Smith’s Epistemology: Meaning, Context and the Nature of Science

The first problem we are faced with when seeking to reconstruct Smith’s epistemology is that, at least on first examination, he does not appear to have a great deal to say on the subject.¹ This in part can be seen as a reaction to the metaphysical system building of scholastic writing, what Smith calls ^t...

He certainly had a surfeit of such metaphysics during his years at Oxford, where he formed a low opinion of much that was taught in many of the European universities, and his views on the subject find frequent expression in the Wealth of Nations. He was also very well aware that lengthy discussions of metaphysical assumptions tended to have a distinctly soporific effect on the general reader, to whom his two major works were addressed. His letter to his publisher, Thomas Cadell, concerning a recently issued work on ‘Ethics and Natural Philosophy’ seems a fair statement of his position, for he found it

... as free of metaphysics as it is possible for any work on that subject to be.

but added,

Its fault, in my opinion, is that it is too free of them. But what is a fault to me, may very probably, be a recommenda­tion to the public. (Letter 261, Corr.)

Whatever the reason, if we are seeking in Smith’s extant work — written or reported — a comprehensive treatment of epistemo­logical and methodological issues, comparable to that presented by his more overtly philosophical contemporaries such as Kant or Hume, we shall be disappointed.

However, fortunately for our present purposes, what he does have to say is sufficiently explicit for us to be able to reconstruct his overall epistemological perspective with a high degree of confidence, and this can then be developed to show that his views on scientific methodology, are in fact a special case of this more general perspective.

This link between language and our perception of reality is in fact a convenient place to start our detailed investigation of Smith’s epistemology. Let us begin with ‘naming’.

Smith sees the act of naming as being both logically and historically amongst the first types of mental activity.

It seems probable that those words which denote certain substances which exist, and which we call substantives, would be amongst the first contrived by persons who were inventing a language. (LRBL, p. 9.)

Thus, he explains, in a primitive state, a man would have need to identify his own particular cave, or source of food, and would need to communicate these concepts to his fellow savage. From this would arise a substantive to cover the concept of this particular cave, or source offood, and these words would serve as a rudiment­ary language amongst the people in question.

The next step would be to generalise these specific substantives to cover classes of objects having similar qualities; to move from the notion of a particular cave, to that of caves in general. As he puts it:

The association of ideas betwixt caves, trees, etc, and the words they had denoted them by, would naturally suggest that those things that were of the same sort might be denoted by the same words. Thus it might perhaps be that those words which originally signified singular objects came to be Special names to certain classes of things. (Ibid.)

Smith is, of course, well aware of the lengthy epistemological discussions to be found in Locke and Hume on the difficulties of moving from the particular to the general, and makes no claim to have resolved these complex issues.² His reservations, however, are not central to his discussions and need not detain us in the present task.

The discussion in the Lectures on Rhetoric and Belles Lettres is in terms of a logical reconstruction of the history of language, but in the Essays on Philosophical Subjects (Astronomy) we find the same process discussed more overtly in terms of epistemology, pre­sented, as is usual with Smith, in the context of a psychological explanation of human propensities.

It is evident that the mind takes pleasure in observing the resemblances that are discoverable between different ob­jects. It is by means of such observations that it endeavours to arrange and methodise all its ideas, and to reduce them into proper classes and assortments. (EPS: Astronomy, p. 37.)

Now on the face of it, the above passages are merely logical implications of the tenets of what we may loosely call ‘Associa- tionism^,.³ If, following Hume, we accept that all knowledge is derived from concrete perceptual experience, presented in the form of sense data — what Hume calls ‘impressions’ — then it is reasonable to suppose that the reflection of these impressions in language will begin with the naming of individual objects, and proceed, by a process of abstraction, to general concepts.

The fact that Smith presents his discussion in the language of Associationism has led some commentators to assume that he also holds the ‘realist’ epistemology that is the normal concomitant of such a view. Bitterman, for example, has claimed that: ‘Adam Smith’s methodology was essentially empirical deriving its inspiration from Newton and Hume in contrast to the rationalistic method of the natural-law school of thought.’⁴

We must, however, go further than this, and probe beyond Smith’s use of language of Associationism if we are to establish his true position. He reveals this clearly in the History of Ancient Logics and Metaphysics where he explains that we can never know the true nature of external objects, so that even at the relatively simple level of ‘naming’ substantives there can be no question of a realist view of perception.

No corporeal substance is ever exactly the same either in whole or in any assignable part, during two successive moments, but by the perpetual addition of new parts, as well as the loss of old ones is in continual flux and succession. Things of so fleeting a nature can never be the objects of science or any steady or permanent judgement. (EPS: Logics, p. 121.)

and further:

No man ever saw, or heard, or touched the same sensible object twice... our sensations, therefore, never properly exist or endure one moment; but, in the very instant of their generation, perish and are annihilated for ever. (Ibid., p. 120.)

