Return to Human Knowledge and Human Nature contents  

 

4. THE EMPIRICIST CASE AGAINST NATIVISM

 

Few empiricists besides Locke have bothered to present explicit arguments against nativism, mostly taking its falsity for granted. In the present chapter I shall consider whether Locke's case is really powerful enough to justify such an attitude.

 

Locke on Innate Knowledge

Locke's main argument against the existence of innate knowledge in Book I of the Essay, is that the various supposed innate truths (for example, 'God exists' and 'Whatever is, is') are not in fact universally assented to. He cites the examples of children and madmen, many of whom will not assent to such propositions if they are put to them. Locke is therefore assuming that innate knowledge would, if it existed, necessarily have to be present in everyone, and also that it would have to be available to consciousness from birth. As we shall see, both of these assumptions are false.

            The argument from madmen is certainly a bad one. To say that something is innate for human beings, is to say that all normal members of the species will possess it, not that all without exception will do so. Consider, for example, the fact of possessing ten toes. This is surely an innate feature of human beings. But some humans will in fact have less, and some more. Some may have lost a toe in an accident; and occasionally babies are born with an extra toe, or with a toe missing. So the fact that madmen (who on any account of the matter are not normal human beings) lack some knowledge that the rest of us possess, does not show that such knowledge is not innate.

            The argument from children is also unsound, but needs to be handled somewhat differently. A natural first response to it, would be to claim that innate knowledge might be latent in children - that is to say: it is there, but not yet available to consciousness. Locke anticipates this reply, and responds by adopting what might be called 'The principle of mental transparency'. He claims that there cannot be anything in the mind which the subject is unaware of. But this principle is surely indefensible. Nor need we commit the anachronism of appealing to Freudian theories of the unconscious to show as much. For consider the every-day phenomenon of temporary memory loss. You may know that you know your mother's birthday, but be unable for the moment to recall it. Then an hour later you may be able to remember it again. If we accepted the principle of mental transparency, we should have to say that you started by having the knowledge of your mother's birthday, then you lost it, and then you acquired it again without any process of learning. This is surely absurd. Rather, we should say that the knowledge was in you throughout, but that for a short period it was not accessible to consciousness.

            While Locke's explicit argument against latent innate knowledge is inadequate, it may seem that he has a valid point nevertheless. For are we really prepared to accept that a new-born infant has its head already stocked with a range of actual (if as yet merely latent) knowledge? One way of developing this point is to notice that you cannot have knowledge of something unless you also believe it. (You cannot know that the Earth is getting warmer unless you at least believe that the Earth is getting warmer.) And what constitutes a mental state as a state of belief is that it is apt to interact with your other mental states (particularly desires and intentions) in such a way as to control behaviour. Thus what makes the difference between believing that the Earth is getting warmer, as opposed to hoping that it is, for example, is that you are prepared if necessary to act on it. You may consider moving to Iceland, or stop using aerosols (depending upon your other beliefs, and on what it is that you want). Then since an infant cannot manifest any of the appropriate behaviour, it would seem that it cannot have articulate beliefs either.

            This argument is perhaps overly swift, however. For even if an infant cannot be said to be born already possessing beliefs, it may be that it is born with a stored stock of propositions. These may start by being inert, but then become knowledge as soon as the child is old enough to be conscious of them. Yet even this may strike one as an extraordinary hypothesis. The idea that the head of an infant is already stocked with a range of articulate propositions may strike one (and does strike me) as just wild. At any rate, if this were the only form which nativism could take, then the burden of proof would surely fall squarely on the nativist to provide some convincing argument for their view. Locke would be quite right that there is a strong presumption against nativism, unless and until we are shown otherwise.

            In fact, however, there is quite a different sense in which innate knowledge may be latent from birth. For it may be innately determined that children develop such knowledge at a certain stage in the course of their normal growth, irrespective of details of education and experience. Compare, for example, the possession of pubic hair. This is surely an innate feature of adult human beings. But it is not present from birth, only making its appearance with the onset of puberty. Similarly, then, in the case of knowledge: it may be that it is not present at all at birth (there being no stored stock of propositions), but that it is innately determined that such knowledge will make its appearance at some particular stage in normal cognitive development.

