SOURCE: "Descartes' Empirical Epistemology," in Descartes: Philosophy, Mathematics and Physics, edited by Stephen Gaukroger, The Harvester Press, Sussex, 1980, pp. 6–22.

[In the essay that follows, Larmore contends that Descartes' epistemology uses experimentation within a framework of a priori principles to advance human knowledge.]

There is something close to a general consensus that Descartes initiated a search for incorrigible foundations of knowledge that deeply shaped modern philosophy and that we have now learned to reject or even ignore. Characteristic of the Cartesian search for certainty, as opposed for example to some tendencies in Greek thought, was that these foundations must be located in individual subjectivity, in our immediate awareness of our own mental states. It implied that unless we could show how our beliefs about the world could be legitimately inferred from this basis, they would have no more rightful claim to being knowledge than would our wildest fantasies.

All the different kinds of errors that lie at the heart of the foundationalist enterprise do not need rehearsing once again. More directly of interest is the fact that a number of philosophers have taken the demise of this enterprise to mean the end of epistemology itself. What else can epistemology be but the search for the incorrigible foundations of knowledge? If that is so, then epistemology indeed amounts only to a subject with a glorious past. But this is not the proper conclusion to draw. The rejected forms of epistemology proved barren because they restricted themselves to the search for incorrigible truths, untainted by the revisability of the empirical truths they were meant to support. To discard epistemology as a dead subject no longer of interest to living philosophy, for this reason alone, merely continues the original error of believing that the theory of knowledge must be kept pure of all dependence upon the empirical sciences.

There are two areas of inquiry whose pursuit would dissociate epistemology from the ideal of a prima philosophia. First of all, we can focus the theory of knowledge upon examples of scientific knowledge in order to formulate criteria of scientific rationality. However, if we are to escape the ideal of a pure epistemology, we must draw out these criteria in a dialectical way from the history of science. Crudely put, we must abstract the criteria from some theories in order both to evaluate other theories in terms of them and to test the criteria against other examples of theories. Otherwise we may find ourselves, as indeed has often been the case in the philosophy of science, stuck with criteria of scientific rationality that no scientific theory has ever met. Obviously, the problems facing this kind of empirical epistemology are immensely difficult. Secondly, we can allow the theory of knowledge to confront what scientific theories imply about the status of our perceptual and experiential image of nature and about the relation between nature and ourselves as knowers. Here the concerns of an empirical epistemology would not be as in the first case methodological, but instead substantive. They would focus upon how we are to understand human knowledge given what we know about the world.

If with these possibilities of an empirical epistemology in mind we look once again at Descartes' theory of knowledge, the traditional picture of Descartes as the founder of a priori epistemology begins to appear importantly incomplete. As I shall show in this essay, his search for an incorrigible foundation of empirical knowledge forms but one strand in his theory of knowledge. There are other epistemological problems for whose solution he deliberately resorted to the results of empirical inquiry. First, it might be recalled, in regard to the project of setting out criteria of scientific rationality, that he recommended his idea of scientific method because he had found it successful. However, I shall be concerned with the more substantive area of his empirical epistemology, especially as it grows out of his attempt at the mathematization of nature. In the light of these generally ignored aspects of Cartesian philosophy, we will no longer be able to foist upon Descartes the onus of having encouraged the idea that an a priori approach is all to which epistemology may aspire. Indeed, for the whole of the seventeenth century the theory of knowledge brought together both a priori and empirical perspectives. In Descartes, the relation between these strands is governed by a conception of method, whereas in Locke, for example, the character of their relation is far less clear. The origin of the idea that epistemology, as a philosophical discipline, must proceed independently of the sciences belongs to a later time. It arises both with Kantian transcendentalism and with the more recent wish to analyse 'the meaning of the concept of knowledge'. One aim of this essay is to indicate why we need a more complex picture of the origins of modern epistemology in the seventeenth century. But, more directly, the aspects of Descartes' empirical epistemology which I shall treat will be among those that can still interest us today.

It is chiefly in his physiological treatises, such as the Treatise on Man and the Dioptrics, that we come upon his empirical epistemology. But in order to understand why at a certain point Descartes let his epistemology become empirical, we will first have to look at his conception of scientific method (Part I). In Part II I shall examine the initial physical problem—the mathematization of nature—with which his empirical epistemology begins, then tracing in Part III the broad implications he drew from that for an understanding of the place of knowledge within the natural order.

Part I Descartes' Conception of Scientific Method

Recently it has become increasingly clear just how erroneous was the traditional view that Descartes thought of physical inquiry as a strictly a priori concern. We can find no better proof of the untenability of that view than to listen to what Descartes himself had to say in the Discourse on Method about the respective roles of the a priori and experience:

I have first tried to discover generally the principles or first causes of everything that is or that can be in the world, without considering anything that might accomplish this end but God Himself…. But I must also confess that the power of nature is so ample and vast, and these principles are so simple and general, that I observed hardly any particular effect as to which I could not at once recognize that it might be deduced from the principles in many different ways; and my greatest difficulty is usually to discover in which of these ways the effect does depend on them. As to that, I do not know any other plan but again to try to find experiments of such a nature that their result is not the same if it has to be explained by one of the methods, as it would be if explained by the other.¹

Thus, according to Descartes, an account of the physical make-up of the world falls into two distinct parts: one we can develop a priori, while the other makes essential use of experience. The 'principles or first causes', that cover the most general features of the world, are something that we can attain without appeal to experience or experiment. In this passage Descartes was referring to what he believed he had already accomplished in his earlier treatise Le Monde. There, from God's immutability alone, he had derived the three fundamental laws of nature:

1 Every bit of matter continues in the same state until constrained to change by encountering some other object.

2 When one body alters the state of another, it cannot give it any movement which it itself does not lose at the same time.

