Synthetic A Priori

Synthetic A Priori

The traditional category consisting of knowledge of synthetic truths is supposed to comprise two types of knowledge: a priori and a posteriori. These are taken to be mutually exclusive and sharply distinct—really nothing like each other (except in both being cases of knowledge). It tends to be supposed that a priori synthetic knowledge is epistemologically problematic, while a posteriori synthetic knowledge is not. Many theorists have sought to assimilate the a priori case to the a posteriori case, in various ways. I am going to suggest an alternative: the converse is true—so-called a posteriori synthetic knowledge is really a subclass of the synthetic a priori. Perceptual knowledge in particular is a type of a priori knowledge: when you come to know an ordinary fact about objects in your visual field, say, your knowledge is a priori not a posteriori. I know this sounds strange, to the point of outrageousness, but bear with me; I aim to lower your eyebrows in what follows. The orthodox classification turns out to be quite misguided upon careful analysis.

Is perceptual knowledge dependent on experience, i.e., conscious sensory experience? Is experience of this kind a necessary condition of perception-based knowledge? There are several reasons to think not. First, it is not the case that there is a type of experience such that it is necessary for perceptual knowledge—as it might be, visual experience. Such knowledge can result from the experiences delivered by quite different senses, some not even possessed by humans. This means we must define what is meant by a sense experience of any possible type, which is not an easy task. Second, there are several types of non-experiential perceptual knowledge: subliminal perceptual knowledge, blindsight perceptual knowledge, and insentient perceptual knowledge (various simple organisms). In these cases, no experience passes through the mind of the knower, yet he knows. If a knower is suitably stimulated by the environment, he or it will know a fact about it, whether conscious experience intervenes or not. Third, an experience may occur but not be the reason the knower knows; he may not know because of the experience, in virtue of it. Epiphenomenalism may be true of conscious experiences, so that nothing happens because of them, or the knowledge may be caused by non-experiential properties of the stimulus such as unconscious computations. We know this can happen in the absence of conscious experience, and it may also occur even when the experience is present; correlation isn’t causation. Experience is simply not a necessary condition of knowledge acquisition. Why should it be, given that some knowledge is entirely unconscious, e.g., knowledge of grammar and innate knowledge. The concept of (perceptual) knowledge doesn’t require that knowledge be preceded by a conscious sensory experience; such knowledge can be experience-independent. By the traditional definition, then, it is a priori—prior to experience, not dependent on it. There might have been perceptual knowledge without any sensory experience at all, just causal informational pathways from world to brain. The knower would still be conscious but not experientially conscious—no visual or other sensory sensations. The only question is whether in our own actual case the knowledge-generating mechanisms essentially involve experiences or is it a matter of extra-experiential processes in the brain and subconscious mind. Certainly, a being could have perceptual knowledge without any contribution from sensory experience. And of course, that idea has met with difficulties relating to the given, inference from experience to belief, and the nature of experience itself. Experience is problematic and clearly not necessary to successful knowledge acquisition. Some theorists think we are better off without it in epistemology; they think no knowledge could be based on it, whatever exactly it is. They would prefer to naturalize and normalize knowledge by eliminating experience. The role of experience in producing knowledge has been massively exaggerated.

But that is not the end of the story: there are good reasons to suppose that perceptual knowledge has other features of the a priori aside from experience-independence. For perceptual knowledge of the external world has everything to do with space and geometry, and geometry is an a priori science. This is a big subject, but I will be brief (it is familiar territory since Kant). Space is the form of our sensibility and geometry is the a priori science of space. Spatial perception is practical geometry. Perception is the science of geometry in action. The visual system calculates size and shape constancies on the basis of distance and other cues. We recognize shapes and compare them to each other. We detect edges and angles. We distinguish curves from straight lines. We see the world as consisting of geometrical objects.[1] Our senses know all about the geometry of space and the objects within it, probably innately. They are mathematical to the core. Geometry is synthetic a priori, so our senses are also synthetic a priori; they have insight into geometrical form. We know geometry a priori, so much of our knowledge of the spatial world is a priori. In fact, we have three sorts of geometrical competence: perceptual, cognitive-intellectual, and imaginative. These are not unrelated but interpenetrate. Our imaginative geometry is heavily a priori, but so is our perceptual geometry. In all three areas we cognize geometrically, and knowledge of geometry is a priori. Therefore, perceptual knowledge is a priori; it involves geometrical “reasoning”. This means that perception is mathematical through and through; it isn’t a simple matter of copying or imprinting, as the classical empiricists supposed. The eyes calculate; they don’t just mimic (they don’t mimic at all). So, perceptual knowledge is synthetic a priori. Combine this with independence from experience and you have a picture of an a priori faculty divorced from the apparatus of impressions and ideas copied from them. Perceptual knowledge is at its core cognitive mathematics. It isn’t raw sense data somehow imprinting themselves on a blank sensorium without any reference to the mathematical; it is more like deductive mathematical proof. Not just the “intelligent eye” but the “mathematical eye”. This is the a priori eye not the a posteriori eye—prior to experience not posterior to it. Perceptual knowledge is not made of experience (whatever that is) but removed from it and made of calculations. It is the kind of thing that rationalists dreamed of. The strict separation of intellect from sense is mistaken; the senses are intellectually active. The geometric faculty extends to them.[2]