We must, therefore, abstract from the dynamic perception of objects to construct ‘specific essences of things’, so that even at this level, the mind must ‘work up’ the raw sense-data of experience and thus make an active contribution to perceived reality. Smith, like Kant,⁵ is postulating an unbridgeable gap between percep­tion and the object represented, so that external reality is essentially unknowable. Indeed he makes it clear that all we can ever be certain of are our personal internal states (EPS: External Senses, p. 141 et seq.) and — because of his somewhat curious belief that touch is somehow more fundamental than the other senses⁶ — the distinction between our own body and external objects (ibid., p. 135 et seq.). We cannot even rely upon the correlation between one sense — such as touch — and another such as sight, for:

As, in common language, the words or sounds bear no resemblance to the things that they denote, so in this other language [sight] the visible objects bear no sort of resemblance to the tangible objects which they represent, and of whose relative situation, with regard both to ourselves, and to one another, they inform us. (Ibid., p. 156.)

Fortunately, despite these limitations, our perceptions do in fact work at a practical level. There must, therefore, be sufficient ‘affinity or correspondence’ between sight and touch, and between our perceptions as a whole and the external world in general, for them to function.

The language which nature addresses to our eyes, has evidently a fitness of representation, an aptitude for signify­ing the precise things which it denotes, much superior to that of any of the artificial languages which human art or ingenuity have ever been able to invent. (Ibid., p. 158.)

and,

In this language of Nature, it may be said, the analogies are more perfect; the etymologies, the declensions and conjunctions... more regular than any human language. (Ibid., p. 161.)

Before proceeding, I would like to make two points which will be developed later in the chapter.

Firstly, the active role that the mind plays in ‘working up’ or interpreting the raw material of experience has fundamental implications for Smith’s views on science. In particular, what he says about the nature of ‘verification’ in science will be shown to have been profoundly influenced by his epistemology. Further support for the present interpretation will be developed in that context.

Secondly, it is important to note that language is the model that Smith uses when presenting his analogies. This will prove to be of relevance when we come to examine those interpretations of Smith that, whilst recognising the central role of analogy in his work, see mechanical analogy as the paradigm case.

Let us return to Smith’s account of the development of language. Despite the difficulties we encounter when forming substantives, they are fairly straightforward, when compared to the other parts of speech.

Here the contribution made by the mind in interpreting percep­tion is much greater than with the formation of even general nouns. Thus:

The quality denoted by an adjective is never seen in the abstract, but is always concorded with some substance or other; the word signifying such quality must be formed from it by a good deal of abstract reflection. (Ibid.)

— and our problems increase as we move into constructing those words which deal with pure relationships, as opposed to qualities possessed by classes of object, so that ^l... whatever difficulty there might be in the formation of adjectives, there must be still more in forming prepositions.’ (Ibid.)

We have, therefore, a distinction between two sorts of mental activity, which are, I think, best seen as opposite ends of a continuous spectrum, rather than discrete categories. On the one hand we have the process of classification, where the material provided by perception is grouped according to similarity of qualities, whilst on the other we have the process of postulating relationships between the objects we have classified. The mind plays a role in both these activities, but that role increases as we move towards the ‘relational’ end of the spectrum. There is also a difference in the kind of mental activity involved as we come to construct relational concepts, for whilst the fundamental act of interpretation by analogy remains, we in addition introduce a process of ‘abstract reflection’ or ‘judgement’, so that a higher degree of reasoning is involved.

The parts of speech, as we have seen, reflect this range of perceptual and mental activity, but Smith goes further than this, for he also holds that the grammatical structure of language displays this interaction:⁷

In every member [i.e. phrase] there are generally three principal parts or terms, because every Judgement of the human mind must comprehend two Ideas, between which we declare that relation subsists or does not subsist; concerning Two of these we affirm something or other, and the third connects them together and expresses the affirmation. (LRBL, p. 17.)

If we now turn to the Essays on Philosophical Subjects we can take Smith’s arguments a stage further, for here again we can find the same concepts presented in terms of human psychology, rather than linguistics. Here Smith tells us not only how we classify, and postulate relationships, but also, why we do so.

Classification he says, makes us feel that we understand the world better. If we are able to group an observation under some generic heading, we no longer feel it strange. We have assimilated our perception into an existing category, into some ⁱ... species or class of things...’ and even though

... we often know no more about them than about it, yet we are apt to fancy that by being able to do so, we show ourselves better acquainted with it, and have a more thorough insight into its nature. (EPS: Astronomy, p. 39.)

The criteria we use here, as Smith makes clear later, are familiarity, uniformity, coherence and generality. What we see must accord with what we already comprehend and understand, so that it can be slotted into the appropriate category, without disturbing our expectations. The role of analogy is crucial here, in that it explains ^l... that which was strange...’ in terms of ⁱ... that which was familiar...’ (EPS: Astronomy, p. 47).