            Locke himself considers ideas related to this one. For he argues against the suggestion that truths may be innate in the sense that one has an innate capacity for knowing them. He also argues against the thesis that innate knowledge may make its appearance in the mind when the subject first attains the use of reason. But his only response to the first suggestion, is that one cannot then distinguish between truths which are learned and truths which are innate, if 'innate' just means that one has an innate capacity to know them. While this may be true, if 'capacity' is understood broadly, it is not an objection to the developmental thesis sketched above. For if the truths which make their appearance subsequent upon experience could not have been learned from that experience, then this will be sufficient reason to count them as innate, as we shall see in the next section. And as for the suggestion that innate knowledge may make its appearance with the onset of reason, Locke's response is mostly to chip away at this as a proposed time at which innate knowledge should appear, which in no way touches the general idea behind developmental nativism.

 

Varieties of Innateness

We have noted that while one form of nativism claims (somewhat implausibly) that knowledge is innate in the sense of being present as such (or at least present in propositional form) from birth, it might also be maintained that knowledge is innate in the sense of being innately determined to make its appearance at some stage in childhood. This latter thesis is surely the most plausible version of nativism. Indeed, there seems no particular reason why we should presume its falsehood. For given that much of the physical growth and development of human beings is innately determined, why should the same not be true of our cognitive development also? It is therefore not at all obvious that the burden of proof is on the defender of this form of nativism to make out their case, rather than on Locke to show that all knowledge is in fact acquired from experience. But we should now notice that this second hypothesis, in turn, admits of two alternative versions.

            First, a belief might be innate in the sense that it is acquired in any course of experience sufficient for forming beliefs at all. To adopt this hypothesis is to allow (as seems likely) that an infant sensorily deprived from birth would never have any knowledge or beliefs, innate or otherwise. We would be allowing that some initial experience is necessary for the mind to operate normally, and for innate knowledge to make its appearance. But we would claim that it does not matter what experiences the child has, provided that they are sufficiently rich and varied for it to acquire at least some beliefs. Let us call this 'The hypothesis of general triggering of innate knowledge' - 'general triggering' because almost any experience will serve, the content of the experiences needing to bear no relation whatever to the content of the innate beliefs which make their appearance as a result.

            The second sense in which an acquired belief might be innate, would be if its existence was inexplicable on any model of learning, its content being such that it could not have been learned from the experiences which give rise to it. To adopt this hypothesis would be to allow, as before, that a sensorily deprived infant would never come to have any beliefs. But it would also be to allow that quite specific types of experience may be necessary for a given innate belief to make its appearance. For example, it might be claimed (as we shall see in chapter 6) that our knowledge of grammatical structure is innate, but that some experience of language is necessary to trigger this knowledge into existence. Or it might be claimed (as we shall see in chapter 8) that our knowledge of the rudiments of human psychology is innate, but that some experience of other humans is necessary for it to make its appearance. (So Tarzan brought up by apes in the jungle would never acquire either of these sorts of belief, although his experience would be sufficiently rich for him to have many other beliefs.) But still the knowledge in question could reasonably be said to be innate, provided that it is impossible to see how it could have been learned on the basis of the experiences in question - for example, if no combination of memory, induction, and inference to the best explanation could have generated that knowledge from such a meagre basis.

            Let us call this second version of developmental nativism 'The hypothesis of local triggering of innate knowledge' - 'local' because specific (content-relevant) types of experience are necessary for the knowledge to make its appearance; but still 'triggering' because the content of the knowledge acquired is so related to the content of the experiences which give rise to it, that the former could not have been learned from the latter. Most of the arguments in support of nativism which we shall consider in later chapters are in fact arguments for local triggering. Notice that Locke himself provides no direct arguments against either of these forms of developmental nativism.

 

Locke on Concept Acquisition

The arguments considered above from Book I of the Essay are not the only ones which Locke uses against nativism. Indeed, they do not even constitute his main argument. Rather, he thinks that it will be sufficient to refute nativism if it can be shown that the hypothesis of innate knowledge is an unnecessary one - that is to say: if he can provide an alternative account of the genesis of all knowledge in experience. This is his strategy throughout the remaining three Books of the Essay.

