Challenges to Empiricism
Introduction
Harold Morick, State University of New York, Albany
1972, Wadsworth Publishing Company, Inc., Belmont, California

The Critique of Contemporary Empiricism

One of the main themes of philosophy and theory of science in the last twenty years is the critical assessment of the foundations of contemporary empiricism. Critics have questioned especially the empiricist conception of the role observation plays in our knowledge of matters of fact. The umbrella contemporary empiricism covers not only instrumentalism and operationalism in the physical and social sciences and "unliberalized" logical empiricism in philosophy but also the linguistic philosophy of Ludwig Wittgenstein and the Oxford school of Ordinary Language Analysis. In philosophy and theory of science, empiricism means the doctrine that experience rather than reason is the source of our knowledge of the world. This technical use of empiricism has obvious affinities with its ordinary use, in which it means the employment of methods based on practical experience rather than on theories or assumed principles.

Because the fundamental doctrines of contemporary empiricism are essentially refinements and modifications of David Hume's basic tenets in the theory of knowledge, the best entrance to an understanding of the foundations of contemporary empiricism is an understanding of the epistemological tenets of this great eighteenth century thinker. Although many of these basic ideas are not original to Hume, it is generally thought that he gave them their clearest, most coherent, and most uncompromising formulation. Let us turn to the groundwork that Hume laid for contemporary empiricism.

1. Hume

In Hume's view, the true nature and scope of ordinary and scientific knowledge can be revealed only by a "science of man," a study of the nature of man's ideas and of the principles of his reasoning processes. He thought that, as with all other sciences, the "science of man" must be founded on experience and observation.

Observation and experience teaches us that all thoughts are derived from past experience. For instance, unlike the blind, we can think about colors because we have seen them. In order to be precise, Hume expresses these points in semi-technical terminology: the nature of man's ideas are to be understood by contrast with the rest of his "perceptions" or mental contents -- namely, by contrast with his impressions. It is from man's impressions that all his ideas are ultimately derived. Roughly speaking, Hume's "impressions" are what we call sensations and feelings: seeing a spider, feeling the pain of excessive heat, feeling anger. To remember or to imagine any of them is, in Hume's terminology, to have an idea; ideas, then, are what we call thoughts. Impressions are our forceful and lively perceptions, whereas our ideas are nothing but their faint copies; a remembered pain, for example, is obviously less vivid a perception than a felt pain.

An empiricist criterion of meaningfulness or significance fol- lows from Hume's doctrine that every idea (i.e., thought) is a copy of a previous impression or set of impressions. According to Hume, we can think of a dragon without ever having seen one only because the idea of a dragon is a complex idea, a combination of simple ideas, each of which is a copy of some preceding impression. The idea of a golden mountain provides another example of a complex idea: "When we think of a golden mountain we only join two consistent ideas, gold and mountain, with which we were formerly acquainted."¹ That no one can have a particular simple idea unless he has experienced the impression of which it is the copy can be illustrated by the fact that blind men have no "idea" of colors, nor do deaf men have an "idea" of sounds. Thus, for Hume ideas divide up into the simple and the complex; all complex ideas are constructions out of simple ideas, and simple ideas are copies of impressions. This doctrine provides a criterion, or test of meaningfulness, a way of cleansing science and philosophy of empty verbiage. If we have "any suspicion that a philosophical term is employed without any meaning or idea (as is but too frequent), we need but inquire, from what impression is that supposed idea derived? And it it be impossible to assign any, this will serve to confirm our suspicion."²

In the following passage, Hume gives his conclusions as to what the science of man reveals about the principles of man's reasoning process.

"All the objects of human reason or inquiry may naturally be divided into two kinds, to wit, relations of ideas and matters of fact. Of the first kind are the sciences of geometry, algebra, and arithmetic, and, in short, every affirmation which is either intuitively or demonstratively certain. That the square of the hypotenuse is equal to the square of the two sides is a proposition which expresses a relation between these figures. That three times five is equal to the half of thirty expresses a relation between these numbers. Propositions of this kind are discoverable by the mere operation of thought, without dependence on what is anywhere existent in the universe.... Matters of fact, which are the second objects of human reason, are not ascertained in the same manner, nor is our evidence of their truth, however great, of a like nature with the foregoing. The contrary of every matter of fact is still possible, because it can never imply a contradiction and is conceived by the mind with the same facility and distinctness as if ever so comfortable to reality. That the sun will not rise tomorrow is no less intelligible a proposition and implies no more contradiction than the affirmation that it will rise. We should in vain, therefore, attempt to demonstrate its falsehood. Were it demonstratively false, it would imply a contradiction, and could never be distinctly conceived by the mind.... All reasonings concerning matter of fact seem to be founded on the relation of cause and effect. By means of that relation alone we can go beyond the evidence of our memory and senses.... A man finding a watch or any other machine in a desert island would conclude that there had once been men in that island. All our reasonings concerning fact are of the same nature."³

In the technical language of contemporary philosophy, Hume is here maintaining that all knowledge about the world is a posteriori. An a posteriori statement is one which can be confirmed or disconfirmed only by experience and observation; for example, we cannot confirm or disconfirm "The sun will rise tomorrow" without looking to see what happens. On the other hand, "Three times five is equal to half of thirty" is a priori -- that is, its truth is "discoverable by the mere operation of thought."⁴ But in Hume's view, any truth discoverable by thought alone is never about the world but only about internal relations between our ideas; in the language of contemporary philosophy, a truth of this sort is analytic. Roughly, an analytic truth is a statement true by definition. Precisely, an analytic truth, as opposed to a synthetic truth, is a truth the denial of which is self-contradictory. For instance, "Three times five is equal to half of thirty" and "All spinsters are unmarried" are analytic because their denials -- "Three times five is unequal to half of thirty" and "Some spinsters are married" -- are presumably self-contradictory. Such truths tell us nothing about the world; they inform us only about the internal relations among our ideas, e.g., that the complex idea of a spinster contains the idea of someone unmarried.