Ifwe are faced with an unfamiliar observation, we automatically attempt to fit it into our network of preconceptions. If we are able to do so, we are content, and consider that we have an adequate understanding of what we see. However, if we cannot so assimilate the unfamiliar observation, we experience a feeling of ‘surprise’ followed almost at once by one of‘wonder’. That is, we are shaken in our beliefs about the world by our inability to classify, and at once begin to speculate as to why this might be. Smith makes clear that the surprise/wonder sequence is unpleasant, and that we seek to remove it by finding some explanation for the aberrant perception, if need be, by a suitable modification of our classifica- tory system.

An extension of the same explanation is offered for our need to establish relationships between the objects we have classified. Here:

When two objects, however unlike, have often been observed to follow each other, and have constantly presented themselves to the senses in that order, they come to be so connected in the fancy, that the idea of one seems of its own accord to call up and introduce that of the other. (EPS: Astronomy, p. 40)⁸

Ideas, says Smith, move faster than objects and if the sequence or development of objects follows the course of our ideas, they appear to be closely connected and thought glides smoothly along. Provided that we can fit our observations into the sequential structure of our expectations, there is no problem. However, when confronted by any new event or relationship that can not be so interpreted, then ^,... the imagination which accompanies with ease and delight the regular progress of nature, is stopped and embarrassed by those seeming incoherences...’ (EPS: Astronomy, p. 50).

We feel such events as a gap or void in the network of our preconceptions; events which should be linked together are seen as disconnected. This anomaly again gives us a feeling of surprise, followed as before by wonder. We seek by all possible means to bridge this gap within the context of our existing perspective. We will, for example, dismiss the observation as a random or chance event, which is thus atypical. If, however, the anomaly persists, we will modify our perceptual framework in order to fit in the new event, and thus soothe the imagination.

Here again the mind plays an active role in postulating new hypotheses about the nature of reality, which ^t... by representing the invisible chains which bind together all these disjointed objects, endeavours to introduce order into this chaos of jarring and discordant appearances’ (EPS: Astronomy, p. 46).

Smith again lays considerable stress on the role of analogy; the principles postulated by the imagination must, as far as possible, be analogous to those we accept already. The criteria as before are in terms of familiarity, coherence, simplicity and generality, and to the extent that they are satisfied, we have a feeling of ‘admiration’ for the new hypotheses.

An important implication of this view is that these criteria function in the context of, and are applied relative to, our existing framework of interpretation. That is to say that meaning is inextricably contextual, and the criteria for an articulation, or modification, of meaning are, in this sense, ‘aesthetic’.⁹ This point will prove to be important as the discussion proceeds.

This then is Smith’s basic epistemological perspective. The picture he paints is of an ongoing interaction between men’s mental activity and perceived reality. The raw material of our perception is furnished by an external reality, but is ordered by the mind which interprets what is ‘seen’ in the context of a pre­existing framework of expectations and beliefs. The causal links we postulate are the produce of ‘abstract reflection’ and the judgement of the mind, and their continuing adequacy as an analogy of ‘reality’ is measured in terms of the coherence they impose on the discordant appearances of nature.¹⁰ This whole dynamic process is reflected in the structure and content of our language.

This epistemological perspective permeates all of the Essays on Philosophical Subjects, and much of the Lectures on Rhetoric and Belles Lettres. It is used by Smith not only to explain the development of language and (as we shall see) science, but also appears in his accounts of logic, history, theatre and painting,¹¹ all of which, like language, are analogies of reality.¹²

This has been clearly seen by Ralph Lindgren who, although approaching Smith from a markedly different perspective from the present intepretation, has come to very similar conclusions on several key points. He interprets Smith in terms of linguistics and maintains that his model is in fact one of‘sign signification’, in which the imitative arts seek to provide analogues of reality by constructing systems of conventional signs. Lindgren’s inter­pretation sees the perceptual anomalies that Smith discusses as indicating ‘... an inadequate semantic formula...’ which makes the observer feel a gap that ‘... divides the simple unity of the objective event from the complex unity of expression and image...’ (ibid.) so that we ‘... attempt to bridge this gap by revising the semantic formula ordinarily associated with the sign object’ (ibid., p. 909 et seq.).ⁱ³

He maintains, I believe correctly, that because Smith sees language, art and science as systems of conventional sign signification, he accepts that there is no possibility of objective confirmation in terms of a correspondence test with reality.

