This paper will discuss Kant’s argument in section 16 for the unity of consciousness. In particular, it will dissect his claim that the analytic unity of apperception is presupposed by the synthetic unity of apperception. It will also briefly characterize the Humean and Cartesian views of self or the unity of consciousness, in order to bring out views which Kant is striving to avoid.
We’ll first provide a general characterization of both synthetic unity and analytic unity, before discussing them as they relate to apperception.
Synthesis is an activity where the raw material received in sensibility is unified (by certain conceptual rules) into a coherent representation (B130). Synthesis is an act of the understanding which combines the raw sensory input into coherent representations. It takes the spherical-sensation and the orange-sensation and combines them into a representation of an orange. The representation of the orange has synthetic unity in that it unites orangeness and sphericalness in some object under the concept “orange”.
The notion of combination is crucial to understanding synthesis and synthetic unity. Combination is an action by our faculty of understanding, it is not an activity of our faculty of intuition. Nothing in pure intuition is exhibits combination, so combination cannot be tied to sensibility and must be tied to understanding. Combination, then, does not lie in the objects, but is an a priori feature of our understanding (because it is an activity). We cannot represent those properties as combined in some object that we have not previously combined in or through our understanding (B130). In this way, combination is the representation of the synthetic unity of the manifold of intuition. For, a complex representation must unite various properties in one object, and this is just what it is for the faculty of representation to engage in combination-activity. But if combination is the representation of synthetic unity, then we cannot ground the representation of synthetic unity in combination (for that would beg the question). Instead, combination arises by “adding itself to the representation of the manifold”. Kant is not explicit about what this means, but there seems to be only one way, on his picture, that combination can “add itself” to the representation of manifold. Namely, that our representation of manifold of intuition combines sensations (for it is not clear what our representation of the manifold of intuition would be if we did not combine the various sensations into coherent representations), and the very fact that our representations are complex in this way constitutes combination’s “adding itself to the representation of the manifold”.
Analysis is the opposite of synthesis. The “dissolution of combination” is analysis. That is, we can only break a representation into its elemental components if that representation was complex, combining various properties to be teased apart. Because analysis breaks down complex representations into their components, analysis presupposes synthesis1 – for in order to have a complex representation, synthesis must be possible, “you cannot dissolve what has not been combined” (B130). A group of objects have analytic unity if, when broken down into their components, they all share some property P. For instance, trees, shrubs, ferns, and watermelon are analytically united under the concept/property “green”. They “belong together” in virtue of their greenness – we break each representation down and see that they share this common element.
We’ll now explain Kant’s argument for the unity of consciousness.
The most significant claim of the passage is that the analytic unity of apperception is presupposed by the synthetic unity of apperception (apperception is self-consciousness). We’ll later see how this allows Kant to establish his conclusion regarding the unity of apperception, but we must first explain how he justifies this claim and what it amounts to. Establishing this claim relies on a crucial premise: that it is possible to attach “I think” to any one of my representations.
If I am representing a keyboard on the desk, I can attach “I think” to it and produce the representation “I think that there is a keyboard on the desk”. “I think” is a special representation with three major components. (1) The “I” indicates that whatever representation “I think” is attached to I am conscious of as mine. (2) I am ascribing a certain representation to myself. (3) The representation (e.g. of the object, say, a pear), is somehow related to me. So we have a power to think of ourselves and ascribe things to ourselves, and this cries for explanation. We will explain this further after we explicate the justification of this initial claim.
Suppose I had a representation that I could not attach “I think” to. Then I am representing something which I can have no thought of representing, for I can never think to myself that I am representing it. If I can have no thought about this representation, then this representation is “nothing to [me]” (B132). That is, it is meaningless and has no content. But all my representations are contentful, so I must be able to attach “I think” to all my representations. This possibility of attachment is the condition under which any representation can be said to be mine (B133).2 If it were not, then there would be representations (of mine) that I could not attach “I think” to, that is, I have representations that I cannot be aware of representing or thinking that I represent them (B133). And it’s not clear how we are supposed to make sense of this – that is, your having representations that “do not belong to you”, that you cannot be aware of. Consequently, I must be able to attach “I think” to all my representations.