With this terminology for Hume's distinctions, we can state succinctly what Hume thought the "science of man" tells us about the principles of our reasoning processes. All reasoning about relations of ideas -- e.g., logic, arithmetic, geometry, algebra -- is analytic and based on the principle of noncontradiction; all reasoning about matters of fact -- e.g., physics, chemistry, everyday factual knowledge -- is a posteriori, and whenever it goes beyond present direct observation and remembered past observation, it consists in generalizing from experience on the basis of the principle of cause and effect. So scientific reasoning consists in empirical generalization, or, more precisely, in induction by simple enumeration: the process of estimating what can truly be ascribed to a whole class of things or events on the basis of what has been observed to be true of part of that class. The following is an example of induction by simple enumeration: Because many cases of smallpox have undergone careful inspection, and because it has been found that all of these cases have been accompanied by a certain virus, we are entitled to infer that the uninspected cases of smallpox are also accompanied by that virus. Hume's own term for this is experimental inference.

According to Hume, these are the principles of human knowledge. They set our knowledge on a firm foundation, and their application enables us to purge science and philosophy of empty metaphysical speculation.

When we run over libraries, persuaded of these principles, what havoc must we make? If we take in our hand any volume of divinity or school metaphysics, for instance, let us ask, Does it contain any abstract reasoning concerning quantity or number? No. Does it contain any experimental reasoning concerning matter of fact and existence? No. Commit it then to the flames, for it can contain nothing but sophistry and illusion.⁵

Unfortunately, Hume's principles threaten to throw out the baby with the bath water. When his principles are applied consistently, they seem to narrow the scope of genuine knowledge almost to a vanishing point. It turns out that on Hume's principles most of what we take to be common knowledge, as well as what we take to be scientific knowledge, will not really be knowledge at all. When Hume applied what he took to be Newton's methods in natural science to the "science of man," he was led, paradoxically, to the conclusion that induction, or scientific method, cannot be rationally justified. Hume was led to this conclusion when he noted that empirical generalization is founded on the principle of cause and effect and that the principle of cause and effect cannot itself be justified rationally. Before we can attend to this claim of Hume's we need to understand his analysis of the nature of the cause-effect relation.

We say that the virus causes the symptoms of smallpox. But what do we mean by this? Roughly, that the presence of the symptoms is invariably preceded by the presence of the virus. Certainly not that there is an a priori relation between the virus and the symptoms; the proposition that the cause of the symptoms is something else, though of course false, is perfectly intelligible and self-consistent. We learn through experience alone that the symptoms are caused by the virus. "And experience only teaches us how one event constantly follows another, with- out instructing us in the secret connection which binds them together and renders them inseparable."⁶ To see what Hume means by "without instructing us in the secret connection which binds [the cause and the effect] together," ask yourself how we discover that we can at will wiggle our thumbs but cannot wiggle our ears. The answer is that we learn only by making repeated tries which are followed by successes in the former case and by making repeated tries which are followed by failures in the latter case. After, say, a dozen tries at each, we know our will has an influence over our thumbs which it does not have over our ears. Yet there is nothing additional we observed in the twelfth try that we had not already observed in the very first try; we observe no subtle connection of "power" or "influence." Were we to observe the presence of such an influence, an otherwise normal man under paralysis ought to be able to know without trying to move a particular limb that he had lost his power to move that limb, for he should be able simply to observe the loss of the power directly. But the fact is that "a man suddenly struck with palsy in the leg or arm, or who had newly lost these members, frequently endeavors, at first, to move them and employ them in their usual offices."⁷ With respect to moving one's thumb, there is something in fact absent at the very first try but present at the twelfth try: an irresistible expectation produced by habit or conditioning, the presence of which gives the illusion that there is more to the cause-effect relation than the invariant sequence of two contiguous events. In summary, to claim that a certain virus is the cause of smallpox is to claim an invariant sequence of symptoms-preceded-by-virus.

Turning from the meaning of a causal attribution, Hume considers its justification. Past instances of invariant conjunction of virus and symptoms lead us to infer that virus and symptoms are related as cause and effect and thus that they will always be conjoined. But what, Hume asks, actually entitles us to make this inference? There are only two types of legitimate inference: a priori demonstration (deduction) and a posteriori experimental inference. The inference from observed past instances to unobserved future instances clearly is not demonstrable. And to say that this inference is experimental is to beg the question, "for all inferences from experience suppose, as their foundation, that the future will resemble the past."⁸ That is, because scientific method presupposes that the course of nature will not change, it can hardly be invoked to prove it. So Hume concludes that experimental inference, which is scientific method, is not rationally justifiable.

But Hume's empiricist principles also lead to skepticism beyond that about the legitimacy of experimental inference. "It is in vain," he said, "to ask whether there be body."⁹ He meant that you cannot know whether there are any physical things -- whether there are houses and trees and mountains, much less whether there are submicroscopic things like atoms and genes. Nor can you know whether there is any you who experiences your impressions and ideas, much less whether there are other people who experience their own impressions and ideas. All you can know is that there are and that there have been particular impressions and ideas. (Even this inventory of Hume's seems overblown, because his principles would seem to imply that you cannot even know whether there ever were any impressions or ideas in the past.) Indeed, if Hume's principles are correct, it is actually meaningless to ask whether there are people and physical objects external to your sensations. This is entailed by the empiricist criterion of meaning: because each and every idea you have is composed of simple ideas, and because every simple idea is a copy of a sensation or impression, the very suggestion of the existence of anything beyond your perceptions is unintelligible. Hume explicitly invokes his meaning-criterion to show that you cannot have any idea of a self who experiences your impressions and ideas. "From what impression could this idea be derived? This ques- tion it is impossible to answer without a manifest contradiction and absurdity; and yet it is a question which must necessarily be answered, if we would have the idea of self pass for clear and intelligible."¹⁰ Now we can see that Hume's principles lead ultimately to solipsism: all you can conceivably know is that there are sensations here and now and maybe that there were sensations in the past as well.