3 Every body tends to continue to move in a straight line.²

The same claim, that the validity of these laws has an a priori basis in an understanding of what it means for there to be a God, reappears in the Discourse and in the Principles as well.³ These laws of nature can be said to be true a priori, of course, only because Descartes thought that he could prove the existence of such a God in a purely a priori fashion, and not by means of some natural theology. Both the causal and the ontological proof take as a premise that I do have a concept of something than which nothing greater can be conceived. That I do have the ideas that I believe I do, whatever may be their material truth, is a result guaranteed by the indubitability of the cogito. Thus, contrary to what has been sometimes suggested, the cogito does play an essential role in the foundation of physical science. It lies at the basis, Descartes believed, of the a priori deduction of the three fundamental laws of nature.

It is important to notice how this a priori part of Cartesian physics lies on a continuum with a priori epistemology. For Descartes, a priori epistemology does not issue simply in a prescription for the kinds of propositions that should serve as foundations (that, of course, is the role that more recent phenomenalist epistemologies have taken on). Instead, the cogito and the proofs of God's existence imply, so he believed, the fundamental principles of physical science themselves. This continuity between a priori epistemology and physics should be borne in mind when we come to consider the continuity between physical theory and Descartes' empirical epistemology. It will become clear that it is his conception of scientific method that orders the a priori and empirical parts of the theory of knowledge and the theory of nature into a single enterprise. Now this a priori physics cannot, as we have seen Descartes admit, yield a complete picture of the physical world. Only the most general features of the world can be ascertained through deduction from the self-evident first principles. For example, from the three fundamental laws of nature he thought he could deduce the laws of impact among bodies. In Principles III, art 46 there occurs a passage where Descartes lists some of the more particular phenomena that we can uncover only through empirical inquiry: the size of the parts into which matter is divided, the speed with which they move, and what circles their movements describe. Clearly, this range of empirical phenomena consists in the numerical values that in any particular case can be given to the variables occuring in the a priori laws of motion and their deductive consequences. In the passage cited from the Discourse at the beginning of this section, he mentions another area of necessarily empirical inquiry. From the a priori laws alone we cannot determine what, in fact, is the mechanical constitution of many of the phenomena we observe. This is the domain of empirical inquiry that will be important in what follows. His empirical epistemology will depend upon understanding the operation of the human eye, for which he will appeal to empirical physiology as well as to a theory of the mechanical nature of light which he found himself forced to justify empirically.

Descartes thus believed that scientific inquiry must begin with an a priori demonstration of first principles and then, once the scope of a priori physics has been exhausted, it must turn to the construction of empirical hypotheses. Since earlier works like the Regulae often suggest a thoroughly aprioristic method, it is with his mature conception of scientific method that I shall henceforth be concerned.⁴

Those explanatory propositions belonging to the empirical part of physical inquiry Descartes himself termed 'hypotheses'. He said that if the consequences of an hypothesis agree with experience and, more particularly, if by way of a crucial experiment they agree with an experimental phenomenon that the deductive consequences of rival hypotheses fail to match, then we have every reason to believe that the hypothesis is true.⁵ (It is to be remembered that the Cartesian idea of deduction is broader than the logical concept of deduction—it covers any sequence of propositions where we perceive 'clearly and distinctly' that the conclusion follows from the premises.)

The hypothetico-deductive method, for Descartes, belongs only to the empirical part of physical theory; it does not touch the fundamental laws of nature and their deductive consequences. There is, of course, the famous passage at the close of the Principles (IV, art 204) where Descartes refers to the whole of his physical theory as an hypothesis whose truth can be guaranteed only by the match between its deductive consequences and experience. This and similar passages have sometimes encouraged the view either that toward the end of his life Descartes had begun to doubt his ability to demonstrate any a priori physical truths or that, in fact, he had never had that ambition.⁶ But this interpretation of the passage is seriously mistaken. At Principles IV, art 205, he says that the hypothetico-deductive method can give us only a 'moral certainty' in the truth of an hypothesis; by this he means that when an hypothesis coheres with the phenomena we have no reason to doubt its truth, though of course it could still possibly be false. But in the subsequent section (IV, art 206) he goes on to claim that about a number of propositions we have more than moral certainty, we have in fact 'metaphysical certainty', once we understand that God exists. These propositions are ones that we can deduce from God's existence and include, not only that we can indeed distinguish the true from the false, but also mathematical truths and physical truths that are equally self-evident. These physical truths are, he says, 'the principal and more general ones'—in other words, the three fundamental laws of nature. Thus, Descartes' position at the end of the Principles does not differ from what he said in the Discourse. His point in the final passages of the Principles where he describes the whole of his physical theory as an hypothesis is simply that, if we were not able to give an a priori demonstration of certain basic physical truths, they too would then have to assume the status of confirmable but ultimately corrigible hypotheses.⁷

