Saturday, August 27, 2011

The Problem of Universals

Introduction to Universals

Every particular thing has a set of properties. The dog is brown. The apple is red. Rex is a dog. Etc. The subject refers to an individual, but does the predicate refer to something as well? Clearly, it does, but what is it? If one dog is brown, and another dog is brown, does the property “brown” refer to the same single entity? Or are they two individual entities of brown? Or is “brown” even an entity at all? In other words, does the property “brown” refer to something, and if so, what is it like? If not, how do we account for resemblance between the properties of different things?

The Nature of Universals

If universals exist, they would be considerably different from particular things. Particular things are generally considered to be concrete, and extended in space. They also can only be in one place at one time. Universals, however, if they exist, are abstract and not extended. They can also be in many places at once.

Reasons to Postulate Universals

A name refers to something. “Rex” refers to an individual dog. If a property, such as “red’ or “dogness” did not also refer to something, then it would be a useless word. So universals would be the thing that the name refers to.

Also, to gain knowledge you need to be able to know something that is unchanging. Universals would be something that holds over and above any individual thing, thus making knowledge possible.

The Problem of Universals

Our language makes use of general predicates: the stop sign is red, the firetruck is red, etc. The predicate has a one-over-many nature. But is this generality real in nature? Or is it just us?

Versions of Realism

Realism states that universals actually exist, as independently existing entities.

Extreme Realism

Plato had the most famous theory of universals. Universals really exist in a third, ghostly, abstract realm, outside of time and space. Any particular thing “participates” in the Form: a brown dog participates in the Form of dog and the Form of brown. So there is the dog, the brownness, and the Form that grounds that brownness. The Form itself also has its own property. I.e., the Form of brown is itself brown.

Strong Realism

Associated with Aristotle. Universals exist, but they only exist in the things themselves. They do not exist in an abstract, timeless, spaceless realm. But they do have the ability to exist in multiple things at once. This version of realism is immune to one of the more devastating criticisms of extreme realism: the Third Man argument.

Objections to Realism

The Third Man argument: this effects only Plato’s extreme realism. If the Form of red, for example, itself needs to be red, then it too participates in a universal. So there is yet another Form above it. But that too would have to be red, and so on to infinity. We never reach an explanatory stopping point. However, strong realism is immune to this, since it says that universals exist only in the particulars themselves.

Another objection to realism is that we should not admit that something exists if we do not need to posit it: Occam’s Razor. Perhaps we can do without universals. For this, we turn to two varieties of anti-realism: nominalism, and conceptualism.


States that universals do not exist.

Predicate Nominalism

This theory says that when we say “Rex is brown”, that’s true of Rex and that’s just the way it is. “Is brown” is just a string of words that can be truly said of Rex, and that’s it. However, this theory basically ignores the whole problem, and so does not solve anything.

Resemblance Nominalism

Properties are explained by putting similar things into sets that resemble each other. So firetrucks, stop signs, and apples all belong to the “red” set. But some things will have multiple properties that will often be paired together: things with a heart and things with a kidney, for example. These things will then be said to be one property, because they are just one set. But clearly they are different properties. Also, sets will overlap. There will be a set with green apples, and that set will also have to include red apples, because they are members of the “apple” set.

But the biggest problem with resemblance nominalism is that it needs to appeal to a universal: resemblance. So it ends up needing universals anyway.

Trope Nominalism

Tropes are universals that have been converted to particulars. On this theory, objects consist of a bundle of properties called tropes. So the brown dog would be a bundle consisting of a brown trope, a dog trope, etc. A bundle of individuals, in other words. No universals necessary. Many of these will look the same, but each one is its own individual. Ones that look the same can be in sets, and can then appeal to resemblance tropes instead of universals, and will therefore avoid the problems with resemblance nominalism.

There are still problems, though. Even though the tropes are individuals, and some of them resemble each other (red tropes, for example), why do they resemble each other? What grounds that? A trope nominalist just says that it’s a brute fact. But it seems that this has not solved the problem at all. The same question just pops up again.


Holds that universals exist only in our minds.The mind creates, say, “dogness” as an abstract concept. But in trying to account for what grounds these similarities, we are drawn back into data about the real world and back into universals themselves.


The debate has raged for thousands of years, but today there seems to be a consensus that either strong realism or trope nominalism are true.


  1. I'm not sure how one can be anything but a moderate (what you call strong) realist.

    As is often the case, I really thought there was something I wasn't getting about trope nominalism that made it viable. Your cogent explanation makes me think otherwise.

    Meaningful language seems to presuppose realism.

  2. Yeah, I kinda think so too. I have a large book on metaphysics that I'd like to read sometime, but prima facile the case for moderate realism strikes me as the most plausible.