Surprisingly perhaps, Hume does not proscribe belief in induction by simple enumeration, belief in the existence of a self, or belief in the existence of an external world for -- on his view -- it is silly to proscribe beliefs that are held on instinct. Fortunately for us, skeptical reason is but a feeble critic when faced with instinct and custom. Were it the other way around, "All discourse, all action would immediately cease, and men would remain in a total lethargy until the necessities of nature, unsatisfied, would put an end to their miserable existence."¹¹

2. Phenomenalism

Recent empiricists have not agreed with Hume's conclusion that the belief in an external world is rationally unjustifiable. Until very recently -- up to around 1950 -- most of Hume's followers thought that his own basic, principles in fact implicitly contained the rational justification for this belief; this argument for belief in the external world is known as phenomenalism. According to this theory, anticipated by Hume and before him by George Berkeley, a physical thing is a kind of construction out of our experiences. Macbeth knew that the dagger he saw was hallucinatory and not real not because he knew that there was no external cause for his visual experience but rather because he knew that if he reached for the dagger he would see and feel his hand breeze right through the space where the dagger seemed to be. So, according to the phenomenalist, to have a visual experience of a real physical thing is to have an experience which belongs to a certain kind of group of experiences. This kind of set of experiences has a constancy and a rich and complex coherence that is lacking in the set of experiences of which hallucinations are a part. As John Stuart Mill, the main figure in nineteenth century empiricism, put it, matter is a permanent possibility of sensation.

"I believe that Calcutta exists, though I do not perceive it, and that it would still exist if every percipient inhabitant were suddenly to leave the place, or be struck dead. But when I analyse the belief, all I find in it is, that were these events to take place, the Permanent Possibility of Sensation which I call Calcutta would still remain; that if I were suddenly transported to the banks of the Hoogly, I should still have the sensations which, if now present, would lead me to affirm that Calcutta exists here and now. We may infer, theretore, that both philosophers and the world at large, when they think of matter, conceive it really as a Permanent Possibility of Sensation."¹²

0ne's instinctive belief that there is an external world of other selves is explained and justified in a similar fashion. The phenomenalist believes that Hume's principles show us that each one of us is nothing but a bundle of experiences having a certain constancy and coherence. This gives rise to the illusion of a simple abiding thing which is over and above the experiences and which "has" them.

It is important to note that the phenomenalist claims not merely that all we can possibly know about daggers, Calcutta, and ourselves is confined to facts about our experiences; he affirms also that all we can possibly mean in speaking of these things is confined to our experiences. To say that there is a jackknife in your drawer at home means that were you or someone else to go home and open the drawer, that person would have such and such experiences were he to do such and such things (were he to reach for it, he would have sensations of resistance, weight, smoothness, and coolness; were he to drop it, he would hear a thump; were he accidentally to nick himself with it, he would feel pain, and so on). That this is what we mean by a real jackknife (as opposed to a hallucinatory jackknife, a dummy jackknife, or a real banana) follows directly from Hume's empiricist criterion of meaning. Hume's criterion, it will be recalled, states that every idea -- and thus the idea of an "external" thing -- is ultimately derived from sense impressions alone.

By the twentieth century, when empiricists had come to take statements or propositions rather than terms or ideas as the basic unit of meaning, the phenomenalist thesis that a physical object is a construction out of sensations was interpreted as the thesis that any statement about a physical object is translatable into a statement about sensations -- that is, any statement about a physical object can be paraphrased into a statement about sensations. Thus, to make a categorical statement that there is along the banks of the Hoogly a particular drainage pipe is really to make a hypothetical statement about sense data. Such a hypothetical statement would be roughly that were a normal observer under normal observation conditions situated to view the banks of the Hoogly, and were he attentively to make such-and-such observations, he would then have such-and-such experiential data. (It is evident that a normality requirement is necessary, for obviously one thing not meant by "There is a new drainage pipe at the mouth of the Hoogly River" is: were a blind man to be suitably located he would have such-and-such visual sense data.) The phenomenalists believed that in these ways they could use Hume's own principles to analyze and justify the belief in the external world.

Although this general scheme was worked out to a remarkable degree of sophistication during the first part of this century (e.g., by Bertrand Russell and A. J. Ayer), the phenomenalist phase of post-Humean epiricism ended by the 1940s, for by that time it had become evident that statements about physical things could not be translated into propositions about actual and possible sense data. If a physical-object statement is to be translatable into a sense-data statement or into a set of sense-data statements, the former must at least be deducible from the latter. But it came to be realized that there is no finite set of statements about actual and possible sense data from which we can deduce even a single physical-object statement. Recall that the translating or paraphrasing statement must be couched in terms of normal observers in normal conditions of observation. There is, however, no finite set of statements that are couched in purely sensory terms and which can express the satisfaction of the condition of the presence of a normal observer. According to phenomenalism, to say in terms of actual and possible sense data that a normal observer is present is to make the hypothetical statement that were a doctor to inspect the observer, the observer would appear to the doctor to be normal. But, of course, not any old doctor will do but only one who himself is a normal observer. If we are to specify the doctor's normality in purely sensory terms, we must make reference to a second doctor who, when inspecting the sense organs of the first doctor, would himself have to have the sense data a normal observer has when inspecting the sense organs of a subject who is a normal observer. And if we are to specify in purely sensory terms that the second doctor is a normal observer, we must refer to yet a third doctor, and so on ad infinitum. In short, an objection fatal to phenomenalism is that its very formulation requires the use of a concept, that of a normal observer, which in principle cannot be defined phenomenalistically. Thus it is in principle impossible for anyone to provide a sense-data translation of even so banal a physical thing statement as "There is a new drainage pipe at the mouth of the Hoogly River."