Descartes' thesis that propositions lacking an a priori demonstration must be treated as hypotheses and tested by means of crucial experiments had an important methodological consequence. If we believe that principles explaining some physical phenomenon can be deduced from other self-evident principles but we do not see yet how the demonstration can be set up, we are not forced to let that part of physical theory lie fallow. Instead, we can admit those principles to the corpus of scientific knowledge if their experimental consequences are borne out. Later, of course, we could return to give them the a priori demonstration they deserve. This is, in fact, precisely what Descartes did in the Dioptrics and Meteorology. Instead of being demonstrated a priori, the mechanical nature of light has in these treatises the status of an hypothesis, from which he sought to deduce both the laws of refraction and, along with physiological data, the operation of the human eye.⁸ In the Discourse on Method, he maintained that in these treatises he has merely withheld the a priori demonstration of this hypothesis that he already possesses. But in a more candid letter to Mersenne of 17 May 1638 (shortly after the publication of the Discourse and the Dioptrics) he confessed that an a priori demonstration of the mechanical nature of light is still only a confident hope.⁹

I shall not comment here upon some of the insights about hypothetico-deductive method that Descartes had acquired at this time, such as the importance of consilient confirmations or the way the experimental confirmation of an hypothesis turns on its comparison with competing ones.¹⁰ Of chief concern for our purposes is that we recognize how a combination of a priori and empirical elements formed an abiding feature of Descartes' mature conception of scientific method. Naturally, there can be no question that he continually sought to render empirical hypotheses as certain as possible. The two principal ways that he considered for increasing their certainty lay either in giving them, at last, an a priori demonstration or in setting up crucial experiments to decide between competing hypotheses. However, he did not believe that every hypothesis could be brought into the first path of certainty. Although he seems never to have ceased hoping for an a priori proof of the mechanical nature of light, he never dreamed of finding this sort of demonstration for other hypotheses, such as how the human eye operates. For this kind of phenomenon we could only try, in accordance with the Fourth Rule of Method, for as complete an enumeration as possible of all the relevant hypotheses; then by appropriate experiments we could hopefully narrow the range of hypotheses to one."¹¹ Clearly, this sort of quest for certainty is one that any rational inquiry must share.

The significant fact, then, about Descartes' mature conception of scientific method is not only that a priori demonstration and empirical testing form the means of justifying different parts of physical theory, but also the a priori area should be explored as far as possible before empirical investigation begins. Even if the ideal of a priori demonstration in physics now seems not just untenable, but perverse, we might still recognize an important truth dimly perceived in Descartes' conception of method. The building of empirical hypotheses should take place within a research programme (like the mechanism expressed in Descartes' a priori laws) that sets down some general constraints on permissible modes of explanation, indicates what are the important problems to tackle, and even has something to say about what will count as an acceptable solution—while itself having a far more indirect relation to empirical confirmation. However, instead of pursuing further this somewhat anachronistic line of thought, I shall now examine how, as his physical science shifts from the a priori to the empirical, Descartes' theory of knowledge takes up a new set of concerns.

Part II The Mathematization of Nature

As I mentioned at the beginning, there has in recent years been an increasing awareness of the extent to which Descartes meant physical inquiry to be empirical. This is so, even if these new treatments of Cartesian physics have often failed to capture, I believe, just what the role of empirical inquiry was for Descartes. But what has gone unnoticed altogether is that a central area of his empirical science has to do with investigating the character of human knowledge itself and its place in nature. Not only nature, but our knowledge of nature as well comes within the scope of inquiry turned empirical. This is what I shall be calling Descartes' empirical epistemology.

In order to understand how an empirical epistemology can emerge for Descartes, we might picture the Cartesian conception of inquiry as a grand circle. A priori epistemology provides the premises for a priori physical theory, but since such theory falls far short of giving a complete account of nature it must be supplemented by empirical hypotheses. But these hypotheses in turn can serve to deepen our understanding of the nature of human knowledge, from whose a priori insights the whole process set out. The path of inquiry, beginning with the a priori truths and then moving into the empirical, doubles back on itself in this way just because—in contrast to much of the philosophy that came after him—Descartes conceived of the theory of nature and the theory of knowledge as lying on a continuum, instead of being wholly different enterprises.

In fact, his empirical epistemology begins precisely at the point at which physical inquiry turns empirical. The character of physical inquiry shifts into a different key with the following questions. Are the qualities attributed to bodies by the a priori laws of nature—the mathematical qualities of extension, figure, and motion—the only qualities that physical bodies really have, contrary to what our perceptual experience would indicate? Or does physical theory concern itself only with certain properties of bodies, while abstracting from others? This problem concerns, of course, the mathematization of nature; since the mathematical qualities in question are geometrical ones (to the detriment of Cartesian physics), more exactly it is a geometricization of nature. Descartes' important insight, either overlooked or not pursued by his predecessors, from Cusanus to Galileo, who had espoused the programme of mathematizing nature, was that the development of this programme must proceed in tandem with a theory of perception that shows both that our ideas of nonmathematical properties, such as colour, resemble nothing in nature and that their occurrence is explicable in terms of a mathematical physics. Furthermore, the mathematization of nature was an empirical project. That is so, because he believed that the needed theory of perception must rest upon empirical hypotheses dealing with how the human eye works and what the nature of light is.