We must now move beyond an interpretation of Smith’s general epistemological position to look specifically at his views on the nature of scientific knowledge. This, in fact, is but a small step for Smith sees scientific inquiry as nothing more than a formalisation of the basic interaction that we have already examined. In other words, as society advances, the efforts of the mind to provide connecting links between objects or events, and thus impose order onto chaos, becomes formalised into ‘philosophy’, which in eighteenth-century usage includes both philosophy and science. Thus: ‘Philosophy is the science of the connecting principles of nature.’ {EPS: Astronomy, p. 45) and, ‘... may be regarded as one of those arts which address themselves to the imagination.’ (Ibid., p. 46)

Smith’s discussion of the development of a science of primary elements in his History of Ancient Physics is worth quoting at length, in that it makes some of the links between his general epistemology and his view of science very clear:

To render, therefore, this lower part of the great theatre of nature a coherent spectacle to the imagination, it became necessary to suppose, first, That the strange objects of which it consisted were made up out of a few, with which the mind was extremely familiar: and secondly, That all their quali­ties, operations, and rules of succession, were no more than different diversifications of those to which it had long been accustomed, in these primary and elementary objects. (EPS: Physics, p. 107)

The role of the imagination, the conventional nature of scientific knowledge, the role of analogy and the need for familiarity, simplicity, coherence and generality, are all, more or less, explicit in this passage, which is closely analogous to Smith’s discussions of the origins of language and of pre-scientific perception.

Further, he also uses the same psychological framework to explain scientific activity in terms of surprise and wonder, and suggests that it is the prime function of science to alleviate wonder and that any other advantages are secondary. As he puts it: ‘Wonder, therefore, and not any expectations of advantage from its discoveries, is the first principle which prompts mankind to the study of Philosophy’ (EPS: Astronomy, p. 50).

Scientific ‘models’ are, therefore, formalised systems of speculative hypotheses which seek to soothe the imagination, by providing linking mechanisms which render a potentially chaotic reality, coherent and meaningful. They differ from everyday perception and interpretation, only in the degree of their abstraction; in what we would now term their reliance on theoretical constructs. The difference between primitive man’s attempts to form a preposition which would express the relation­ship between two objects, and the system of, say, Kepler, is thus only one of degree. The same epistemological framework covers both, but they function at different levels of abstraction.

The transition from one system, or model, of scientific interpre­tation to another — say from that of Kepler to that of Newton

— can thus be seen as a change from one interpretive structure of preconceptions and expectations to another. In other words, scientific laws and theories are intrinsically contextual; they function in terms of a particular perspective, take their meaning from that context, and are judged by standards that are largely internal.¹⁴

To the extent that we accept what has been argued already concerning Smith’s epistemological position, it will be clear that the rejection of one scientific perspective in favour of another can­not simply be the result of objective comparisons with reality. Since the ‘real chains’ of causal connection are inaccessible, verification, or refutation, must involve more than merely ‘empirical’ observation. There must of course be a process of observation, from which anomalies may arise, but what we see

— or indeed look for — will depend upon the primary principles which form the interpretive context in question. Smith is, in fact, quite explicit about this point. For example, when discussing the relative merits of the astronomical systems of Copernicus and Ptolemy, he says that the former was more accurate due to obser­vational errors in the system of Ptolemy. This ^ζ... ought naturally to have formed a prejudice in favour of the diligence and accuracy of Copernicus in observing the heavens...’ (EPS: Astronomy, p. 76) but, ‘... it ought to have formed none in favour of his hypothesis; since the same observations and the result of some calculations, might have been accommodated to the system of Ptolemy’ (ibid.).

The rejecting of Ptolemy’s system was not, therefore, because of empirical evidence against it. Why was it in fact rejected? Copernicus’s system, says Smith, offered an easier method of calculating the position of the planets in order to predict their movements, and:

The superior degree of coherence which it bestowed upon the celestial appearances, the simplicity and uniformity which it introduced into the real direction and velocities of the Planets, soon disposed many astronomers, first to favour, and at last to embrace a system which thus connected together so happily, the most disjointed of those objects that chiefly occupied their thoughts, (ibid.)

The criteria used are thus, as in Smith’s basic epistemology, coherence, simplicity and generality. Familiarity is not explicitly mentioned here, but later in the same essay Smith develops the theme in a discussion of Kepler’s system of laws, and their associated model:

The imagination, when acquainted with the law by which any motion is accelerated or retarded, can follow and attend to it more easily, than when at a loss, and as it were, wander­ing in uncertainty with regard to the proportion which regu­lates its varieties; the discovery of this analogy therefore, no doubt, rendered the system of Kepler more agreeable to the natural taste of mankind. (EPS: Astronomy, p. 89)

However, the analogy was lacking in simplicity and was thus not fully adequate.

Smith’s discussions of the development of astronomy and physics make it quite clear that he sees scientific activity as an ongoing dynamic process, in which successive systems are first modified in order to preserve the internal coherence of their primary principles, and then rejected in favour of those which offer more plausible explanations. Science takes place within the nexus of metascientific preconceptions and assumptions under­lying the model in question, and in the face of anomalies, we modify peripheral aspects of our model in order to preserve its coherence and generality. What we postulate as an acceptable modification will depend upon the nature of our existing perspec­tive, and our subsequent observations will also be interpreted in this context.