If I see, or have a representation of a pear, I can attach “I think” to the representation such that I represent, “I think that there is a pear”. If I represent a counterexample to the modal invariance lemma, I can also represent “I think that there is a counterexample to the modal invariance lemma”. (That is, we can attach “I think” to representations in the intuitions of both space and time.) All my representations bear a necessary relation to that special representation of mine (me being the relevant subject), the “I think” (insofar as I must be able to attach “I think” to them, regardless of whether or not I actually do [B132]). Because I can attach “I think” to all my representations, all my representations I can ascribe to myself, and so all my representations share the property of “being mine”. It is important to note, however, that the “I think” representation is a concept belonging to the understanding and not the sensibility (for you will never perceive your “I” or “think” concept from any appearance in sensibility); moreover it is an activity and not a passive occurence. Again, we’ll need some account of what kind of concept this is and whence it comes, since it cannot be gotten through experience (this suggests that it will be a priori, as we will later show).
Having established this crucial premise, that it is possible to attach “I think” to any one of my representations – that is, that all my representations necessarily conform to the condition that I can attach “I think” to them, the condition “under which alone they can stand together in one universal self-consciousness” lest we have representations which do not belong to us – Kant can now argue for the claim that the analytic unity of apperception is presupposed by the synthetic unity of apperception.
Recall our earlier characterization of analytic unity, where objects are united under some property they share. Recall, too, that all my representations are such that I can attach “I think” to them. Then all my representations have analytic unity. That is all my representations can be united under the “I think that such-and-such” concept. This is crucially different from our examples of trees and watermelons being analytically united under green, for greenness is a sensible property which is importantly different from the property of being mine. There is no sensation of “being mine” that I could point to which accompanies my representations. It’s not as though all my purported representations come with an identifying red tag that I use to discriminate between those representations that are not mine. “Being mine” is not a property of objects which is represented by some feature. But each representation of has the property of having “I think” or “is mine” attached to it – even though this is not sensible. But it is this property that all my representations share, and consequently it is this property which gives apperception its analytic unity – all my representations belong together in virtue of being mine. In this sense, they have analytic unity. This is to be regarded as the analytic unity of apperception. How is the analytic unity of apperception possible, and whence comes the “I think”? The subsequent portion of Kant’s argument offers some insight.
Apperception contains a synthesis of representations (B133): whenever “I think” is attached to my representation the result, “I think that R” is synthetic, bringing together “I think” and R (and R’s relation to me [in virtue of the “I” in “I think”]). For self-awareness or apperception requires not just that we think R, but that we be able to think that we think R, or else we would have no use for the “I” in “I think”. But to be able to think that we think that R, Kant argues, we must be conscious of our representations’ being synthesized. Why? Suppose I am not consciouus of my synthesizing of representations. My experience of empirical reality, varied as it is, will never present me with some relation to my identity as a subject. (For, as we showed earlier, there no “red tag” present in outer sense representing yourself as a subject of diverse representations.) But we do have a concept of our identity as a subject, and since we just saw this cannot be obtained on the supposition that we are not conscious of the process of synthesizing representations, it must have something to do with our being conscious of the process of synthesizing representation.
But how do we bring about this relation to our identity as a subject? It comes out of our conjoining or combining representations as we do and, moreover, by being (able to be) conscious of this process of combination.
This entails that I can only represent to myself the identity of my consciousness in these representations insofar as I am able to unite the manifold of representation under one consciousness. That is, the former – the analytic unity of apperception – is only possible in virtue of the latter – the synthetic unity of apperception, the ability to attach the “I think”.
