Classical empiricism maintains two main theses: all concepts are acquired by experience and are not innate; and all knowledge is based on experience. Classical rationalism by contrast maintains that some or all concepts are innate and not derived from experience; and some knowledge at least is not based on experience. These are independent propositions: it would be possible to hold that all concepts are innate and that all knowledge is based on experience. That is, one could be a rationalist about the acquisition of concepts (including the content of experience) and an empiricist about the basis of factual or propositional knowledge. Everything about the way we perceive and understand the world could be innate, but still we only know truths about it by means of perception. This position—rationalist empiricism—avoids the traditional empiricist claim that our means of representing the world is derived by abstraction from sense experience, but it accepts that all justification is experiential. How we see things comes from within the mind, but knowledge of facts requires commerce with the world outside the mind (except when it is knowledge of the mind). The constituents of thought (knowledge, perception) are innately given, but all factual knowledge results from some sort of experience of reality: nativism about concepts, empiricism about knowledge. Alternatively put, ideas are innate but beliefs are acquired by experience. This is in many ways an attractive position given the difficulties of the empiricist theory of concept acquisition, i.e. abstractionism, and given the obscurity of rationalist accounts of knowledge. It allows us to have a uniform theory of knowledge that possesses a fair degree of intelligibility while avoiding the problems inherent in the idea that concepts can be derived from objects by some sort of imprinting or copying procedure. The rationalists were right about one thing; the empiricists were right about another. Yet the position is seldom if ever occupied: instead we get a rigid divide between pure experience theories and pure reason theories. The rationalist will accept that some knowledge is based on experience, but the knowledge that isn’t so based is conceived as radically different from empirical knowledge. Wouldn’t it be better if we could develop a more unified picture of human knowledge without such a sharp dichotomy? So we should look with favor on the project of articulating a rationalist empiricism that has a more workable theory of concept acquisition and a more seamless picture of knowledge, thus combining the best elements of both traditions.
But that project faces a formidable challenge: how can knowledge that is traditionally deemed a priori be assimilated to empirical knowledge? How can mathematics, say, be based on experience? The senses alone cannot deliver mathematical knowledge; for that we need a faculty traditionally labeled “Reason”. One approach to this problem for a general empiricism is to claim that human knowledge forms a seamless whole that collectively faces the test of experience. We adopt a holism of human knowledge and then claim that mathematics is justified by its place in the totality of knowledge as it confronts experience (see Quine). The advantage of this approach is that we can apply an intelligible framework for understanding all knowledge (the senses operating in intelligible ways) and at the same time deliver a uniform picture of all knowledge (we don’t have to postulate two completely different ways of knowing). Such a picture could be combined with nativism about the building blocks of knowledge to produce a streamlined and appealing theory of the whole of human knowledge: so we get the best of both worlds, rationalist and empiricist. But the holistic doctrine is hard to accept, and the account of the actual epistemology of mathematical knowledge quite implausible. Couldn’t we find a way of thinking that achieves the same end but without the implausible claims? Can’t we broaden the notion of experience so as to include mathematical (and other) knowledge—why must it be restricted to sense experience, and indeed to bodily sense experience? The old empiricists held that the five human senses are the sole origin of knowledge, with vision preeminent, but why insist on that limitation? Certainly we would want to include other possible senses, and why restrict ourselves to these? For one thing, we need to make room for introspective knowledge—knowledge by reflection, as the empiricists put it. Here it is natural to invoke the idea of an inner sense: I can sense my own mental states and come to know about them. I thus come to know about them by experience (in one sense of “experience”)—not by logical deduction or revelation or sacred texts. I can receive data about my own mind: information, evidence. It seems to me that I am in pain and so I form the belief that I am in pain. The OED gives three definitions of “experience”: “practical contact with and observation of facts or events”, “knowledge or skill gained over time”, and “an event or activity that leaves a lasting impression”. Each of these can fit the case of introspective knowledge: I can be practically aware of my mental state, can gain knowledge of it over time, and it can leave a lasting impression on me. We don’t need to limit the notion of experience to the particular types of experience characteristic of the five bodily senses. Nor is the notion of a sense so limited: that is just empiricist dogma. So we already extend the concept of experience beyond the five senses, along with allied concepts. What about the concept of mystical experience, or experiences of cosmic despair, or experiences of the oneness of nature?