            Now, although I denied above that there is a general presumption against the truth of nativism (at least in either of its more plausible developmental versions), it does seem to me that Locke is on strong ground here. For suppose that both Locke and the nativist could provide equally good explanations of the knowledge we actually possess. In the case of the nativist (but not of Locke) there would still be something left over that needed explaining - namely, how it is that some of our knowledge comes to be innate. We already know that some knowledge is derived from experience, and we know roughly how this takes place (through perception). So the hypothesis that all knowledge comes from experience leaves nothing further in need of explanation. In contrast, if some knowledge is innate, it still remains to be explained how it comes to be so. Therefore, other things being equal, Locke's hypothesis is the one to be prefered, since it leaves less in need of explanation.

            In fact Locke does not focus very directly on explaining the acquisition of knowledge from experience. Rather, most of his efforts are directed towards showing how all our concepts (ideas) may be derived from experience. He here assumes, I think, that no knowledge can be innate if no concepts are.¹ So if he can show that the hypothesis of innate concepts is an unnecessary one, he believes that he will thereby have shown that the hypothesis of innate knowledge is also unnecessary.

            The idea that innate knowledge requires innate concepts is certainly a very plausible one. For it is clearly the case that knowledge itself requires concepts. Knowledge (at least in the sense which concerns us) is essentially propositional - it is always knowledge that such-and-such is the case. So you cannot possess such knowledge unless you also possess the concepts involved in the proposition that such-and-such. You cannot know that grass is green unless you possess the concepts grass and green. But in fact, whether this dependence of knowledge upon concepts extends also to the dependence of innate knowledge upon innate concepts, turns on what exact concept of the innate is in question.

            Clearly there can be no innate knowledge in the sense of stored propositions unless there are also innate concepts. For the constituent concepts of those propositions will also have to be present in the mind from birth if the propositions themselves are. Nor can there be general triggering of innate knowledge by experience unless there is also general triggering of innate concepts. For remember that the experiences which give rise to such knowledge need bear no relation to it in content, and so would not be the sort of experience from which one could derive (that is, learn) the concepts in question either. Matters are quite different, however, when it comes to the thesis of local triggering. For on this account, it may be that concepts are learned from appropriate and relevant experience, but are then triggered into items of knowledge which go far beyond the content of the experiences which gave rise to the constituent concepts. It may thus be that while no concepts are innate, some knowledge is. For it may be that while the data which gives rise to our knowledge of some subject-matter is not sufficient to sustain the view that we learned that knowledge, it may still be sufficient to support the view that we learned the constituent concepts.

            Thus Locke's assumption that there can be no innate knowledge without innate concepts is false. In which case his arguments against concept-nativism will not necessarily undermine knowledge-nativism. Nevertheless, it is worth considering his theory of concept-acquisition in its own right. For after all, most nativists have in fact held, not only that some knowledge is innate, but also that some concepts are. Moreover, the most plausible case of innate knowledge, to be defended in chapter 8 (namely, knowledge of the general principles governing our own and other people's psychology), will in fact be such as to involve the claim that there are innate concepts also.

            Locke believes that we derive simple concepts from experience by abstraction (complex concepts can then be formed from simple ones by definition). The idea is that from a sequence of experiences we are to isolate the various features they have in common. We are to do this by ignoring differences of time, context and so on, and by noticing and isolating recurring aspects. Consider this analogy. Suppose that you are taking part in one of those psychological experiments where you are handed a sequence of cards on which geometrical shapes of varying colours have been printed. Some are red triangles, some green triangles; some are blue circles, some red circles; and so on. As you are handed each card the experimenter says either 'This is a grink' or 'This is not a grink'; your task being to acquire the concept grink. What you would do, of course, would be to attempt to spot resemblances between those cards which contain a grink, using those which do not to disconfirm your hypotheses. This will be how Locke thinks of concept-acquisition in general. For what you would in effect be doing in this experiment, is abstracting from the sequence of your experiences the common feature of all grinks.