3. Contemporary Empiricism

This brings us to the second and most recent major stage of post-Humean empiricism. Contemporary empiricism begins with the rejection of Hume's assumption that what we directly experience are always our own sensations. Physical things actually constitute the objects of perception, where "physical things" is taken in the wide sense that includes not only daggers and drainage pipes but also all the other sorts of objects common within the natural world -- shadows, mirror images, and thunder claps. Hume and the phenomenalists were not unaware of this common-sense belief, of course, but what convinced them it was not sound is called the argument from illusion. Hume and the phenomenalists reasoned as follows. We sometimes undergo illusions -- that is, we sometimes perceive physical things to have properties they do not really have. When approaching a table, what is in our visual field gets bigger, but of course the table does not grow. Thus what we see, which really is getting bigger, must be something else; but the only something else it could be is an experiential datum private to the perceiver -- a mental image representing the table. Thus we perceive not the natural world but only mental images of it. We also sometimes fall prey to hallucinations: what we see may look like a table (or a dagger) when actually there is no table present. So what we perceive cannot be a table, because by hypothesis no physical thing is present where we are looking. Thus the only alternative is that we are perceiving one of our own sense data. Moreover, because what we see when hallucinating a table is qualitatively indistinguishable from what we see when a table is actually present (or else we would not be fooled by the hallucination), what we directly see even when a table is present is not the table but rather our personal image of it. There are other variations of the argument from illusion, but the above cases bring out the essence of the argument. We may now turn to the contemporary empiricist objection to it.

As the contemporary empiricist sees it, neither the case of illusions nor the case of hallucinations requires us to introduce into the world special objects -- impressions or mental images of tables, etc. -- in addition to tables and the rest of the usual array of physical objects. In the illusion example above, we do not have to describe the state of affairs as one in which there is a queer something, a sense datum, which actually is growing larger. All we need say is what in fact in real life we actually do on such occasions say: "It looks as if the table is getting larger (although of course it isn't)." And in the case of hallucination, we can say: "I thought I saw a table, but actually I didn't see anything since there is nothing there to see." And in general, if we take concrete cases of illusions, hallucinations, dreams, afterimages, etc. one by one, we can show that in each case there is no need to introduce sense data to account for any aspect of our experience.

Indeed, in the view of the typical contemporary empiricist, Hume's doctrine that the only objects of our experience are perceptions or sense data is not merely unnecessary to account for the facts of experience but it results in an incoherency as well. What brings the contemporary empiricist thus to the complete and utter rejection of one of Hume's basic principles is one or another version of what is called the private language argument, an argument based on or derived from some remarks of Ludwig Wittgenstein. The core of this argument is as follows. If all your experiences are only of your own impressions and ideas, as Hume believes, and it all words derive their meaning ultimately from impressions, another basic doctrine of Hume's, then the language you have acquired cannot possibly be understood by any other person. In order for another person to understand what you mean by table or red and in order for you to understand what he means by these words in his language, you and he would have to be able to experience one another's perceptions. But this is impossible on Hume's principle that the only objects of a person's experiences are his own impressions and ideas. According to Hume, because you use red to stand for a sense impression private to you and another speaker of English uses this word to stand for a sense impression private to him, it is possible that what he calls red is what, were you able to have his sense impressions, you would call green. Thus Hume's principles lead to the strange conclusion that English, say, is not a single language understood by millions of people but rather a kind of federation of languages, each of whose users may use words in a way different from every other language user. But the critic of Hume points out that any such doctrine which entails that we speak in private language must necessarily be false, because private languages turn out to be a logical impossibility.

Why are private languages impossible? Any language involves rules of use. A noise or a mark which is not used in accordance with rules of correct application is not a word. Thus, for example, were a child or a monkey to make the noise red in no consistent manner with respect to the presence or absence of red objects or with respect to the presence or absence of any other observable phenomenon, the noise red would not be used as a word of a language. So the words of a purportedly private language would have to be used consistently if they really are words. But in a private language, the notion of the consistent or inconsistent application of, say, red is without meaning. For there is no possible way you could tell that you now use red to stand for the same kind of sense impression for which you have used it in the past. You cannot ask other people whether you are consistent, for that is ruled out: your language is supposed to be intelligible to you alone. Nor can you bring back and pass under review the sense impression to which yesterday you attached the sound red. You have only your own memory impressions to rely on, which means that you have no independent justification whatsoever for your belief that today you use red to apply to the same kind of sense impression to which in the past you affixed red. The point of the critic of Hume is not that memory impressions per se cannot be used to justify or support claims we make; the point is rather that those memory impressions, the credibility of which can be vouched for only by yet more memory impressions, can provide only the illusion of justification. If you wonder if you are right about when the train leaves, you might call up a memory picture of the timetable in support. But the memory picture can thus serve as support of your recollection of the train's departure time only because the memory picture itself can be tested independently of your own personal memory impressions: you can phone the train station, ask a friend, or get the timetable and look at it. In contrast, the Humean private-language user, who has for support of any memory impression only other memory impressions, is like a man who gets extra copies of the New York Times to prove that what was said in the first copy is true.

To sum up, the structure of the argument against private language is that of reductio ad absurdum: assume as true for the sake of argument what you actually take to be false -- in this case, that there is or can be a private language -- and then show that this assumption leads to its denial, namely, that there cannot be such a language. This argument against private language was influential in bringing about the transition from Hume and phenomenalism to contemporary empiricism.

So the transition to contemporary emipiricism is made by replacing Hume's doctrine that private sense impressions are the objects of direct perception with the view that publicly observable physical things are the objects of perception. Observation claims about these, rather than about sense data, provide the foundations of empirical knowledge. That is, contemporary empiricism is Humean empiricism without sense data. Like Hume, the contemporary empiricist holds that a priori knowledge is analytic; he holds that observation -- now of publicly observable things -- together with memory is the only source of empirical or a posteriori knowledge; he holds that any reasoning taking us beyond this source, i.e., any nondemonstrative (nondeductive) reasoning is basically empirical generalization from observations; and he holds that all meaningful ideas must ultimately come from experience.