To be sure, the hypotheses that Descartes advanced about the structure of our perceptual system, in order to meet the mathematization of nature problem, form part of physiological theory. To what I am calling his empirical theory of knowledge belong, rather, the broad implications he drew from this to describe the relation between our scientific and perceptual images of nature as well as our relation as knowers to the natural order. Perhaps it may be objected that these are not 'philosophical' issues, supposedly because their pursuit must proceed against the backdrop of our knowledge of nature. Definitions of what counts as 'philosophical' are never very fruitful. Their usual intent is to exonerate the philosopher who makes them from having to learn anything about the areas of inquiry they exclude. Problems are a better guide than definitions. If our problem is to understand the relation between scientific knowledge and experience and the place of knowledge in nature, then a philosophical treatment of this problem is one that tries to approach it in the broadest possible way, making use of anything that may be appropriate. This was also Descartes' conception of philosophy, as the use of both a priori and empirical approaches to understand mind and nature in a book entitled The Principles of Philosophy would indicate. In this section, I shall discuss his mathematization of nature and the consequences he drew from it for an understanding of the relation between the scientific and perceptual images of nature. Descartes also exploited his physiological work to describe the place of knowledge within the natural order, and this I shall discuss in the subsequent section.

First, let us see just how the mathematization problem emerges as Cartesian physical inquiry becomes empirical. The three fundamental laws of nature and their deductive consequences are true a priori and characterize any possible physical world. That there does indeed exist such a world is something we infer, according to Descartes, from the fact that we experience many of our ideas as something passive, as a mental state caused by external objects, and that we have a divine guarantee that whatever we so clearly and distinctly perceive to be true must be true. Thus, once we see that there is a world of objects and movements, we may then conclude that it falls under the rubric of a 'physical world' governed by such a priori laws.¹² Here the mathematization problem first presents itself. Do objects really have only the properties mentioned by these laws?

Significantly, Descartes did not try to establish the mathematization of nature apart from an appeal to empirical considerations, at least in his mature period. He frequently extolled the greater clarity and distinctness enjoyed by perceptions of extension, figure and motion, in contrast to the obscurity affecting perceptions of colour or of hot and cold. But in none of these passages did he make use of this greater clarity to establish the mathematization of nature; it is always some other point that he was concerned to make.¹³ In fact, there is a letter that Descartes wrote to Chanut, several years after the publication of the Principles, in which he said explicitly that in that work the proof that ideas such as those of colours are not resemblances comes only at the end of the fourth part, at Principles IV. arts 189–98, where he refers to the physiological account of perception, given in such previous works as the Dioptrics, to prove the mathematization thesis.¹⁴ Since these physiological hypotheses belong to the empirical part of physical theory, the mathematization of nature, for Descartes, is an empirical hypothesis.

Thus, it is also clear that for Descartes the mathematization of nature depends upon empirical scientific hypotheses about the physiology of perception, and not merely upon everyday observations. This is, in general, an important point just because some philosophers, for example Jonathan Bennett, have claimed that the thesis that colour-ideas do not resemble objective properties of bodies does not require any 'recherché scientific information'.¹⁵ According to Bennett, reflection upon obvious empirical facts shows that the perception of an object as having some colour does not hang together with the rest of our knowledge in any way so systematically as does our perception of it as having some shape. From this he believes that we may infer that colours do not inhere in the things themselves. But, however poorly entrenched our colour-predicates may be, this argument does not have the force that Bennett thinks it has. At most, it could serve only to render easier the acceptance of the thesis that colourideas are not resemblances once that thesis has been independently confirmed on scientific grounds. Thus, Descartes was on the right track when he rested his mathematization of nature upon physiological hypotheses.

Before looking at the use that Descartes made of these hypotheses in his empirical epistemology, we must first see just what was the explanation of colour-perception that he presented in the Meteorology. He traced the perception of different colours to the differing rotational velocities of the light-corpuscles which, interacting with our eye in a mechanically explicable way, cause us to have colour-ideas. We may indeed speak here of the rotational velocities of the light-corpuscles, since only in regard to being 'transmitted' instantaneously can light be but a tendency to movement. This explanation is an empirical one in that both the mechanical nature of light and the account of how the eye reacts to these light-corpuscles and transmits their 'movements' to the brain and then the mind are, according to him, hypotheses that must be confirmed by experience. Now the reason he offers for the causal connection between the rotational velocities of lightcorpuscles and ideas of colours is a rather slender one: such velocities form the only remaining degree of freedom for the corpuscles and colour is the only aspect in which our perceptions of light vary.¹⁶

But whatever the shakiness (not to mention the falsity) of his explanation of colour-perception, Descartes went on to draw from it, and the mathematization of nature it made possible, an important philosophical consequence. This first result in his empirical epistemology is one that even an adequate physiological explanation of colour would inspire. Descartes was not content with claiming merely that our perceptual belief in colours is false. What he did was to set up a generalized concept of representation, according to which there are a number of ways our representations may represent features of nature besides resembling them. The physiological explanation of colour-ideas shows, that, even if they do not resemble actual features of nature, there are nonetheless interconnections among them that represent real relations in nature: the closer to red in the spectrum a colour is, the faster, according to Descartes, the corresponding rotational velocity of the light-corpuscles. Moreover, he might naturally have gone on to speculate about what must be the actual constitution of an object in order for it to reflect light-corpuscles of a certain rotational velocity; then a colour-idea would represent something of the object, though without at all resembling it in that regard. But this was one of the rare cases where Descartes did not seize an opportunity to put forth an hypothesis.