Having established that the analytic unity of apperception is only possible by the synthetic unity of apperception, Kant claims that the synthetic unity of apperception (with regard to the manifold of intuition) is generated a priori and, consequently, serves as the ground of the identity of apperception – that is, the ground of the identity of a persisting self. Why is the synthetic unity of apperception qua the manifold of intuition a priori? (1) It must precede a priori all determinate thought, and (2) combination is not a thing which lies in objects themselves. In order to have a determinate thought, the thought that such-and-such is so, I must have a “such-and-such” and an “is so”. That is, I need a representation synthesizing an object and a property (or properties within some object) before I can have a thought which unites a property with an object. In this way, the synthetic unity of the manifold is a necessary condition on having any determinate thoughts. This suggests that combination is a necessary condition on thought. If combination is a necessary condition on thought, then it cannot be a feature of objects themselves and must be an activity of our minds. Thus, synthetic unity (of apperception) is a priori. And because the analytic unity of apperception is possible only because of the synthetic unity of apperception – that is, analytic unity of apperception is grounded in the a priori synthetic unity of apperception – the analytic unity of apperception is, too, a priori. So the unity (both synthetic and analytic) of apperception is a priori. And the unity of apperception is the identity of the enduring subject.
At this point, it will be useful to consider two alternative accounts of the self in order to bring out the salient features of Kant’s view of apperception as well as clearly show what Kant is not saying. First, we’ll characterize the Humean view and demonstrate Kant’s issues with that. Then we’ll characterize the Cartesian view, and explain how Kant would reply.
The Humean view is that one has awareness of self through a sensible intuition. The “self” one is aware of, however, is not a persisting self, but rather a mere collection of distinct representations. When you look internally, the only self that is present is whatever current state-of-affairs is being represented. A new representation (one that, say, occurs an instant later) is a new, distinct representation. Consequently it relates to a new, distinct subject. There is no persisting self; rather, there is a continuum of different selves relating to the continuum of representations experienced. Every thought of the form “I think that such-and-such is so-and-so” must have a different subject as the referent of “I”.
Kant rejects the Humean view because it does not account for unity of thought. All my representations I ascribe to a single thing, which can consider all of them (the representations), judge and think about them accordingly. This speaks to a certain “unity of thought” present in us. There is a stronger sense in which we must account for the unity of thought, however. The perception of an ice cube melting follows a necessary temporal sequences in which it gets smaller and forms a puddle, regardless of however else your mind may relate to it. Whereas the perception of various faces of a cube as you move around it follow no necessary temporal sequence. The former perception, insofar as it has a necessary sequence of perceptions, highlights the unity of consciousness. That is, these representations are necessarily united in consciousness in a certain way – and this is what cannot be accounted for on the Humean view. The Humean picture, wherein “I” always takes a new subject as its referent, cannot account for this unity of thought, because there is no unity of subject. Consequently there is no reason that perceptions should follow this kind of necessary temporal sequence. Kant, however, in arguing for the unity of apperception provides a successful account of the unity of thought for there is a persistent self who bears all the representations.
The Cartesian view maintains that one has awareness of a self through an intuition of a self as a persisting entity bearing representations. I have a nonsensory intuition of a single thing which bears the representations, and call this thing “self”.
Kant rejects the Cartesian account because it relies on a nonsensory intuition. On Kant’s picture, all intuitions are sensible. If there is an intuition of a single thing which bears all the representations, then there is some sensation corresponding to that single thing. But as we explained earlier, there is no sensation or “tag” in our experience which delineates my representations from those which are not mine. If there is some nonsensory thing which bears all the representations, then, it cannot be an intuition. This would suggest some sort of conceptual insight. But if this is so, then the concept must be a priori (for we saw how it could not be obtained through experience). Fortunately, Kant furnishes some argument for this, as we saw when explaining how the synthetic unity of apperception is a priori.
To recapitulate, Kant argues for the unity of apperception in the following way. We can attach “I think” to any representation. All my representations have analytic unity because they all belong to me. All my representations have synthetic unity because I can attach “I think” to any of them. The former is true in virtue of the latter. And the latter is grounded in something a priori. Because the unity of apperception, both analytic and synthetic, is a priori, we can know that we have a persistent self or identity.