We can see how Hume's principles, cleansed of the doctrine of sense data, enable the contemporary empiricist to deal with the problem of the external world: once one sees that sense data are a myth born from conceptual confusion, the problem of the external world seems to vanish. There is then no question of whether we are rationally justified in believing that beyond the iron curtain of our sense data there is a world of tables, houses, mountains, and so on, because in every normal waking hour of our lives these items are precisely what we directly confront in perceiving. In this way contemporary empiricism justifies our instinctive belief in the existence of ordinary objects, the existence of which in no way depends upon our perceptions of them. But the nonphilosopher believes that there is more to the world beyond his own mind and conscious experiences than just the ordinary things like houses and mountains; he believes that there are other selves, with their own conscious experiences, and he believes that there are unobservable physical things which are not part of our day-to-day world, namely, the entities of theoretical science such as atoms, electrons, and genes. How does contemporary empiricism deal with this aspect of the problem of the external world?

The solution to the problem of justifying one's belief in the existence of other selves is implicit in contemporary empiricism's justification of our belief in the existence of everyday physical things. Self is simply a misleading label for persons, for flesh and blood creatures we confront every day in perception and in interaction. Like the account of our knowledge of ordinary physical things, this account of our knowledge of our self and of other selves accords with common sense: persons are a species of physical thing of which oneself is a member, and the others of which anyone but a hermit directly confronts every day.

However, a problem arises for the contemporary empiricist when he tries to explain one's knowledge of another person's experiences. Because the contemporary empiricist cannot observe or experience another person's experiences (he cannot feel his pains, for example), how on strict empiricist principles can he justify his day-to-day beliefs that, e.g., right now Jones is in pain? In order to see the contemporary empiricist's answer we must turn to his "linguistic" version of Hume's empiricist criterion of meaning: every genuine descriptive word must be either an ostensive word such as red or it must be a term definable in terms of ostensive words. An ostensive word cannot be defined verbally but only by pointing out examples of what it is to which the word applies. Thus one understands a descriptive word only if one knows the observable situations to which it can correctly apply.¹³ Now Hume's criterion of meaning -- when combined with the contemporary empiricist doctrine that the objects of perception are public and intersubjective -- yields the requirement that if a word such as pain is to be counted as the genuinely descriptive word it obviously is, then it must in some way refer to publicly observable circumstances. One teaches a child what pain means by pointing out cases of crying with pain, grimacing with pain, etc.; and in the view of contemporary empiricism, this is no mere happenstance. If people did not behave in certain characteristic ways when having a pain, the very concept of pain could not become a part of our conceptual scheme. Thus we cannot separate pain-behavior from the very meaning of pain. Some contemporary empiricists believe that this entails that pain is nothing but a certain pattern of behavior. These are the operationalistic (or behavioristic) psychologists, linguists, and social scientists. However, most contemporary empiricists do not hold that pain means merely pain-behavior. In their view, the behavior is only an aspect of the meaning of pain -- that aspect which guarantees its status as a genuinely descriptive word. Because the state of being in pain has both an inner and an outer aspect, we can rationally account for knowledge of a person's pains and other states of consciousness whether that person is oneself or someone other than oneself. The pain Jones feels, so to say, from the inside is one and the same thing as what those of us in Jones' vicinity can observe from the outside. So it is roughly in this way that contemporary empiricism resolves the "other minds" aspect of the empiricist problem of the external world -- the problem of accounting on strictly empiricist principles for all levels of our knowledge of the existence of anything outside of one's own states of consciousness.

With respect to justifying or analyzing the nonphilosopher's belief in the individually unobservable submicroscopic entities of theoretical science -- atoms, electrons, genes, and such--- strict empiricist standards oblige contemporary empiricists to conclude that theoretical talk is a kind of a convenient shorthand fiction for talk about the behavior of such genuinely real and observable things as the movement of a meter-pointer or the path of a streak in a Wilson cloud chamber. We cannot see or touch an individual electron for precisely the same reason that we cannot see or shake hands with the average man.

4. The Critique of Contemporary Empiricism

I noted at the beginning of this essay that criticisms of the principles of contemporary empiricism focus mainly on the contemporary empiricist conception of the role observation plays in obtaining factual or empirical knowledge. I have been describing what this conception is in the context of the Humean and phenomenalistic empiricism, which leads up to it. Although empiricism has never lacked critics, the present-day questioning of its foundations started to gain momentum only around 1950. These current criticisms issue in general from science-oriented philosophers, from historians of science, and from scientists who have argued that the principles of empiricism in fact conflict with actual scientific methodology and with certain substantive results in linguistics and in the psychology of perception. This group, which includes all the contributors to this volume, challenges and/or revises empiricism in light of the methodology and the results of scientific practice.

As we have seen, empiricist epistemology from Hume on is guided by the idea that observation provides a maximally certain and conceptually unrevisable foundation of empirical knowledge, a foundation that supplies the basic premises of all our reasoning and without which there would not even be any probable knowledge. This foundation picture of factual or a posteriori knowledge has come under severe attack on the grounds that it is not merely simplistic but that it is incorrect and consequently misleading. In the view of the critics, if one looks at actual scientific inquiry one sees that sense experience does not function as the foundation of knowledge at all. Instead, it serves as a stimulus, provoking us to make revisions in our system of beliefs -- revisions from which our observation claims are not themselves immune.

Critics have raised three not entirely independent objections to this foundations picture. First, observation claims derive their credibility from background assumptions. It is simply false, the critics say, that what a scientific theory does is to explain established "facts" revealed by sense experience. The question of what the facts are is not one which can be answered except within a theory or network of background assumptions. Thus, for example, when we establish on the basis of eyewitnesses that it is a fact that Jones actually did go through a red light or that his dipped-in-acid litmus paper actually turned red, we presuppose that there actually exists at least one normal observer. This assumption, one which we all make, cannot be justified by either of the two empiricist-certified types of reasoning. Because this assumption is about what Hume calls "matters of fact and existence" and is thus not a truth about relations of ideas, it cannot be demonstrated. Nor can we support it by enumerative induction, for such extrapolation from observation claims presupposes the reliability of those claims and thus can hardly be invoked to support them. Thus there is no sharp distinction between observation claims, which were purported to report only what is directly observed, and theoretical claims, which assert more than what is directly experienced; every claim in one way or another transcends what is presently and directly experienced.