Smith specifically instances the ‘earth centred’ model of ‘Celestial Spheres’, where continuing anomalies were interpreted in terms of the existing model, which was, as a result, repeatedly modified. Eventually: ‘The system had now become so intricate and complex as those appearances themselves, which it had been invented to render uniform and coherent.’ (EPS: Astronomy, p. 59.)

The system thus became aesthetically unacceptable, so that the imagination was no longer soothed, and was eventually rejected in favour of the helio-centric model.

Smith’s belief in the conventional nature of scientific know­ledge — in the mind’s active contribution to the totality of percep­tion — extends even to the model of Sir Isaac Newton, which in the second half of the eighteenth century was the paradigm case of‘hard-science’. Newtonian Mechanics, he says, is the accepted system of the day; the framework within which contemporary science is taking place (EPS: External Senses, p. 140). This makes it difficult to recognise that Newton’s model, like all systems before it, is a conventional interpretive perspective, so that

^ζ... even we, while we have been endeavouring to represent all philosophical systems as mere inventions of the imagina­tion, to connect together the otherwise disjointed and dis­cordant phenomena of nature, have insensibly been drawn in to make use of language expressing the connecting prin­ciples of this one, as if they were the real chains which Nature makes use of to bind together her several operations. (EPS: Astronomy, p. 105)

Given this interpretation of Smith, it is clear that there cannot be any fundamental distinction between the physical, and the social sciences. Both are seen as extensions of man’s basic interaction with nature, and must, therefore, share the same underlying epistemology. This is very clear in the Essays on Philosophical Subjects, where Smith uses exactly the same format when discussing, say, Plato, Aristotle and Descartes as when speaking of Kepler or Newton. Further confirmation is provided by the general approach of both the Theory of Moral Sentiments and the Wealth of Nations (and indeed of the reported Lectures on Jurisprudence') where we will search in vain for any suggestion that the sciences of morals, or the ‘general principles of law and government’ are in any significant sense different from the laws of physics.

Indeed this interpretation gains further support from the one distinction Smith does make, for he makes clear that the moral (and by implication, social) sciences are more precise and rigorous than the physical sciences. Thus in the Theory of Moral Sentiments, when discussing the work of Mandeville, he says:

But how destructive soever this system may appear, it could never have imposed upon so great a number of persons... had it not in some respects bordered upon the truth. A system of natural philosophy may appear very plausible, and be for a long time very generally received in the world and yet have no foundation in nature... [but]... it is otherwise with systems of moral philosophy, and an author who pretends to account for the origin of our moral senti­ments, cannot deceive us so grossly, nor depart so very far from all resemblance to the truth. (TMS, VII.ii.4.14)

Thus to the extent that a distinction can be made, it is clearly the moral sciences that are ‘harder’ than the physical sciences.

Two further points need to be made about this passage: Firstly, on the present interpretation, the reason for this distinction lies at the heart of Smith’s epistemology, in his view of the difference between internal states, and external objects. The physical sciences are concerned with external events, so that even though the criteria for evaluating such a system are anthropocentric — being a function of the model’s ability to soothe the imagination — the interconnections and relationships postulated are, as it were, at one step removed from our internal states. This is not, however, fully true of the moral (and I will argue social) sciences, which, for fundamental principles, appeal to internal states and feelings. True, they move rapidly beyond this in postulating that other people are formed along similar lines, but at least we have the possibility of checking the plausibility of the analogy we need to make, against our own internal perceptions.

This interpretation is supported by Smith’s explanation of the distinction made in the quoted passage, which is in terms of an analogy which distinguishes between ‘... a traveller [who] gives an account of some distant country...’ in the case of the physical sciences, and ^,... an account not only of the affairs of the parish we live in, but of our own domestic concerns...’ (ibid.) in the case of the moral sciences.

Secondly, ‘resemblance to the truth’ in the quoted passage should be interpreted as meaning that a more coherent, familiar and general analogy is needed to convince us of the adequacy of a moral system. This again finds some support later in the same passage, when he says:

The author who should assign, as the cause of any natural sentiment, some principle which neither had any connection with it, nor resembled any other principle which had some such connection would appear absurd and ridiculous to the most injudicious and inexperienced reader, (ibid.)

So far it has been argued that Smith’s approach to science is fundamentally homogeneous, in that he sees the physical, moral and social sciences as formalised extensions of the basic interaction found at the level of everyday observation. To the extent that this is accepted, it follows that there can be no ‘Adam Smith Problem’, in the sense of there being significant differences of methodological approach within his major works.