How should we view the mental faculties involved in producing so-called a priori knowledge? There are four cases to consider: knowledge of meaning, logical knowledge, mathematical knowledge, and ethical knowledge. I propose assigning these to separate mental faculties: they are not homogeneous and deserve to have their own specific module. Then our question is whether they can be seen as lying on a continuum with other types of knowledge; in particular, can they be said to involve types of experience? I think they can: we have semantic experiences, logical experiences, mathematical experiences, and ethical experiences. That is, our consciousness can be affected in specific ways according to the relevant subject matter: there is something it is like to think about these things. When we acquire these types of knowledge our consciousness plays a role—we have a kind of basic awareness of the properties and facts in question. I am aware of what my words mean; I feel the compulsion of a logical deduction; I notice mathematical relations; I have a sudden sensation that such and such would be very wrong. Of course, these states of consciousness are elusive and hard to put into words, but it is undeniable that consciousness reflects the subject matter we are thinking about. We learn about these things over time by attention and practice, so that the knowledge comes more easily to us; and we do it without divine assistance or by recourse to supposed authorities. The individual using his or her natural faculties can come to have knowledge of the kinds in question—a key tenet of the empiricist philosophy. So why not say we have such knowledge “by experience” (notmeaning by the kind of experience proper to the five senses of the body)? We just need a more relaxed and inclusive notion of experience—one that is not at all alien to common sense. Then we can formulate a more liberal form of empiricism that includes types of knowledge usually labeled a priori. Knowledge arises from the individual’s natural faculties operating without divine assistance or deference to supposed authorities: so we know these things by “personal experience” not by external means (custom, the church, our parents, God’s messages inscribed on the soul). Our own consciousness is the source of all our knowledge—whether employed about the physical world or about more “abstract” matters. We have experience of many things and thereby come to know about them, not all of it connected to the five senses.
There is a kind of epistemological spectrum at work here. Suppose we think of touch as the most basic and tangible way of knowing things: things actually come into contact with our body in a mechanical manner (mechanism applies to knowledge by means of touch). Then we might rank taste and smell as the next senses on the continuum, since they too involve actual contact with the sense organs. Hearing comes next with sound waves in the air impinging on the tympanic membrane etc. Sight is a further step away from the paradigm of touch, being a distance sense (the idea of photons hitting the retina is far removed from the idea of observable bodily contact and is not part of the ordinary concept of vision). The next step brings us to introspection, in which the idea of physical contact starts to look misguided (at this point some people start to speak of the a priori). Not so far away from this we reach knowledge of meaning, which strikes us as a form of introspective knowledge, since meaning is a psychological fact of some sort. Closely following this we get to logical knowledge, which flows from knowledge of meaning, logical consequence being a type of meaning relation. Then we have a bit of a leap to mathematical knowledge: now we seem to be speaking of a realm of objects bearing properties about which we have information, but which make no contact with our body at all. Finally, we reach ethical knowledge: here we lose any analogy to the case of touch, and mechanism has been thoroughly defenestrated. There isn’t a sharp dichotomy, more a series of similarities and differences. In all cases we have some type of experience proper to the subject matter in question—some condition of individual consciousness—and knowledge springs from this basis. So empiricism applies to all cases—relaxed non-bodily empiricism. It is true that mechanism runs out of steam half way up the ladder, thus leaving us bereft of a theoretical structure that seems to render things intelligible to us; but mechanism is a doctrine long ago abandoned in physics and shouldn’t shape the way we do epistemology. Is it a tacit mechanism that lies behind our tendency to favor old-fashioned empiricism? All knowledge, we think, must spring from mechanically intelligible causes, and the bodily senses, particularly touch, conform to this model (or seem to). But once we abandon that ideal we can allow that all knowledge belongs on a continuum: it is all based on experience in the broad sense. Thus empiricism turns out to be true of all knowledge no matter how “abstract” or “a priori”. In fact no knowledge is a priori in the sense of being independent of all experience—though it can be independent of the experiences of the five senses (but so what?). Knowledge is a priori only relative to a type of experience; in this sense knowledge obtained by vision, say, is a priori relative the experiences delivered by the other four senses. We have a mathematical sense that produces mathematical experiences, and that sense is a priori relative to the other senses: it doesn’t depend on them. That is, we have conscious data in all cases that lead to knowledge—sense-data if you like (if we accept