            As it stands, however, Locke's account of concept-acquisition appears viciously circular. For noticing or attending to a common feature of various things presupposes that you already possess the concept of the feature in question. Thus in order to notice that Peter, Paul and Mary have something in common - namely, that they are all freckled - you must already possess the concept of being freckled. (This is not to say that you must already have a word for 'freckled'; but you must know in general how to distinguish people who are freckled from people who are not.) Even more obviously, if one thinks of concept-acquisition as being a matter of formulating and testing hypotheses (for example, 'Is a grink any four-sided figure which is either red or green?'), then some concepts must already be possessed in advance (namely, the concepts which figure in the hypotheses).

            While Locke discusses abstraction in terms which suggest that the processes involved will be conscious ones, it would be open to him to respond to the above objection by denying this. He could say that his theory is really that prior to acquiring any concepts there are processes which are somewhat like those of noticing resemblances, ignoring differences of context and so on, only that they are nonconscious ones. This saves the account from circularity. But notice that it is now apparently to concede that there are innate concepts after all, since nonconscious noticing of a resemblance presumably requires a nonconscious concept of the feature in question. So it is doubtful whether Locke would find this defence of his position satisfying. But then how else is he to defend it? Whilst he remains wedded to the language of abstraction, it appears that he must be committed to the mind's possession of conscious or nonconscious concepts prior to the process of abstracting a concept from experience.

 

Complex Concepts

If we set aside the worry about circularity outlined above, then Locke's theory can seem quite powerful. In particular, it has the resources to rebut many of the arguments presented by rationalists in favour of the thesis that there are innate concepts. For example, in the Phaedo Plato argues that the concept straight must be innate because, first, judgements of 'almost straight' presuppose prior grasp of the concept straight, and secondly, because no object in the world of our experience can be better than almost straight. Locke can reply by denying the implicit assumption that straight is a simple idea, which would have to be learned directly from experience if it is learned at all. Rather, he can say that what we derive from experience is the comparative concept straighter than, by observing pairs of unequally straight things. We can then introduce the concept straight by definition, as a thing straighter than anything else could be.

            Locke can reply similarly to Descartes' argument in his third Meditation, that our concepts of God's perfections must be innate, since they plainly could not have been acquired on the basis of experience. Locke can maintain that what we acquire from experience are the various comparative concepts better than, more powerful than, more knowledgeable than, and so on. The idea of God may then be introduced by definition, as the one and only person who is better, more powerful, and more knowledgeable than anything else could possibly be.

            However, Locke's account of concept-acquisition does face a number of further difficulties. The most notorious is that there are many concepts which cannot be abstracted directly from experience, and yet which appear not to be definable in terms which can be so abstracted either. Consider, for example, the concept of causation, which was treated at length by Hume.² All that is immediately observable in a case of A causing B is that the one precedes the other. Hume suggests that the remainder of our concept is made up from observing regular concurrences of such events. Then to say that A caused B will be to say that it preceded it, and that all events of the same type as A precede events of the same type as B. Yet clearly this falls short of our intuitive concept of a cause, which includes the idea that causes somehow necessitate their effects. Hume has to conclude that this idea is an illusion, brought about by our own habits of mind in expecting an event of type B when we see one of type A. Since our psychology is such that we cannot avoid expecting B whenever we see A occur, Hume suggests that we mistakenly assume that it is B itself which is unavoidable, given that A has occurred. He claims that what we do is to project a feeling of psychological necessity on to the world.

            There are many reasons for rejecting Hume's account of the matter. One is that it tacitly assumes that we do have a concept of causal necessitation, if only as holding between psychological events. For what, otherwise, is the feeling of necessity a feeling of? Another point is that one may come to believe in a causal relationship between events after observing just a single instance, when there has been no opportunity for a habit of expectation to be formed. For example, I may see someone fall down stairs and come to believe that this caused their broken leg, although this is the first occasion on which I have observed anything of the sort to occur. Another argument is that we may wonder whether A caused B, where this is clearly more than a matter of wondering whether all events of type A precede events of type B. Yet in such cases we are obviously not wondering whether our minds are necessitated to expect B given an observation of A.

            In fact what is really involved in our idea of causation are, at least, counterfactual and subjunctive conditionals, as we have already noted in chapter 3. To say that A caused B is to imply that if A had not happened, then B would not have.³ It is also to imply that if in other sufficiently similar circumstances an event of type A were to happen, then an event of type B would happen also. Yet it is impossible to see how the concepts of such conditionals may either be abstracted from experience, or defined purely in terms which may be so abstracted.