Second, according to the contemporary critics of empiricism, not only do observation claims derive their credibility from a network of background assumptions but they also derive their very meaning from this network. Thus observation terms, it is claimed, do not retain constant meanings through theory changes. For example, the scientific revolution of the sixteenth and seventeenth centuries replaced the Aristotelian conception of color as a continuous and perceiver-independent property of physical things with the conception of color as a certain power of physical things to stimulate an organism's sense organs. The critics hold that "being red" no longer means quite the same thing as it did to Aristotle, and thus that an observation claim like "The (litmus) paper turned red" has undergone a meaning-change.

Because this second objection applies to observation language, it applies to observation words as well as to observation claims. It will be recalled that the contemporary empiricists translated Hume's assumption that there are simple ideas into the claim that there are ostensive words -- words which refer to and mean a simple experiential content -and which imply nothing beyond that experiential content. But if there is no sharp distinction between observation claims and theoretical claims, then ostensive words are a myth. For when one points to a red thing and tells a child, "That is what red means," one has merely pointed out one aspect of the meaning of red. To be red is to be able to appear in a certain multiplicity of ways -- under a multiplicity of conditions for observation for a multiplicity of kinds of observers (normal ones, color-blind ones, etc.) -- and not just in a single way. Thus it is claimed that observation has no monopoly on meaning; instead, it is held that our theoretical claims and concepts give meaning to our observation claims and to experiential concepts such as color concepts.

Third, the critics of empiricism argue that not only are so- called observation claims and concepts theory-laden, but also our very observations are theory-infected. The principles of empiricism are proffered against the background presupposition that a person's perceptions are unaffected by the beliefs he has, by the assumptions he makes about the objects that he is observing. That is, empiricism rests on the background assumption that there is an absolutely stable and invariant correspondence between perceptions and the stimuli which produce them. The critics of contemporary empiricism believe that the results of empirical psychology disconfirm this empiricist pre-supposition. Observations are not "givens" or "data" but are always interpretations in the light of our background assumptions. The idea of unambiguous objects of perception is a myth. Take the familiar examples from gestalt psychology of the duck-rabbit figure and the figure which can be seen either as a black cross on a white background or as a white cross on a black background.¹⁴ What someone sees in looking at such figures is not independent of his beliefs and expectations. For example, a person who is unacquainted with the ambiguity of such figures and who expects to see a rabbit-picture sees just that.

The result of these criticisms is that if they are correct, one must reject or severely alter the "foundations" picture of empirical knowledge, which lies at the heart of the empiricist vision. The empiricist picture is that all empirical knowledge is founded upon a set of independently intelligible and independently credible "observation claims." The critique suggested here, it sound, shows that it is a myth that there are such "basic" claims, and that the doctrine that all empirical knowledge could rest on such claims is incoherent.

5. The Critics and Their Criticisms

Any categorizing of the criticisms and revisions of contemporary empiricism can claim only a certain convenience, for these criticisms and revisions are largely interconnected. In particular, the issues of the first section of this collection spill over into those of the second section. In the first section the essays all deal with ontology, or the theory of reality, and its relationship to empiricist epistemology. There is also much in these essays about the relationship between scientific inquiry and empiricist epistemology.

In "Empiricism, Semantics, and Ontology" Carnap distinguishes questions which are "internal" and those which are "external" to a conceptual scheme or language framework. He argues that while we can determine what things there are in the world by observation and by logical or mathematical demonstration within a given conceptual scheme or language framework, we can determine whether or not the conceptual scheme itself "exists," i.e., whether or not it provides a true map of reality, only on pragmatic or practical grounds. For example, given our actual conceptual scheme, "Are there brick houses on Elm Street?" is a factual question which is answerable by looking and seeing; whereas "Are there physical objects per se?" is not a factual question, for given our choice of a convenient language framework, its affirmative answer is merely analytically true, true "by definition." Carnap's position contrasts markedly with the empiricist view that it is an obvious factual truth about the world, revealed through the testimony of the senses, that there are physical things. Thus, whereas for the traditional empiricist only what is directly observable is real (everything else, it will be recalled, is a construction out of these foundational elements), for Carnap the empiricist's question as to which kinds of entities do and do not occur as immediate data is entirely irrelevant for ontology. So when seeking the best conceptual scheme, Carnap suggests we should ask not whether sense data, physical objects, numbers, etc. are observables but rather whether each of these categories serves as an efficient instrument of inquiry.

In "Two Dogmas of Empiricism" Quine argues that in "Empiricism, Semantics, and Ontology" and elsewhere Carnap fails to carry his pragmatism to its logical conclusion. The reason for this, Quine thinks, is that Carnap is unwilling to give up the empiricist synthetic-analytic dogma that some statements are about the world, whereas others are merely about language. Because the distinction between statements about the world and statements about language is illusory, says Quine, ontology is in fact a branch of science (although more abstract than most other branches, to be sure). Like any branch of science, its ultimate criterion of adequacy should be a pragmatic one. In Quine's view the analytic-synthetic dogma is ultimately one with the dogma of reductionism -- the dogma that every descriptive statement is individually a report or summary of direct experience. He thinks that it is a mistake to think that scientists compare single individual statements with the world. We confront reality only with a totality or network of statements and beliefs, so that statements ought to be viewed as ingredients of a total system rather than as isolated reports or summaries of experience. And the ultimate purpose of the total enterprise of science is a pragmatic one: "Each man is given a scientific heritage plus a continuing barrage of sensory stimulation; and the considerations which guide him in warping his scientific heritage to fit his continuing sensory promptings are, where rational, pragmatic."