Numerous dichotomies have, however, been postulated; for example, that the Theory of Moral Sentiments relies on natural law and a benevolent Providence, whilst the Wealth of Nations does not (Viner¹⁵); that the genuine science in Smith’s work is confined to (parts of) the first two books of the Wealth of Nations (Schumpeter¹⁶); that a distinction can be made between Smith’s inductive approach and his deductive explanation (Cropsey¹⁷), or that we must distinguish between Smith the sociologist and Smith the economist (West¹⁸). The list is by no means exhaustive.

It is significant that all of the quoted studies proceed by establishing contemporary definitions of scientific methodology, and then seek applications of the approach in question in Smith’s major works. It is, however, very much more difficult to find support for any of these views within Smith’s own writing on methodology, either in his explicitly epistemological treatments in the Essays on Philosophical Subjects, or in more down-to-earth discussions in his ‘applied’ work. Let us allow Smith to speak for himself, by juxtaposing two quotations, one from each of these sources.

Firstly, his well-known advertisement to the sixth edition of the Theory of Moral Sentiments, where he makes reference to the final paragraph of the first edition, where he said

... and that I should in another discourse endeavour to give an account of the general principles of law and government, and of the different revolutions which they had undergone in the different ages and periods of society, not only in what concerns justice, but in what concerns policy, revenue and arms and whatever else is the object of law.

This clearly implies (as the editors of the Glasgow edition point out) that the ‘other discourse’ would ‘continue the sequence of thought set out in the Theory of Moral Sentiments’.¹⁹

But Smith continues:

‘In the Enquiry concerning the Nature and Causes of the Wealth of Nations, I have partly executed this promise; at least so far as concerns policy, revenue and arms. What remains [is] the theory of jurisprudence... (Ibid.)

There is no evidence of any perceived dichotomy here; the Theory of Moral Sentiments, the Wealth of Nations and the proposed work on jurisprudence are presented as being three aspects of the same programme, grouped according to subject, not methodologi­cal approach.

Add to this the second quotation, taken from the Essays on Philosophical Subjects (Logic). Here he is discussing interpretations of Plato, saying that some authorities have sought to identify hidden meanings in his work. He refers to the

... strange fancy that, in his writings there was a double doctrine; and that they were intended to seem to mean one thing, while at bottom they meant a very different, which the writings of no man in his senses ever were, or could be intended to do. {EPS: Logic, p. 122.)

If we consider Smith’s own writings on methodology in the light of the above passage we are, I think justified in agreeing with Donald Winch²⁰ who has concluded that ‘... there is no Adam Smith problem in the original sense of a fundamental incompatibility of the ‘sympathetic’ ethic of the TMS and the ‘selfish’ ethic of the P½V.^,21

and that:

The latter work can, therefore, be regarded as a specialised application to the detailed field of economic action of the general theories of social (including economic) behaviour contained in the earlier work.

If we now strip Smith’s statements on epistemology of their psychological terminology, and concentrate on their logical and taxonomic status, it becomes clear that they closely conform to the modified Lakatosian format presented in Chapter 2.

We can begin with the metascientific world-view (the nexus of our preconceptions and expectations) which is not in itself testable, but which forms a ‘model’ or perspective within which observa­tions can be interpreted and classified. The analogue of Lakatos’s ‘negative heuristic’ here would be those fundamental principles, which can’t be changed without the abandonment of the entire world view. Examples here are explicitly provided by Smith; the belief that the earth is the centre of the universe in the ‘System of Celestial Spheres’, the alternative assumption of the heliocentric model, or the principle of gravity in Newton’s model and the associated mathematical laws.²² Smith’s own model of sympa­thetic interaction within a developing socio-economic framework will be shown, in due course, to be of precisely the same nature.

The positive heuristic, as with Lakatos, is then the range of problems generated by the world-view, which are seen as requiring attention. We form hypotheses about what sorts of relational links might exist between given classes of event, and these postulations are based upon our existing preconceptions. If observational anomalies challenge these hypotheses, our world-view also specifies the sorts of modification that will be consistent with its overall perspective. An example here would be observations made from within a geocentric model of the universe. Our world-view will firstly suggest the sort of astronomical observation we should be undertaking in order to confirm (or refute) our expectations of the relationships which might exist. Should these observations repeatedly ‘refute’ our hypotheses, as for instance when a planet appears to reverse its direction of travel, our world-view will specify the ways in which we can modify our peripheral hy­potheses, whilst leaving the protected core intact. We may thus add yet another epicycle to our model, but may not question its central assumptions. The fact that Smith conducts his discussion of this process in terms of psychological concepts — surprise and wonder — should not prevent us from seeing that a close analogy to our ‘positive heuristic’ exists.