            Indeed, it is arguable that there is more to the concept of cause even than this. For example, David Armstrong maintains that causation is best understood as a relation of necessitation between immanent universals.⁴ This relation is held to imply, and hence explain, the truth of the counterfactual and subjunctive conditionals mentioned above, rather than being constituted by them. Then it is because A is made up of properties which necessitate the properties making up B, that it is true that if A had not happened B would not have, and also true that if an event similar to A were to happen, so would an event similar to B. Moreover, Armstrong argues that the notion of causal necessitation has to be taken as primitive (that is, as indefinable). He also concedes that it cannot be acquired from experience, without I think noticing that he is therefore committed - very plausibly, in my view - to the claim that the concept of cause is innate.

            Other concepts besides causation have led to problems for classical empiricists. Thus both Berkeley and Hume drew the conclusion that our concept of mind-independent continuously-existing physical objects is illusory, because they were (rightly) unable to see how such a concept could be derived from experience, or defined in terms of concepts which can be so derived. Since one cannot have experience of an unexperienced object, it is hard to see how one could derive the concept of such an object from experience. But then neither can that concept be defined in other terms derivable from experience. The closest we could get would be to say that an unexperienced object is the continuously existing cause of our episodic experiences. But this is plainly too broad (quite apart from the problem of how we are supposed to have acquired the concept of cause). It cannot distinguish between the chair, as cause of my experience of it, and Descartes' all-powerful demon.

            It is worth remarking here just how powerfully these empiricists must have been convinced of the thesis that there are no innate concepts. For rather than give up this thesis, they were prepared to deny that it is possible for us to conceive of a mind-independent physical reality. But in the absence of any convincing argument in its support, the proper conclusion to draw is surely that it was their anti-nativism itself which is false.

 

Simple Concepts

In fact problems arise for empiricists even in connection with the very simplest concepts, such as those of colour. For it is false that all instances of a given colour share some common feature. In which case we cannot acquire the concept of that colour by abstracting the common feature of our experience. Thus consider the concept red. Do all shades of red have something in common? If so what? It is surely false that individual shades of red consist, as it were, of two distinguishable elements: a general redness together with a particular shade. Rather, redness consists in a continuous range of shades, each of which is only just distinguishable from its neighbours. Acquiring the concept red is a matter of learning the extent of the range.

            Nor will it help to say, as Berkeley does,⁵ that the concept of red is not an idea of a common feature abstracted from differing red things, but is rather an idea of a particular shade of red which is then used as a representative of the whole range. For there is nothing in the particular shade itself which can give you the extent of the range. Nor can this be learned from experience. There is nothing in experience which can tell you where in the spectrum red begins and ends. It would seem that the boundaries between the various colours must somehow be specified innately, unless they can be explained as taught social constructs of some kind.

            This last remark suggests an alternative strategy which is available to empiricists for explaining our possession of concepts, and it is worth considering why they have not, in general, been inclined to pursue it. The strategy would be to appeal, not to abstraction, but to some sort of linguistic training. Why should empiricists not say, consistently with their anti-nativism, that we are taught to classify things in the way that we do? On this view, possession of concepts would still arise out of experience by a process of learning. But the experience in question would not be (or not primarily) of the things to which the concepts apply, so much as of the norms which are prevalent in the person's language community. A child's first fumbling use of words would gradually be refined and perfected through a process of reward and correction. It would be, for example, by mistakenly describing an unripe tomato as 'red' and being put right by its parents, that a child would acquire its grasp of the boundaries between the colours.

            However, the obvious question arising for such an account would be this: from where did our teachers, in their turn, acquire their concepts? The answer, in terms of the theory, is equally obvious: from their teachers. But now we have a problem. Plainly the sequence of past teachers cannot be infinite, since the human race has not always been in existence. So it appears that there must have been some person, or group of people, who were the first to use simple concepts, without having been taught to do so. But then we shall be landed back with some version of abstractionism again, if we are to avoid commitment to nativism. For those first users of concepts will somehow have to have acquired their concepts directly on the basis of their experience. Certainly the problem of concept-acquisition cannot be solved merely by pushing it back into the past. It is for this reason (among others)⁶ that most empiricists have not taken very seriously the idea that we acquire concepts through linguistic training.