Both in Le Monde and the Dioptrics this distinction between representation and resemblance is laid out explicitly. There, for example, he compared ideas of colour to scripts or languages that bear a systematic relation to what they represent without resembling it.¹⁷ Descartes needed the generalized concept of representation to make sense of the relation between the scientific and perceptual images of nature. Although his physiological theory shows that certain of our ideas are not resemblances, it also shows how they do, in fact, represent actual properties of nature. In other words, while rejecting our 'natural interpretation' of colour-ideas, what Descartes calls our 'natural belief that takes them as resemblances, the theory places a new interpretation on them that indicates how they do represent. In this way, only, could he do justice to the fact that our ideas of colour prove useful in guiding our activities in the world. Descartes' general concept of representation expresses a view that we, too, must adopt if we are to understand how modern scientific theory at once characteristically corrects our perceptual image of nature and yet must ultimately be tested against our perceptual experience. As in the case of Descartes' explanation of colour-perception, the scientific theory that refutes our 'natural interpretation' of what we perceive is one that purports to explain why we have the perceptions or ideas that we do; this explanation we can understand as a new interpretation that tells us how our ideas really do represent. Yet the new interpretation is not tested against sentences expressing the natural interpretation it refutes (an incoherency often used by instrumentalists to discredit the idea that the correcting theory could count as being true). It is tested against an account of what perceptual ideas we do have.

In the past an exclusive concern with Descartes' a priori epistemology has portrayed his theory of representation as if it strove chiefly to determine with what right we can come to know that a representation is true or not. Because the empirical dimension of his epistemology was then overlooked, his need to examine the different kinds of representation went unnoticed. Indeed, Descartes considered the generalized concept of representation one of his most important discoveries. To the absence of this concept he traced the failure of the older view that perception occurs through objects transmitting 'intentional species' to the mind; on that view perception could be a matter only of whether an idea resembles an object or not.¹⁸ The generalized concept of representation, explaining how the mathematization of nature is possible, is thus the first key concept of Descartes' empirical epistemology.

Part III The Natural Setting of Human Knowledge

The second set of issues belonging to Descartes' empirical epistemology are ones that have to do with the place of human knowledge within the natural world. This area of his empirical epistemology arose because his physiological theories led inevitably to localizing the mind at a determinate position within the causal order of nature, namely in the vicinity of the pineal gland.¹⁹ Indeed, this conflicted head-on with his a priori distinction between mind and body, where spatial location was supposedly a distinctive feature of bodies alone. However, it is not with this conflict between a priori and empirical developments and the inadequacies of Cartesian dualism that I intend to deal, but with his empirical epistemology. Before we look at these further aspects of it, we must first take a glance at the theory of ideas that he worked out on a priori grounds and that served as the background for how he further exploited his physiological work for epistemological ends.

It is well known that the Cartesian concept of an idea is quite broad in scope, meaning as it does any sort of representation, but chiefly the content of a thought or a perceptual content. Ideas arise from two sources, either from the innate capacities of the mind (these are the ideae innatae) or from experience (these are the ideae advenitae, or adventitious ideas). When we use any of our ideas to re-interpret or combine other of our ideas, we end up with constructed ideas (or ideae factae). As for the nature of ideas themselves, Descartes often, when hurried, treated them as immediate objects of thought or perception, in the sense that they are mental items separate from the acts of thinking or perceiving them. But his more considered view (justifiable, as we shall see, within his physiological theory) was what we might today call an adverbial theory. Then he understood ideas as features of the mental acts themselves of thinking or perceiving, and not as separate items toward which those mental acts are directed. For example, in this spirit he defined an idea as 'the form of any thought (cogitatio) … by the immediate awareness of which I am conscious of that said thought', just after he had defined a thought as the mental operation (operatio) of thinking, perceiving, or willing.²⁰ On this view, we are immediately aware of our ideas only because we have immediate reflexive awareness of our thinking, and not because our ideas are separate items uncommonly close to our acts of thinking or perceiving.

Descartes used his physiological work to deepen his account of the character of human knowledge by placing human knowledge in its natural setting. He did this by examining the role of perceptual ideas in our empirical knowledge of nature. Remember then on a priori grounds Descartes believed that he could prove (in the Sixth Meditation) that the causal dependence of perceptual ideas on external objects is just as clear and distinct as our having such ideas at all. What he did in his physiological treatises, the Treatise on Man and the Dioptrics, was to show just how this causal chain proceeds. In general, he drew a mechanistic picture of how the impingement of the light from the object upon the eye causes certain movements to be transmitted along the optical nerve; these cause movements in the animal spirits of the brain, which in turn induce a particular movement of the pineal gland, which immediately gives rise to a particular idea in the mind.²¹ Now, to be sure, knowledge for Descartes consists not simply in having an idea but in the judgement that the idea is true or false. Thus, the causal chain of perception that results in a perceptual idea yields a necessary, but not a sufficient condition for empirical knowledge. The perceptual idea must figure in a judgement for there to be knowledge; and that perceptual idea may also be variously interpreted before a judgement is made. But we can think of Descartes' physiological work as placing empirical knowledge in its natural setting because the possibility and even the scope of such knowledge depends upon the physiology of perception. Furthermore, the mental acts of interpreting and judging, just as much as the ideas upon which they operate, occupy a determinate position in the causal order of nature. They, too, must take place in the vicinity of the pineal gland—contrary to what the distinction between mind and res extensa would seem to require.