What does this do to traditional epistemology? A lot. Empirical knowledge is a priori! Does this mean that all knowledge is a priori, given that analytic knowledge is also a priori? That would be one response, with some new distinction being erected, or none. We simply accept that all knowledge is a priori—a totally rationalist epistemology. That would not be unprecedented, what with Plato et al. But there is another possibility that should not go unmentioned, namely that analytic knowledge be regarded as a posteriori in some respects.[3] Suppose we think of introspection as a type of proprioception—both are types of self-knowledge. We have proprioceptive knowledge of the body, of the person, and of the mind—as in “My arm is bent”, “I am going to the store”, and “These thoughts are chaotic”, respectively. Among the introspective cases we have knowledge of conceptual truths like “Bachelors are unmarried males”. Then we can ask whether such knowledge is a posteriori or a priori. It could be argued that analytic knowledge is based on something akin to experience, because there cannot be introspective knowledge without conscious sensations; there is no subliminal perception of one’s own mind, or blindsight knowledge, or unconscious knowledge. You feel your own mind when you know it, necessarily. Second, introspective knowledge is not spatially configured, not geometrical, not mathematical. It fits the old empiricist picture of knowledge as qualitative and reactive (“Pain here”); there is no calculation or ratiocination in it. When you have analytic knowledge, you are not making spatial inferences or recognizing shapes. Thus, your knowledge is not of the typical a priori form. If so, it is a posteriori knowledge, as the empiricists supposed (“knowledge by reflection”). What they missed is that knowledge of analytic truth is a type of introspective (proprioceptive) knowledge—the introspection of concepts and meanings. Analytic knowledge turns out to be a posteriori, while synthetic knowledge turns out to be a priori. Or better, what remains of the old dichotomy is best characterized in this way. No natural name for the new distinction comes to hand, and maybe there is no sharp distinction that preserves the old dichotomy. We might just say that all perception-based knowledge has an a priori character, while analytic knowledge lacks the geometrical form and content of perceptual knowledge. The trouble is that the philosophers of yore who forged the old distinction had a naïve picture of perceptual knowledge and a narrow model of analytic knowledge. If we want to sum this all up, we could say that a priori knowledge is far more pervasive and basic than we thought. Geometry is the general form of our synthetic knowledge, while analytic knowledge approximates more to the old stimulus-response theory of empirical knowledge.[4]

[1] We also see the world as colored, and color is not part of orthodox geometry. But we see colored patches as extended and shaped, so they are still subject to a geometric form of understanding. We perceive secondary qualities as laid out in space.

[2] Perception also involves statistics and probability theory, but I won’t go into this.

[3] See my “Analytic A Posteriori”.

[4] One way to put the point of this essay is that human knowledge falls into natural epistemic kinds in which reality and appearance can diverge. Perceptual knowledge can seem experience-dependent because of its experiential phenomenology, but in its real essence it is a priori, owing to its geometrical substructure. We must look beneath the surface to see its underlying nature, as revealed by science. Similarly, knowledge of analytic truths can seem a priori because of its stripped-down phenomenology, while really being a posteriori. We can’t go by how it strikes us now; we must remember its genesis and mental development. The old distinction between a priori and a posteriori knowledge is too surface-oriented; we must consider its deeper structure. Thus, synthetic knowledge is geometrical-computational-mathematical, while analytic knowledge is reflexive-immediate-sentient. Zombies can have synthetic knowledge but not introspective analytic knowledge. Self-knowledge is inherently conscious-sensational, but external-world knowledge is not. Roughly.