We also find in Smith’s epistemology, the same emphasis on the dynamic character of scientific knowledge; as in Lakatos’s model, there is a continuous interaction between hypothesis and observation (see Chapter 2) and the ‘refutation’ of a world-view is a long term process which is dependent upon its adequacy to adapt in the face of anomalies. A degenerative world-view, such as the system of celestial spheres is thus one which can only defend its central core (as specified by the negative heuristic) by ad hoc hypotheses which eventually render the model aesthetically un­acceptable, within its own framework of reference. A progressive world-view is one which can accommodate anomalies without such a loss of coherence, thus preserving its internal elegance. The adequacy of the genetic empirical content generated by a world-view is thus evaluated in terms of explanatory power and overall plausibility, rather than in terms of verified predictions.

Moving beyond Lakatos into the realms of qualitative change is complex and we will need the arguments to be presented in Chapters 4 and 5 to complete the discussion; however, some points have begun to emerge, and these can be given a preliminary presentation at this stage, and developed more fully as the work proceeds.

Firstly, the transition from one system of interpretation to another can be regarded as a qualitative change of the type discussed in Chapter 2 above. Since meaning is intrinsically contextual, there will be a discontinuity when we move from one conceptual framework to another. An obvious example of this is the change in the way we ‘see’ the movement of a planet when we move from a geocentric to a heliocentric world-view. A par­ticular hypothesis, therefore, has meaning only within the context of the system that generates it. Further, it is assessed by standards of aesthetic acceptability that are grounded on the perspective of the system in question. It follows from this that relational hypotheses (and arguably, even classificatory systems) cannot be fully comprehended in isolation from the world-view that generates them, and, further, that alternative systems and the concepts they employ are, in this sense, incommensurable.

Secondly, we shall see in due course that Smith sees social institutions in very much the same way as he does other systems, and treats them as if they conformed to this same, basic, epistemol­ogy. That is to say, they too are intrinsically contextual, deriving their meaning and relevance only within the context of a specific socio-historical framework. Given this view, it follows that any ‘model’ constructed in an attempt to comprehend the development of social institutions must be ‘dialectical’ in the sense defined above (Chapter 2). That is, it must depict the qualitative transi­tion of the socio-economic framework which generates specific institutional patterns.

This, it will be argued, is why Smith invariably links social and moral science to a historical treatment, and seeks not just to establish general principles, but also (as he says in the advertise­ment to the Theory of Moral Sentiments already quoted) to give an account of‘... the different revolutions which they had undergone in the different ages and periods of society’.²³

Thirdly, it will also be argued that the specific model that Smith constructs to mirror this process of qualitative transition, is itself consciously constructed as a ‘system’ of the type described in his epistemological discussions. He has thus attempted to build a model which links together, and explains in terms of primary principles, the development of many aspects of human society. His model is one of sympathetic interaction, which finds its context within a progressively evolving socio-economic framework. It depicts the institutions generated by this process as being appro­priate to a particular socio-economic phase, but as becoming outmoded as society develops, and eventually being modified or replaced. As we shall see in Chapters 5 and 6, Smith measures this socio-economic development in terms of the degree of specialism, or ‘division of labour’, attained, and relates the institutional framework of society to this economic base.

This same basic format is used in all of Smith’s historical explanations. It is used in the Wealth of Nations to explain the progress of ‘policy revenue and arms’; in the Theory of Moral Sentiments to explain the development of morality and general rules of conduct, and in the Lectures to explain the evolution of jurisprudence.²⁴

Finally, given Smith’s belief in the conventional nature of scientific knowledge (discussed above), it follows that the adequacy of such a system cannot be judged in terms of normal empiricist standards of verification, which would require some form of realist epistemology. As we might therefore expect, Smith holds that its adequacy is to be judged, not in terms of predictive power, but rather in terms of its ability to provide a plausible analogy of the actual process of socio-historical development. That is to say, it must explain this evolutionary process by means of an analogy that is familiar, simple, coherent and general.

The final task in the present chapter is to conduct an examina­tion of these criteria, in terms of the perspective that has been presented so far.

The criteria put forward by Smith are closely interlinked. Coherence, for example is largely assessed in terms of familiarity, simplicity and generality, and simplicity is, at least in part, assessed in terms of generality. What is being evaluated is the adequacy of a particular model to function as an analogy of a fundamentally inaccessible reality. The criteria are thus, at least in part, self-referring; being established within the context of the model, they are, in this sense, aesthetic.

Familiarity, for example, is measured in terms of the perspective in question. The addition of a further epicycle in a geocentric world-view would be a familiar analogy and thus acceptable, but to use the same hypothesis to explain observational anomalies in the Newtonian model would be aesthetically unacceptable. In other words, modifications on the periphery of our model must conform to our metascientific perspective, by being the sort of ‘event’ specified as acceptable by the primary principles of the system in question.

Familiarity is closely linked with both simplicity and generality. The addition of a familiar analogy to extend or modify our model will not destroy its simplicity or detract significantly from its generality. However, if we were to introduce unrelated, ad hoc, modifications, the model would rapidly lose its acceptability.