I shall discuss two ways in which Descartes used his physiological work to fill in the natural setting of empirical knowledge. The first lay in his analysis of the causal link between the pineal movements and the resulting perceptual idea. Of course, he did believe that we cannot come to understand in what, in this case, the causal operation consists. But he did say something about the relation between pineal movements and perceptual ideas that is precisely as sophisticated as we should desire. In the Dioptrics he insisted that although the movement of the pineal gland causes the idea in the mind, it is not then the pineal movement that we perceive. The immediate object of perception consists in the content of the idea, while the pineal movements act upon the mind in such a way as to cause the mind to have such an idea.²² To believe that the pineal gland causes the mind to have a perception by causing the mind to perceive its movements would be to suppose, he wisely pointed out, that the mind itself has an eye to perceive those adjacent movements. Such passages indicate how successfully Descartes was able to integrate the representationalist and physiological components of his theory of perception.

Moreover, the very same kind of argument could have been used to justify his adverbial theory of ideas, the far more acceptable form of representationalism. No more than the mind has an eye to perceive brain-states does it have an eye to perceive ideas as separate mental items. Unfortunately, none of the passages that I have found where Descartes carefully laid out his adverbial theory indicates for what reason he preferred this account. It would not be unreasonable, however, to conjecture that behind this account lay his physiological claims about the relation in perception between brain-states and the mental states they cause.

In this way, Descartes' physiological speculations helped him to fill in his central epistemological thesis that our knowledge of the world takes place by means of our having ideas. Notice that his representationalist theory of empirical knowledge, as put forth on a priori grounds alone, was compatible with a range of different accounts. A priori introspection yields that adventitious ideas, that lie at the basis of our empirical knowledge, depend causally upon external objects. But this point is compatible with ideas being either the way things themselves look, or 'intentional species' of things transmitted without alteration to the mind.²³ Both of these alternatives were ruled out in virtue of taking mental states as caused by brain-states. But Descartes went on to clarify just what this causal relation means for the character of perceptual ideas. It appears plausible that this clarification led Descartes to his adverbial conception of ideas.

There is a second and more important way in which his physiological work contributed to an empirical account of human knowledge. His physiological investigation of vision showed, not only why our colourideas are not resemblances, but also under what conditions even the perception of the mathematical qualities of bodies can go astray. In the Dioptrics he showed how the accuracy of distance-perception diminishes when the object is either too near or too far and that bright objects appear closer than they actually are because the intensity of the light causes the same contraction of the pupil that occurs when it is focussed upon nearby objects.²⁴ Having ascertained the range of accuracy of the eye, he went on in the Dioptrics to show how the use of glass lenses, in telescopes or microscopes, could increase our access to the actual mathematical properties of bodies.

The important epistemological consequence that Descartes drew from this aspect of his physiological work lay in his coming to conceive our visual system as simply one kind of optical receptor among others. Our natural organs of perception he treated as lying on a continuum with what he called the 'artificial organs' that can supplement the deficiencies that nature has left us with.²⁵ He listed four conditions that any optical receptor should meet, and envisaged that sometimes different organs, whether natural or artificial, might satisfy some of these conditions better than others. These four conditions were that the receptor produce images that do not distort features of the object, that these images be detailed, that the light forming the images be strong enough to move the fibres of the optical nerve, and that the images represent at the same time as many different objects as possible. In short, what he was working at was a generalized concept of an optical receptor, under which our visual system would fall as simply one among other, 'artificial' ones. This concept formed part of an overall generalizing of our perceptual systems. Descartes' generalized concept of representation, which I discussed in the previous section, was intended to make sense of the fact that much of our perceptual experience represents, without offering resemblances of things. Now, when it comes to those perceptual ideas that can be resemblances—the ideas of the mathematical properties of perceived bodies—we see him generalizing along this axis as well. Artificial organs can, under a great many conditions, yield us resembling images, where our natural organs fail.

This overall generalization is an exceptionally important aspect of Descartes' empirical epistemology. By stressing how in many ways our perceptual image of nature proves inaccurate and how, even where it does offer us resemblances of the way things are, it is far less serviceable than the instruments we can construct, it served to undermine the traditional conception (deriving from both Greek and Christian sources) that God or nature has given us the perceptual organs we have because they naturally display the nature of the world we desire to understand. In short, this aspect of his empirical epistemology served to de-teleologize our perceptual system. This is a much-neglected aspect of the break with teleology characteristic of modern physical science, and yet it proved just as significant as the rejection of teleological theories of motion. Its importance lay, not least of all, in recognizing that progress in our knowledge of nature will come, not from the mere observation of nature, but from experimentation. Thus, Descartes thought that the senses should be subservient to the intellect, not simply because in ordinary life we make perceptual errors, but because more fundamentally we must take our perceptual experience as only an indirect access to the actual structure of nature. In the seventeenth century, it was Locke and Robert Hooke who chiefly continued the Cartesian break with the teleology of perception; Hooke recommended that 'The footsteps of Nature are to be trac'd, not only in her ordinary course, but when she seems to be put to her shifts, to make many doublings and turnings, and to use some kind of art in indeavouring to avoid our discovery.'²⁶ In fact, they went further in this development than Descartes himself. He was willing to de-teleologize our perceptual system probably only because he believed (it seems in contrast to Locke and Hooke) that he had a divine guarantee for the ability of our intellect to understand the world.

