Proving the Self

Proving the Self

Is it possible to prove that the self exists? First let’s consider the existence of material objects, particularly the relationship between property instantiation and object-hood. An ordinary material object, such as a table or a bee, does not instantiate a single property but a collection of properties—a cluster of properties. Search the world high and low and you will not find a material object that confines its attentions to a single property; there is always co-instantiation as well as instantiation. There is always a type of fullness to the cast of properties that any material object instantiates. Without that the alleged object is too thin to count as a genuine object; it is nothing but a property attached to a location. If the property were to cease to be instantiated, the object would cease to exist, so that identity through change is ruled out. That is not our notion of a material object: an object can lose some of its properties and still exist, simply because it has other properties to anchor its existence. Co-instantiation of properties is constitutive of the existence of an object: an object is a unification of properties, a co-presence, an integrated cluster. Nothing is a genuine object if it restricts itself to a single property: a flying object, say, is never just a flying object; it will also have such properties as weight, size, shape, color, wings, and bristles. If someone were to claim that there could be a flying object that has no further properties, we would doubt that the word “object” was the appropriate word. The object-hood of a flying thing (such as a bee) requires the presence of attributes other than flying (and whatever attributes logically follow from this). Always and everywhere material objects, properly so-called, come attached to a cluster of properties, typically quite rich; they are not merely instances of a single property (trait, attribute). We could say they have a multi-dimensional nature.

            Now consider the Cogito: “I think, therefore I am”. We are being asked to accept that if something has a certain trait, viz. thinking, then it exists as a thinking thing. We can move from the instantiation of a property to the existence of a thing that does the instantiating: if there is the property of thinking, then there is a thing that thinks. Or again, if there is the activity of thinking, there is a thing (object, substance) in which the activity occurs. This move has always been found problematic: why does the process of thinking entail a thing that underlies the process?  [1] Mere thinking is too exiguous a basis to ground the ontology of selves. The point could be put as follows: why should we interpret “I think” as entailing predication of an object rather than being merely a feature-placing sentence? If the sentence “It’s raining” is true, then a certain feature is present at a certain place, but it doesn’t follow that there is a raining thing: there is no object that instantiates raining, just a certain meteorological activity occurring at a certain location. Similarly, it may be said, the Cogito is only entitled to the claim that there is thinking going on at a certain location (“It’s thinking here”) not the claim that there is a thinking thing, i.e. a self. Feature placing does not entail object-predication. Activity does not entail an agent. Process does not entail a substance. You can’t derive a something from a doing. Thus the traditional Cogito is invalid: the premise is too weak to support the conclusion. It takes an ontological leap across a logical chasm.

            I suggest that the problem here stems from the attempt to derive the existence of a thing from the instantiation of a single property (trait, activity, dimension). This is simply not rich enough to ground the notion of a thing; as with material objects, we need to add a range of properties co-instantiated with the given property. If that were not possible—if all we did was think—then indeed we would not be entitled to talk of a thinking thing; feature placing would be the preferred interpretation of the “I think” of the Cogito. The inference “It rains, therefore there is a thing that rains” is clearly not valid; and the skeptic would be well within his rights to insist that the “I think” of the Cogito should be regarded similarly. But actually we have resources beyond the usual thin interpretation of the Cogito: for not only do I think, I also feel, sense, imagine, and will—among other things. That is, I instantiate many psychological traits that are not entailed by thinking as such: I am a bundle of traits not just a single trait. But this kind of clustering is exactly what grounds talk of thing-hood. Thus I exist as a thing because I manifest a cluster of mental traits. Co-instantiation is what justifies the move to thing-hood, not instantiation singly considered.

I propose, then, what may be called the “expanded Cogito”: “I think and I feel and I sense and I imagine and I will, therefore I am (a conscious thing)”. My status as a conscious thing, not merely a congeries of free-floating mental activities, turns on the fact that I (a single entity) instantiate all of them. The mental properties are instantiated by the same thing, and that thing is precisely a thing. What is doing the work here is the clustering not the constituents of the cluster: you can’t derive the substantial self from the properties in the cluster considered singly, but you can derive that self from the fact that they form a cluster. Talk of things is precisely a way to register such clustering; without it all we have is the distribution of features at locations. The Cogito needs beefing up from the latter to the former, and we have the resources with which to accomplish that. I am a thing because I am many things. The traditional Cogito imputes too little structure to psychological self-attribution, regarding it as essentially one-dimensional (“thinking”), and as a result fails to sustain the assertion of thing-hood (and hence of the self). The cure is to recognize that multiple simultaneous attributions are true, and hence we have the ontological basis necessary for a claim of thing-hood—just as in the case of material objects.

            And it isn’t merely that multiple attributions are true; we also know them to be true. Not only do I know that I think and know that I feel (etc); I also know that I am all these things simultaneously. Thus I know the psychological fact that grounds my claim to thing-hood: I am presented to myself as a combination, a clustering, a bundle. This means that I can use that fact in proving the existence of my self: I know that I think and I know that I feel, but I also know that I think and feel—I know that I have many psychological traits at the same time. So I know what is necessary to infer the conclusion of the Cogito. Perhaps this is why we tend to go along with the Cogito on first hearing without analyzing it too closely: we have the resources to make it come out valid, so we don’t notice that the usual formulation leaves it vulnerable. I know myself to be a unitary thing because I am aware that I combine a number of separate psychological traits—that I am center of instantiation. I am aware of myself as a centered cluster of attributes and that’s why I assent to the thesis that I am a thing—not simply because I am aware of my thoughts in isolation from other psychological traits. Thus we read the traditional Cogito in the light of the expanded Cogito. In any case, I know of my simultaneous instantiation of distinct psychological properties as well as I know of my instantiation of those distinct properties, so I know what is necessary in order to provide the desired proof.

            This account of the epistemology of the self differs from other accounts. It differs from the traditional Cartesian account because it locates the operative premise in the clustering not in the items clustered, but it agrees that knowledge of the existence of the self is inferential: there is a definite move from “these properties are clustered together” to “there is a thing that underlies the cluster”. So the account does not postulate direct knowledge of the existence of the self: that is, we can provide a discursive proof of the self, as it is understood in the Cogito. Nor does the account ground knowledge of the existence of the self on some kind of immediate impressionof the self; the only impressions here are of specific mental traits (and possibly of the fact that they come in clusters). The correct way to reconstruct the epistemology of the self in the Cartesian style is via the expanded Cogito, and that has the form of an inference. My knowledge that I exist is therefore not like my knowledge that I think: my knowledge of the latter admits of no proof, being immediate, while my knowledge of the former does admit of discursive proof—even if I never explicitly go through such proof in my daily life. This seems intuitively correct: the self really isn’t a given in the way the contents of the mind are. There are intelligible forms of skepticism about the self, and thus we need a proof to combat them. The traditional Cogito came close but foundered on the objection from thinness; the expanded Cogito gets over that objection. We can thus reasonably argue that we exist.

Colin McGinn  

  [1] This is usually called “the Lichtenberg objection”: how do we move from events of thinking to a substantial thing that thinks? But the objection had already been made by Gassendi and others. The question may be put as follows: how do we justify the Cogito without presupposing a scholastic metaphysics of substance and accident?