that we have a mathematical sense and a moral sense). The main dogma of old-style empiricism is the doctrine that only the five bodily senses deserve to be called senses. But a sense is just a faculty for receiving and processing information—for generating knowledge. So it is possible to be as rationalist as you like about mathematical and moral knowledge and still be an empiricist about them. Someone who thinks that we literally see numbers (but not with our physical eyes) is a firm rationalist but at the same time a staunch empiricist (he believes that mathematical knowledge springs from the unaided individual consciousness). He thinks we need more than just mathematical concepts in order to have propositional mathematical knowledge—we need to do the kind of seeing he has in mind—so he thinks that experience is necessary to knowing mathematical facts. His view is that mathematical concepts are innate but we also need mathematical experience in order to obtain actual mathematical knowledge—so he is a rationalist empiricist about mathematics. And he does not appear to contradict himself. Similarly an ethicist can hold that moral perception is necessary for moral knowledge, not just a stock of innate moral concepts; so she is in effect an ethical rationalist empiricist (we don’t get our moral beliefs from religion or tradition but from our own inner consciousness). She might indeed stress the importance of moral experience, so affirming her commitment to moral empiricism (it’s not all a matter of logical deduction in the style of Kant). Innate endowment gives you the building blocks; experience enables you to form these into actual knowledge. So it might reasonably be maintained: the result is a mixture of rationalist and empiricist tenets, suitably modified and extended. A nice result is that there isn’t the kind of sharp dichotomy envisaged by traditional epistemology: all knowledge is experience-based, but the experiences are of different types. Plato’s slave boy in the Meno learned Pythagoras’s theorem by experience (guided by Socrates), i.e. by applying his mind about geometrical concepts; he didn’t know this theorem before encountering Socrates that day—though he did already possess the necessary concepts. He is not essentially different from someone coming to know that it’s raining: such a person knows this fact by experience but all the concepts he employs are innately based (as we are supposing). We thus obtain a pleasingly uniform account of knowledge that builds in the best insights of rationalism and empiricism. The empiricist can be right that no knowledge of facts is strictly speaking innate, since some kind of eliciting episode of conscious experience is needed; yet the rationalist can also be right that all knowledge depends on a stock of innate concepts not put there by experience. Babies don’t actually have explicit mathematical and ethical knowledge, i.e. true justified beliefs in these things, but they do have the conceptual resources to have such knowledge once suitable experience comes their way. Both sides win: knowledge is always a mixture of the innate (rationalism) and the experiential (empiricism). All propositional knowledge is dependent on experience, but no concept is so based.
 We don’t have to say that every complex concept is innate qua complex concept; we can hold that all primitiveconcepts are innate. We can thus allow for post-natal concept combination. It will still be true that all concepts are composed wholly of innate concepts—none is acquired by any process of abstraction from sense experience.
 We might also list a modal sense and an aesthetic sense. It seems to us that some things are necessary and some contingent, and it seems that some objects are beautiful and some are not. We accordingly form beliefs about the modal and the aesthetic based on these appearances. Neither of these senses can be assimilated to the five bodily senses. Note that we don’t just de novo form beliefs and possess knowledge in these cases (as in others); we also experience presentations—ways things seem, appear, strike us. This is part of what makes it natural to talk of a sense: there is a way things seem and a belief based on this. Thus the concept of a sense has a wider application than traditional empiricists have supposed (it need not be causal, for example). Consider also the proprioceptive sense and the temporal sense: neither of these is quite like the Big Five and yet they are senses in good standing.
 In the case of knowledge of language, i.e. grammar and syntax, we might hold that the framework of categories and rules of combination is innate but that we only form propositional knowledge of language by being exposed to linguistic data in the course of our upbringing. We are not, on this view, born knowing that such and such a string is grammatical, but we are born with a mastery of the concepts and principles that enter into such judgments. (The case of knowledge of grammar may not be typical of other types of knowledge traditionally considered in epistemology, and requires a separate discussion.)
 If we ask who was basically in the right in the traditional debates about rationalism and empiricism, I would say the rationalists. The type of empiricism I defend here is quite far removed from anything the old empiricists had in mind: they were dedicated to the five bodily senses and insistent on the sensory origins of concepts. The “empiricism” I defend is really thoroughly rationalist in spirit (though not in letter!). But it does provide a way to unify all propositional knowledge conceptually: all such knowledge is true experientially justified belief.