In the third selection Quine argues that there is no way we can tell for sure whether or not members of an alien culture divide the world into the kinds of objects that we do, or even whether or not they "cut up" the world into objects at all. Suppose they say "gavagai" whenever a rabbit passes by. Does this response to the rabbit-stimulation indicate that gavagai means rabbit for them? Apart from guesses about many other words of their language as well -- what Quine calls "analytic hypotheses" -- there can be no definite answer to this question. Within the context of one such system of analytic hypotheses gavagai may come out as "rabbit," but within other systems of hypotheses it may come out something like "rabbit-event," "rabbit-stage," "undetached rabbit-part," or something even stranger from the point of view of our ontology. Thus from the point of view of the question of the preconditions for translation, Quine challenges the empiricist doctrine that experience alone can reveal to us what kinds of things there are in the world. Quine also attacks empiricist ontology in a subtler way. The analytic-synthetic distinction holds that some propositions are true in virtue of their meaning. Bolstering the idea that there are meanings is the idea that translation consists of finding a sentence in, for instance, English that has the same meaning as a sentence in the alien language. If Quine is right, such a conception of the translator's project makes no sense, because it is a senseless idea that one could translate gavagai into English apart from any system of hypotheses at all.

Sellars, in "Empiricism and the Philosophy of Mind," argues that the scientific account of the world ought to take precedence over the common sense ontology of everyday life; for example, because colors have no place in the scientific framework, we ought to say that strictly speaking, i.e., scientifically speaking, they do not exist. He argues that those who find it preposterous to question the validity of our common-sense categories are victims of the empiricist myth that language hooks on to the world basically by means of ostensive definitions.

In "Brains and Behaviour" Putnam attacks the most plausible form of behaviorism in contemporary empiricism. This "soft" behaviorism, it will be recalled, holds that part of the meaning of any psychological term is its "outer" aspect -- namely, behavior. Thus recall the example of pain: the "soft" behaviorist argues that we can teach children the meaning of pain only because people behave in certain characteristic ways when they feel pain; and from this the "soft" behaviorist infers that pain-behavior becomes part of the very meaning of pain. By various examples and arguments Putnam tries to show that a consideration of how scientific theories actually are constructed and supported leaves "soft" behaviorism as an enormously implausible theory. If we wish to theorize in accordance with the practices of the sciences, the plausible thing to conclude is that pain-behavior is in no way part of what we mean by pain; rather, it is merely the effect of a publicly unobservable event -- viz., a feeling of pain. More broadly, Putnam is here arguing against the contemporary empiricist doctrine that it is absurd or meaningless to say that there are publicly unobservable entities. In Putnam's view, there are cases in which the most plausible way to scientifically explain a "fact" is to posit the existence of entities which may be publicly unobservable.

In the second section of this collection the attention of the authors is focused directly upon the relationship between science and empiricist epistemology. According to Popper's "Science: Conjectures and Refutations," the traditional empiricist problem of justifying inductive inference is a pseudoproblem emanating from Hume's misconception of the nature of scientific method. Hume saw correctly that it is impossible to infer a theory from observation claims, but he failed to see that from this one should not conclude that a theory cannot be refuted by observations. Once we see that scientific method is essentially the method of trial and error, not essentially empirical generalization as the empiricist would have it, we see also that science is a perfectly rational activity. Rather than generalizing from observations, we jump to conclusions, make bold conjectures; we venture theories which we test not by trying to infer them from some imaginary foundation but by setting up experimental situations in which we try to refute or falsify them. Thus by the method of trial and error, or more aptly, conjectures and refutations, we use experience to weed out false theories.

In his compressed note, "Science without Experience," Feyerabend uses some discoveries and inventions of recent science and technology (computers, subliminal perception) to suggest the perhaps heretical possibility that science without experience is conceivable. This is a way, he suggests, to show the falsity of the empiricist doctrine that experience furnishes the foundations of knowledge. This note of Feyerabend's provides a good orientation for the perspective of his earlier and much fuller "How to Be a Good Empiricist." Feyerabend's recipe in this earlier essay is that a good empiricist is one who tries to proliferate theories rather than one who contents himself with merely sharpening the fashionable theories of the day. Because the good empiricist obviously cannot pull out of a hat a full-blown alternative to the going theories, he must start with the development of an alternative vision to the picture embedded in the going theory; he will try to formulate a general view not yet directly connected with observations. In a word, to be a good empiricist one must be a critical metaphysician. Traditional empiricism, in its attempt to rid science and philosophy of metaphysics, ignores the actual (and fruitful) way in which scientists carry on their enterprise in real life: "Elimination of metaphysics, far from increasing the empirical content of the remaining theories, is liable to turn those theories into dogmas." A one-theory system in science, says Feyerabend, would be as unproductive as a one-party system in the state.

In "incommensurability and Paradigms" Kuhn defends the antiempiricist view that scientific frameworks are incommensurable with one another in roughly the same way that natural languages are incommensurable. Quine had argued that in translating, say, from Hopi to English, we cannot avoid gerrymandering Hopi to fit a mapping into English. Similarly, Kuhn argues, what Newton said must be "warped" to fit our Einsteinian space-time framework. Because there is no such thing as a theoretically neutral description of anything, "the choice of a new theory is a decision to adopt a different native language and to deploy it in a corresponding different world." Unlike English and Hopi, perhaps, it might appear that there are words in common -- words at least spelled the same -- between the languages of old and new scientific frameworks. Besides sharing certain scientific words like mass, Newton and the later Einstein used English and thus seem to share the observational vocabulary of English. But according to Kuhn and Feyerabend, terms like mass or red typically take on an altered meaning within a new theory, for words are implicitly defined by the theories which employ them.

In "Duhem, Quine and a New Empiricism" Mary Hesse argues that this controversial thesis of meaning-variance between theories is a welcome antidote to the traditional empiricism in which statements, terms, and predicates are taken to have atomically independent and therefore stable meanings. She believes that by noting some of Quine's suggestions (which in key respects are refinements of Duhem's) holders of the meaning-variance thesis can avoid the paradoxes which would seem to make the thesis preposterous. Feyerabend and Kuhn are right to reject the traditional empiricist view that the meanings of terms are unaffected by the theories in which they are used: the meaning of a term in one theory is not the same as its meaning in another conflicting theory. Quine's doctrines support this view, she thinks, and they enable its defenders to avoid what would seem to be some absurd implications. Some of these implications are that no observation claim containing the term in one theory can contradict a statement containing the term in another theory and that there are no crucial experiments between theories. Hesse also questions to what extent Quine can reasonably be called an empiricist.

