms and crafts concrete things that reflect them. So, the universals are causally responsible insofar as they inspire the demiurge to create as he does.

Moreover, since real resemblance groupings are ultimately causally linked to the universals via the demiurge, the epistemic problem disappears as well. We can know about them by inferring their existence from the real resemblance groupings for which they are ultimately causally responsible.

If I am correct that this theistic realism has what it takes to deal with the problems of unintelligibility and causal isolation, the implications are significant. For one thing, it would mean that realists should think twice before minimizing the importance of the demiurge to Plato's original theory, to which they owe their realism. This may mean being compelled to choose between realism and atheism, which would be bad news for realists openly hostile to theism, like Russell. For another, this provides natural theologians with the basic constituents of an argument for theism: if we need universals, but can't have them without theism, then it looks like we need theism too.

Two 'Problems of Universals'

Back when I taught Philosophy A-level, I noticed that students were usually able to explain Plato's Theory of Forms, or at the very least, to tell a story about how all concrete, individual horses all somehow participate in or reflect a perfect, unchangeable 'Form of the Horse'. However, when asked why Plato, or anybody else, would believe such 'forms' to exist, answers proved rare.

Unfortunately, A-level students aren't alone in their confusion. Philosophers tend to agree that the motivation for believing in 'forms' -- which they now call 'universals' -- comes from a perceived need for a solution to a so-called 'Problem of Universals'. However, on top of disagreeing about whether universals exist, they also disagree on the exact nature of the 'Problem' they are supposed to solve.

I have found at least two candidates for being the 'real' Problem of Universals, which I will call the logical and explanatory problem respectively.

The logical problem of universals goes something like this: consider two Lions, Liam and Lionel. Liam is a lion. Lionel is a lion. But Liam is not Lionel. How is this possible? How can Liam and Lionel both 'be' the same thing if they are distinct from each other? After all, we know that if x = y and y = z, then necessarily, x = z (the transitivity of identity). The natural answer is that, when we say that Lionel is a lion, and that Liam is a lion, we mean that despite being distinct, they have something in common. But what is this 'something' which they share? Clearly, it can't be the specific, concrete traits and body-parts which each lion has. Liam doesn't 'share' his mane with Lionel, Liam has his mane, and Lionel has his, they don't take turns to wear the same mane. And so it is for all their other traits and body-parts. Plato and his supporters conclude that the 'something' they share is their exemplification of the universal 'lion'.

Thus understood, the problem of universals is a logical problem, because universals are brought in to dispel an apparent logical contradiction. Universals answer the question of how it is (logically) possible for two distinct things to 'be' the same thing. In this version, the problem is sometimes referred to as the 'One Over Many' problem. Whatever Plato himself thought, many since him have espoused realism (here, the view that universals exist) on the basis that it is necessary to solve the logical problem -- see Bertrand Russell's classic The World of Universals for an example of this.

On the other side of the fence, proponents of nominalism (the view that there are no universals) also often agree that the problem is a logical one (e.g. Rodriguez-Pereyra 2000), though they disagree that other-worldly, abstract universals are needed to solve it. In its simplest form, the nominalist response says that what Liam and Lionel share is a resemblance: they are said to both be lions because they resemble each other. But that clearly isn't enough. Liam the lion also resembles Terry the tiger, but obviously they can't both be said to be lions. Nominalists have put forward many ways to improve the simple response, but the strongest of these, in my opinion, is the one that appeals to tropes. A trope is sometimes referred to as a 'particularized property'. Think of two red balls: the two balls may both be red, but the redness of the first ball is distinct to the redness of the second ball. If I hide the second ball under the table, you can see the redness of the first ball, but not that of the second ball. Thus, we say that each ball has a 'red-trope'. Such tropes are readily accessible and part and parcel of the world around us, not shut away in their own realm. Importantly, they do not share the same red-trope. Rather, they each have their own, though the two red-tropes may resemble each other exactly. This is the trope nominalist solution to the logical problem of universals: it is logically possible for two distinct things to 'be' the same thing if the have tropes that resemble each other. Liam and Lionel are both lions because Liam has a lion-trope and Lionel has a lion-trope, and both lion-tropes resemble each other.

Not everyone is happy with the idea of a trope, some even calling it incoherent (Levinson 2006). But it seems to me that their existence is perfectly plain -- right now as I write, I see the blueness of the sky outside, the redness of the table cloth, the 'rectangleness' of my notebook, and so on. In any case, I think the trope nominalist successfully solves the logical problem of universals, in showing that it is logically possible for two distinct things to 'be' the same thing, without recourse to universals.

As I have said, however, not everyone agrees that the problem of universals is a logical problem. Paul Gould, a proponent of realism, argues that the question that realists and nominalists must answer is the question of " what a metaphysical explanation of qualitative and resemblance facts would be" (2012). The world we live in is one in which different things come under the same names or categories. The explanatory problem of universals is the problem of the "explanation" of this interesting feature of reality. Of course, some philosophers will say that this is just due to a contingent feature of language, i.e. its tendency to regroup different things under the same concepts (this view is held by conceptualists and predicate nominalists). But many other philosophers would insist that our conceptual groupings reflect real categories and resemblance relations. Indeed, both realists and trope nominalists would say that distinct things fall under the same concepts because they really resemble one another in certain ways. Realists argue that this real resemblance between different things is due to them all exemplifying the same universal. Trope nominalists, on their side, contend that it is due to individual things having tropes that themselves resemble one another (e.g. the redness of ball 1 resembles the redness of ball 2).

So, do both realism and trope nominalism solve the explanatory problem of universals? Not, says Gould, to the same degree of satisfaction. To see why, notice that the trope nominalist account explains the resemblance of distinct things by pointing to the resemblance of their respective tropes. But in that case, it cannot really explain why the world contains resemblance at all. Such an explanation would be viciously circular, as it would involve explaining resemblance by appealing to more resemblance. The realist goes further in answering the explanatory problem: the world contains real resemblance relations, because groups of things are linked to the same universals by the exemplification relation. Therefore, while trope nominalism must ultimately regard resemblance as an unexplained brute fact -- because the resemblances between tropes are themselves unexplained -- realism does answer the question of 'why is there resemblance at all?', since the exemplification of universals by concrete things is not itself a resemblance, making the explanation non-circular.

Gould concludes that "realism regarding universals is explanatorily superior to its nominalistic competitors". Of course, the debate between realists and trope nominalists cannot end here, as realism has problems of its own. The most damning of these, I think, have to do with the apparent impossibility of universals disconnected from our world somehow causing the world to exhibit resemblance, and of being known to exist by us (I hope to address this is another entry). But what's interesting for our purposes is that our understanding of the problem of universals, i.e. whether we regard it as a logical or an explanatory problem, bears on the question of whether we ought to think that Plato may have been right after all.

Levinson, J. (2006). Why There Are No Tropes. Philosophy, 81(04), p.563.

Gould, P. (2012). The Problem of Universals, Realism, and God. Metaphysica, 13(2), pp.183-194.

Rodriguez-Pereyra, G. (2000). What is the problem of universals?. Mind, 109(434), pp.255-273.