Since a characteristic feature of modern physical science has been not just its extension, but, in quite fundamental regards, its correction of our perceptual image of nature, physiological theories that de-teleologize perception have played a vital role in its development. In other words, put more generally, the modern theory of nature has required a theory of human knowledge as it exists within the natural setting physical theory describes. To that extent, modern epistemology had to have its empirical dimension, at least as long as it remained in contact with the growth of science. One of Descartes' unsung merits lies in his having perceived so distinctly and so fruitfully the need for an empirical epistemology.

The generalized concept of representation, the relation between perceptual ideas and the brain-states that cause them (as well perhaps as the adverbial theory of ideas), and the generalized concept of an optical receptor are the key features of Descartes' empirical epistemology. It is perhaps not surprising that his empirical epistemology has gone unnoticed for so long. Only recently has the myth been exploded that Cartesian physical science was thoroughly a priori. I have sought to show, what has not really been recognized, that for Descartes empirical inquiry was concerned not simply with a deeper understanding of nature, but also with a broader understanding of the nature of our knowledge of nature as well….

Notes

¹Discourse on Method, VI, AT, [Oeuvres de Descartes, ed. C. Adam and P. Tannery, 13 vols., Paris: Vrin/CNRS, 1879–1913], VI, pp. 64–5 (HR, [The Philosophical Works of Descartes, trans. Elizabeth S. Haldane and G. R. T. Ross, 2 vols., Cambridge: Cambridge University Press, 1970], I, p. 121).

²Le Monde, VII, AT, XI, pp. 37–45.

³Discourse on Method, VI, AT, VI, p. 64 (HR, I, p. 121); Principles, II, art 36–42.

⁴ In Regulae, rule XII (AT, X, p. 427; HR, I, p. 47) Descartes does imply, as in the case of the nature of the magnet, that sometimes we must rest content with hypotheses that are only empirically confirmable; but this passage is surrounded by other comments (AT, X, pp. 419–28; HR, I, pp. 41–7) that imply that all scientific knowledge must be deduced from self-evident 'simple natures'. Since the Regulae is so obscure a work, I have chosen to discuss Descartes' conception of scientific method as it emerges with the Discourse. In Le Monde, four years before the Discourse, he boasted that from the three fundamental laws of nature he could deduce a priori a complete account of nature (AT, XI, p. 47).

⁵Discourse on Method, VI, AT, VI, p. 65 (HR, I, p. 121); Principles, III, art 43–4.

⁶ For a recent statement of the view that by the end of the Principles Descartes had surrendered the idea that any physical truths can be demonstrated a priori, see D. Garber, 'Science and Certainty in Descartes', in M. Hooker (ed.) Descartes: Critical and Interpretative Essays (Baltimore, 1978, p. 146). Garber takes Principles, IV, art 206 to indicate that Descartes was 'uncomfortable' with having just abandoned, in the previous section, the possibility of a priori physical truths; in contrast, I take it to express Descartes' simply having finished the thought he began in the previous section—without a knowledge of God all of science would be hypothetical, but we do know God and He lends metaphysical certainty to the basic principles of physical science. For the view that, throughout the whole of his writings, Descartes considered physical science as thoroughly empirical and hypothetical, see A. Gewirth, 'Experience and the Non-Mathematical in the Cartesian Method', Journal of the History of Ideas, II (1941), pp. 183 ff.; also E. Cassirer, Das Erkenntnisproblem, Vol. I (Wissenschaftliche Buchgesellschaft, 1974; originally 1922), p. 469 ff. R. M. Blake, 'The Role of Experience in Descartes' Theory of Method', in Theories of Scientific Method, Seattle, 1960, claims that both a priori demonstration and experimental confirmation serve to justify the three fundamental laws of nature. This is an interesting idea, but the passages Blake cites are not convincing. An account generally similar to the one that I have presented may be found in L. J. Beck, The Method of Descartes (Oxford, 1952), pp. 239 ff., as well as in L. Laudan, 'The Clock Metaphor and Probabilism', Annals of Science, XXII (1966), pp. 73 ff.

⁷ Furthermore, Descartes believed that if we did not know the existence of God we would have no right to believe that the experimental confirmation of hypotheses had anything to do with their being true. Thus, in Principles, III, art 43, he traces the link between confirmation and truth to a divine guarantee; but in the next section, where no mention is made of God, he begins to hedge on whether hypotheses may be no more than practically useful (as opposed to true).

⁸Dioptrics, I, AT, VI, p. 83.

⁹Discourse on Method, VI, AT, VI, p. 76 (HR, I, pp. 128–9); To Mersenne, 17 May 1638, AT, II, pp. 134 ff. (PL, pp. 55–6).

¹⁰ For consilience, see To Morin, 13 July 1638, AT, II, pp. 196 ff. (PL, pp. 58–9); for comparative confirmation see Discourse on Method, VI, AT, VI, p. 65 (HR, I, p. 121).