The third section of this book is devoted entirely to the Chomskian attack on the empiricist view of how language and knowledge are acquired. In the opening symposium Chomsky defends his claim that the empiricist account is inadequate; Putnam and Goodman attack this claim. Chomsky argues that the rules relating the underlying "deep" or logical structure to the surface structure of a sentence are so complicated and abstract as to challenge the adequacy of any empiricist stimulus-response or associationist account of how they are learned. We must suppose that the child has innate knowledge of the structure of his language -- that is, that it has been programmed into his brain at birth. Only in this way can we account for the intellectual feat of his acquiring the rules of the grammar of a natural language.

In the critiques of Chomsky in this section, Putnam and Goodman hold that this innateness hypothesis of Chomsky's is probably unintelligible at bottom and is anyhow needless; an empiricist account of knowledge and language acquisition is perfectly capable of explaining the facts. In his discussion note, "Linguistics and Philosophy," Quine adds to this that Chomsky's criticism of empiricism is a red herring. He says that as a matter of fact empiricism cheerfully embraces innate mechanisms: "Externalized empiricism or behaviorism sees nothing uncongenial in the appeal to innate dispositions to overt behavior, innate readiness for language-learning." Thus when we blow away all the froth, Chomsky says nothing with which a behaviorist or externalized empiricist can disagree. "What would be interesting and valuable to find out, rather, is just what these endowments are in fact like in detail." And with respect to that question, Quine implies, Chomsky has nothing helpful to tell us. Finally, Edgley, in his essay "Innate Ideas," suggests that Chomsky illegitimately infers from certain empirical facts of language learning that we must have innate knowledge that certain things are true when all Chomsky could hope to infer is that we must have innate knowledge of how to apply certain abstract principles of grammar.

The last two selections in this collection show how Chomskians reply to these sorts of objections. If Fodor is right, then Quine is mistaken in suggesting that behaviorism can account for the empirical facts about language. "Behavioristic accounts of perception as involving the disposition to produce discriminative responses to physical differences are plausible only when isomorphism obtains between perceptual distinctions and specifiable stimulus differences." But when we hear, for example, a speaker say "The man hit the colorful ball," we perceive the longest pause on either side of hit, "although in acoustical fact, normal utterances of this sentence contain no energy drop at all." What we are in fact unconsciously doing is parsing this sentence in accordance with its conceptual or syntactic structure. This parsing cannot be explicated in terms of a "disposition" to provide a verbal report because normally we cannot say what we are doing; nor do our parsings correspond with any kind of characteristic nonverbal patterns of behavior.

"Perhaps the clearest and most explicit development of what appears to be a narrowly Humean theory of language acquisition in recent philosophy is that of Quine." With these words, Chomsky, in the concluding essay, launches a sustained and Complex attack on the adequacy of Quine's attempt to use behavioristic stimulus-response psychology to "externalize" Hume's theory of language acquisition. Chomsky holds that Quine's definition of language as a "complex of dispositions to verbal behavior" is no more than an empty verbalism. In addition to his criticism of Quine, Chomsky ferrets out and criticizes Wittgenstein's empirical assumptions about language.

Notes

David Hume, An Enquiry Concerning Human Understanding, in Hume, Enquiries Concerning the Human Understanding and Concerning the Principles of Morals, 2d ed., L. A. Selby-Bigge, ed. (Oxford: Clarendon Press, 1902), p. 19. The quotations from Hume's writings in this introductory essay follow that of the editions cited except that certain archaisms of spelling, punctuation, and the use of capital letters have been corrected.
Ibid., p. 22.
Ibid., pp. 25-26.
lbid., p. 25.
lbid., p. 165.
lbid., p. 66.
Ibid.
Ibid., p. 37.
Hume, A Treatise of Human Nature, ed. L. A. Selby-Bigge (Oxford: Clarendon Press, 1888), p. 187.
Ibid., p. 251
Hume, Enquiries, p. 160.
John Stuart Mill, An Examination of Sir William Harniiton's Philosophy, in British Empirical Philosophers, ed. A. J. Ayer and Raymond Winch (New York: Simon and Schuster, 1968), p. 552.
Some contemporary empiricists expand this meaning-criterion to statements: one understands a descriptive statement only if one knows the observable situations which would make it true. This move has not gained wide allegiance from contemporary empiricists, however.
The sketch given in the text is but a rough one of the empiricist doctrine of ostensive definition. Wittgenstein and others have made this position sophisticated. For example, Wittgenstein has argued persuasively that ostensive definitions can be used only with a pupil who already has a background of linguistic abilities.
For those already acquainted with the linguistic philosophy of Wittgenstein, I should mention that Wittgenstein's conception of criterion is nothing but the conception of ostensive definition qualified as noted above: where the falling of the barometer is a "symptom" of rain, its looking like that outside is the "criterion" of rain, the phenomenon to which we must point in order to explain the meaning of rain (see, e.g., Wittgenstein, Philosophical investigations (Oxford: Blackwell, 1953), s. 354).
Figure from Philosophical Investigations by Ludwig Wittgenstein, 1953. Reprinted by permission of Basil Blackwell, Oxford.
Bruce Aune has pointed this out in Knowledge, Mind and Nature (New York: Random House, 1967), pp. 266-67. This book of Aune's as well as his Rationalism, Empiricism and Pragmatism: An introduction (New York: Random House, 1970) provides an excellent discussion of many of the issues treated in this collection. (Note: footnote marker seems to be missing from the original text.)

Challenges to Empiricism Introduction Harold Morick, State University of New York, Albany 1972, Wadsworth Publishing Company, Inc., Belmont, California