Both empiricists and Aristotelians hold that “nothing is in the intellect that was not first in the senses,” as the hoary Aristotelian dictum puts it. But they disagree about what this means.
For Aristotelians, sensation leads to the formation of “phantasms” (very roughly, mental images), and from these objects of the imagination, the intellect in turn abstracts universals. So, for example, the perceptual experience of various triangles leads to the formation of various mental images of triangles, and from these the intellect derives an abstract general concept of triangularity. Such general concepts are nevertheless irreducible to mental images. For any mental image applies, strictly speaking, only to certain members of a class; for example, any particular mental image of a triangle is necessarily going to be either of an isosceles , scalene, or equilateral triangle, of a black, blue, or green triangle, and so forth. But the abstract concept of triangularity applies to all triangles without exception. Mental images are vague and indeterminate when they concern something complex; for instance, a mental image of a chiliagon (a 1,000-sided figure) cannot be distinguished from a mental image of a 1,002-sided figure, or even from a mental image of a circle. But the concept of a chilaigon is clearly distinct from the concept of a 1,002-sided figure or the concept of a circle. There are things we can form no mental image of at all – abstractions like law, economics, and love, for example – that we nevertheless have clear concepts of. And so forth.
Since Wittgenstein, contemporary philosophers have been pretty well inoculated against the error of supposing that concepts could possibly be reduced to images of any sort. (Though they hardly needed Wittgenstein to tell them that – Platonists, Aristotelians, Thomists, and rationalists have emphasized it for centuries, or even millennia.) And yet such a reduction, a view known as “imagism,” was a key feature of the classical empiricism of Locke, Berkeley, and Hume. (Sometimes it is disputed whether Locke was really committed to it, though not very plausibly, I think.) Berkeley and Hume rightly concluded that imagism entails that there are no abstract concepts, for every mental image is always inherently particular rather than universal. (Here as elsewhere Locke failed to see all the implications of his position.) So-called “general ideas” are for them really just particular ideas annexed to a general term (together, in Hume’s case, with a disposition to refer to all ideas resembling that particular one by the term in question).
This is, however, simply a reductio ad absurdum of the (imagist) theory that leads to it. For we simply and obviously do have abstract or universal ideas – triangularity is one of them, and there are countless others – and any theory that implies otherwise is thereby proved to be false. Furthermore, Berkeley’s and Hume’s attempts to defend their account of “general ideas” are notoriously unconvincing. Berkeley says that a particular idea stands for or represents other ideas of the same sort. But what is it for one idea to stand for or represent other things? And in virtue of what are the various represented particular ideas “of the same sort”? The obvious answer to the first question seems to be “It is to be an abstract concept under which particular things fall,” and, to the second, “In virtue of exemplifying the same abstract concept.” That is to say, the seemingly obvious answers to these questions presuppose the existence of the very thing Berkeley denies. Or at least, if these answers are wrong, Berkeley needs to provide some alternative answer, which he does not. He also says that a particular idea is made general when the mind selectively attends to some features of it and not others. But how can this “selective attention” to some features possibly fail to presuppose an abstract concept of that which has precisely those features and not the ones attended to? Again, the view seems to presuppose precisely what it denies, and again we are given no reason to suppose otherwise.
Hume basically takes all this on board, and only calls greater attention to the problem by saying that we have a disposition to call “resembling” ideas by the same general term. For we want to ask: In virtue of what do they resemble? By exemplifying the same abstract concept, surely; or if not, we need to be told why not, and we’re not.
Now you could take the imagist position in question to form the basis of an argument for nominalism: Concepts are images; images are particular rather than universal; therefore concepts are particular. (The claim that everything else is particular too would then be argued for independently as a “mop-up” operation, since the universality of concepts is arguably the hardest case for nominalism.) Or you could take imagism to presuppose nominalism: Concepts, like everything else, must be particular; mental images, which are particular, are the only things for concepts plausibly to be if concepts are not universal; so concepts must be mental images. Either way, imagism is just false – having been decisively refuted over and over again by (as I have said) Platonists, Aristotelians, Thomists, and rationalists – and both arguments are therefore unsound.
All the same, Berkeley and Hume rode this dead horse amazingly far. It arguably underlies all of Berkeley’s arguments against the possibility of material substance. For the concept of matter, Berkeley says, would have to be something that abstracts away from all ideas of sensation, and (he claims) there are no such abstract (i.e. non-imagistic, non-particular) ideas. Similarly, Hume’s skepticism about material substance depends on the idea that we cannot trace the idea of material substance to any impression – which would be the case only if we assume (as Aristotelians would not but classical empiricists would) that ideas or concepts cannot be anything more than a faint copies of sensations (and thus necessarily fail to transcend the particularity of sensation), rather than being abstractions from them (and thus capable of transcending that particularity). The claim that all ideas or concepts not only begin from sensation but cannot outstrip it also underlies Hume’s claim that we can have no idea of necessary connection because we have no impression of it. This can be so, again, only if ideas or concepts are faint copies of impressions (that is to say, if they are, more or less, mental images) and thus cannot transcend the particularity of the impressions from which they derive.
If the mind’s conceptual powers do outstrip the limits classical empiricism would put on it, however – as they obviously do given that we have many concepts that cannot be accounted for on classical empiricism – there is no reason to take seriously the bizarre and/or skeptical conclusions vis-à-vis causation, substance, etc. that classical empiricists derived from their false conception of these limits.