            However, we should beware of the suggestion that there must have been a first concept-user, if present concept-users get their concepts from others. For compare the following. What makes someone a member of the human species? One obvious answer is: being born of human parents. This looks equally vulnerable to the charge of merely putting a problem off, on the grounds that there must have been at least two first humans who were not human by virtue of having human parents. But in fact, as we now know, creatures that were recognisably human evolved gradually, in small steps, from creatures which were not. So it is possible that something similar may hold in the case of concept-acquisition as well. It may be that what was recognisably a use of concepts evolved gradually, from the use of grunts and growls which were plainly non-conceptual. In which case an empiricist could explain concept-acquisition in terms of linguistic training, without having to be committed to some form of abstractionism in explaining how the first concepts were acquired. But this is, so far, merely a promise. It remains for an empiricist to show how concepts could arise gradually out of something non-conceptual. We shall return to the issue in chapter 7. For the moment, it is enough to have noted the weaknesses in the classical empiricist accounts of concept-acquisition.

 

Why be Anti-Nativist?

While the empiricist case against platonism is powerful, as we saw in chapter 3, its case against nativism is very weak by comparison. Not only are the direct arguments against nativism unsound, but the attempt to explain how all concepts may arise out of experience itself faces severe difficulties. This is not to say, of course, that nativism is then shown to be true. It is simply that the case against it is unproven. We may then remain puzzled as to why empiricists such as Locke and Hume should have been so convinced, nevertheless, that nativism must be false.

            It might be replied that there is no special problem about this: they were simply misled by bad arguments. But I find this response unsatisfying. For after all, Locke and Hume were both of them extremely intelligent men. So it remains possible, at least, that they may have had other, more powerful, reasons for rejecting nativism. Perhaps these may have gone unmentioned as a result of some sort of political expediency. For example, it may have been that their real reasons would have placed them in direct opposition to the Church.

            Thus, one hypothesis might be that the early empiricists' rejection of nativism was part of their more general Enlightenment belief in the perfectability of man, and should be seen in contrast with the traditional Christian doctrine of original sin. But this proposal is hardly very satisfactory either, since there is no intrinsic connection between perfectability and anti-nativism. Enlightenment thinkers could equally well have maintained that while we have innately-given knowledge, and innate faculties which are structured in such a way as to embody information about the world, our knowledge and attitudes nevertheless admit of indefinite extension and improvement.

            It is true that if the mind were literally a 'blank slate', as early empiricists seemed to maintain, then human nature would be almost unlimitedly malleable, for good or ill. The only constraints would be those of capacity (there may be limits to how much knowledge a human mind could contain, for example), and those imposed by the properties of the mental medium itself (some sorts of knowledge might be more difficult to acquire on the basis of general learning principles, for example). So the denial of nativism, if correct, would provide some sort of guarantee of human perfectability.

            Endorsements of nativism, in contrast, would admit of at least two possible versions, implying either perfectability on the one hand, or inherent imperfection on the other. But even given a prior commitment to perfectability, this would be a very poor reason for rejecting nativism altogether. For it is just as plausible that the explanation of perfectability might be an innately given but indefinitely improvable nature. Moreover, the current proposal would require us to attribute to empiricists a belief in human perfectability which is apparently lacking in independent support, but which stands in need of it. For they cannot simply take for granted the falsity of Christian versions of nativism, unless they have some independent reason for rejecting nativism as such.

            A rather different sort of proposal would be that the early empiricists' reasons for rejecting nativism might have formed such a fundamental part of their outlook as never to have been consciously articulated. Certainly it is common enough in philosophy for thinkers to allow themselves to overlook the weaknesses in their explicit arguments for a thesis, precisely because they have already become convinced of the truth of that thesis on other, and less readily articulable, grounds. Charity requires us to hope that something of this sort may be true in connection with the classical empiricist rejection of nativism. I shall return to the issue in the next chapter, and again in chapter 9.

 

On to chapter 5.