¹¹ In the article cited above, D. Garber claims that at the time of the Discourse Descartes believed he could enumerate all possible hypotheses consistent both with the a priori principles and with the phenomena to be explained and then, by crucial experiments, he could show with deductive certainty which hypothesis was correct. But none of the passages cited by Garber rules out the interpretation that, according to Descartes, we should try for as complete an enumeration of possible hypotheses as we can; and this (if we leave aside the additional idea that they must be compatible with principles that are a priori) would hardly indicate that Descartes did not take the hypothetical method seriously (as, on his interpretation of the passages, Garber maintains). When in Discourse, V (AT, VI, pp. 40–1; HR, I, p. 106), Descartes writes that 'I have always remained true to the resolution I made … not to admit anything as true which did not seem to me clearer and more certain than the demonstrations of the geometricians', he is referring to principles (as the rest of the sentence makes clear), and in particular to the three fundamental laws of nature (as the subsequent sentence makes clear). This passage is used by Garber to support his claim that Descartes believed at this time that he could make the truth of his hypotheses certain.

¹² Although at this point the a priori laws of nature are known to apply to the physical world, there remains the problem how they may in fact be applied by us. Descartes' solution would lie in his theory of 'natural geometry' (see Nancy Maull's paper below).

¹³ See e.g. Third Meditation (AT, IX, p. 34; HR, I, p. 164). Probably as a result of the traditional view of Cartesian physics as thoroughly a priori his mathematization of nature is usually seen as a priori, not empirical. Cf. e.g., A. J. Kenny, Descartes (New York, 1968), p. 207.

¹⁴ To Chanut, 26 February 1649, AT, V. pp. 291–2:

It is necessary to remember, in reading this book [the Principles], that although I consider nothing in a body besides the sizes, figures, and movements of their parts, I claim nonetheless to explain there the nature of light, of heat and of all the other sensible qualities; so that I presupposed that these qualities are only in our senses, like tickling or pain, and not in the objects that we perceive, in which there is nothing but certain figures and movements, that cause the perceptions that we call light, heat, etc. This I did not explain and prove until the end of the fourth part….

[my translation].

¹⁵ J. Bennett, Locke, Berkeley, Hume (Oxford, 1971), p. 105.

¹⁶Meteorology, VIII, AT, VI, p. 334.

¹⁷Le Monde, AT, XI, pp. 3–4; Dioptric, IV, AT, VI, pp. 109–14.

¹⁸Dioptric, IV, AT, VI, p. 112.

¹⁹ See Treatise of Man, AT, XI, pp. 131, 143; Principles, IV, art 189; Passions of the Soul, AT, XI, p. 352 (HR, I, p. 345).

²⁰Reply to Second Objections, AT, IX, p. 124 (HR, II, p. 52).

²¹ At times, Descartes wrote that the perceptual idea occurring at the end of this sequence must be 'innate'; what he meant was that, since the figures and the movements in the sense organs and the brain give rise to ideas that do not resemble them, the mind must have an innate faculty that governs what the content of the perceptual ideas corresponding to these figures and movements will be. See Notes Against a Program, AT, VIII, pp. 358–9 (HR, I, pp. 442–3). Clearly, Descartes is not denying here that the knowledge of the world we gain through perceptual ideas is empirical.

²²Dioptric, VI, AT, VI, p. 130:

Now although this picture, in being so transmitted into our head, always retains some resemblance to the objects from which it proceeds, nevertheless … we must not hold that it is by means of this resemblance that the picture causes us to perceive the objects, as if there were yet other eyes in our brain with which we could apprehend it; but rather, that it is the movements of which the picture is composed which, acting immediately on our mind inasmuch as it is united to the body, are so established by nature as to make it have such perceptions.

[Translated by Olscamp, p. 101 in Discourse on Method, Optics, Geometry, and Meteorology (Indianapolis, 1965).]

Descartes did not have this insight from the beginning, since in an earlier work like the Treatise of Man he suggested that we perceive directly events in the brain; this is because he then thought of the ideas themselves as patterns in the animal spirits of the brain (AT, XI, pp. 176–7). N. K. Smith errs by attributing this earlier position to the whole of Descartes' thought, in his New Studies In the Philosophy of Descartes (London, 1952), p. 147.

²³ By rejecting the view that our perceptual ideas are the 'looks' of the things themselves, Descartes' physiological account of perception broke with our everyday understanding of perceptual knowledge. Ordinarily (in the case of vision) we believe that we perceive objects directly. Thus, what we perceive of an object, we think, is how the object itself looks in that situation; even if I know that that elliptical shape is actually a circular one, I believe that from this angle the object looks that way. The reason why this view comes so naturally is that in seeing an object we see ourselves seeing it, we see our bodies in a certain position vis-á-vis the object. It is this reflexive element that leads us to believe that we can see the object itself as it is causing us to see it. On the everyday view, see J. L. Austin, Sense and Sensibilia (Oxford, 1962).

²⁴Dioptric, VI, AT, VI, p. 144 ff.

²⁵ For this whole discussion see Dioptric, VII, passim.

²⁶ This passage is from the preface to Hooke's Micrographia (London, 1665). From Locke, see Essay Concerning Human Understanding, Book II, Ch XXIII, 12; there he says that God fitted our senses for our practical welfare, and not for our knowledge of nature (cf., however, Essay, Book IV, Ch IV, 4). For the same idea in Descartes, see Sixth Meditation, AT, IX, p. 66. Aristotle, as is well known, urged that the theory of nature should remain in harmony with 'ta phainomena'.