Smith holds that to conform to this requirement for simplicity and generality a model must use a small number of inter-related basic principles, to explain as wide a range of phenomena as possible. This view finds clear expression in the Lectures on Rhetoric and Belles Lettres, where he makes a distinction between two forms of ^l... didactic writing containing an account of some system...’. {LRBL, p. 145).

We may, he says

... in Natural Philosophy, or any other Science of that sort... either, like Aristotle go over the Different branches in the order that they happen to cast up to us, giving a principle, commonly a new one, for every phenomenon; or in the manner of Sir Isaac Newton, we may lay down certain principles known or proved in the beginning, from whence we account for the several Phenomena, connecting all together by the same Chain. The Latter, which we may call the Newtonian method is undoubtedly the most Philosophi­cal, and in every science, whether of Morals or Natural Philosophy etc. is vastly more ingenious, and for that reason more engaging than any other. (LRBL, pp. 145-6)

Similar views are expressed in the Essays on Philosophical Subjects where, in the Astronomy (p. 99 et seq.), Smith praises Newton’s mechanics at some length for the universality of its basic prin­ciples; and an analogous position is maintained in the Logic, where Smith concludes that:

In every case, therefore, Species or Universals and not Individuals, are the object of Philosophy. Because whatever effects are produced by individuals, whatever changes can flow from them, must all proceed from some universal nature that is contained in them. (EPS: Logic, p. 119)

and he goes on to say that this fundamental requirement of natural philosophy also has its reflection in metaphysics which ^l... considered the general nature of Universals...’ and logic, which:

... from the general nature of Universals, and of the sorts into which they were divided, endeavoured to ascertain the general rules by which we might distribute all particular objects into general classes... (ibid., p. 120)

Simplicity has another aspect besides restricting the number of basic principles used to explain a class of events; it also refers to the complexity of the analogies used. Kepler’s model was ‘too difficult’ to be fully coherent, whilst that of Descartes was more satisfactory, being ^t... ingenious and elegant, tho’ fallacious...’ (Letter 5, Corr.).

Coherence is, at least in part, measured in terms of the other three criteria and is closely related to Smith’s view that we interpret new observations in terms of our existing expectations. If a new event can be assimilated into our system without breach­ing the rules of simplicity, generality and familiarity, then it conforms to our expectations and the perspective retains its coherence. Observations that cannot be so accommodated lead to a feeling of incoherence; of wonder and surprise at an event that makes us examine our framework of beliefs.

All of this will be shown to be of relevance when we come to examine Smith’s ‘impartial spectator’ whose impartiality is measured largely in terms of coherence, and thus of the other criteria we have examined. His objectivity lies in the universality and generality of his judgement relative to the existing nexus of appraisal. If coherence is to be preserved, the judgement made in any particular case must be appropriate to the social context — the dynamic element in Smith’s model stems from the fact that he sees social context as evolving through qualitatively different phases, as the division of labour increases.

Before concluding this chapter, there is one final point to be made about the role of analogy in Smith’s work. His frequent citing of Newton’s model as a paradigm case of analogy, together with his view that: ‘Systems in many respects resemble machines.’ (EPS: Astronomy, p. 66) has led some commentators to suppose that mechanical analogy is the basic requirement. Campbell, for example, follows Smith’s view that ⁱ... all theories should be a type of mechanical analogy, becomes in the end a sterile methodological principle.’²⁵

On the present view, however, this needs qualification. The basic requirement as I have tried to show, is for any form of analogy which is adequate to depict reality. It may be mechanical, but it could also be linguistic, perceptual, artistic or of any other form provided that it satisfies the necessary criteria.

Indeed, as Lindgren has pointed out, in so far as Smith has any ‘basic’ model it lies in the way in which language functions as an analogy of reality, so that ‘... he adopted language, not mechanics as the model of inquiry.’²⁶

If systems resemble machines, it is because both resemble the same fundamental interaction which takes place in the formation of a language. Smith makes this clear in the Lectures on Rhetoric and Belles Lettres when he says:

The languages in this have made advances a good deal similar to those in the construction of machines. They are at first vastly complex, but gradually the different parts are more connected and supplied by one another. (LRBL, p. 13)²⁷

We may, therefore, conclude that whilst mechanical analogy may conform particularly well to the criteria of familiarity, generality, simplicity and coherence, it is not for Smith, in any other sense, fundamental.

This chapter has covered a lot of ground and before proceeding a summary of the central issues would seem appropriate.

Firstly, it has been argued that Smith sees man as being in an ongoing process ofinteraction with a nature which is, in significant respects, inaccessible. The mind plays an active role in this process by constructing classificatory systems and relational analogies which seek to order, and link together, an otherwise chaotic flow of perceptual experience. Observation is conditioned by the expec­tations and preconceptions implicit in our current perspective (system). Modification of our systems is a more or less continual process, which is assessed by criteria which can be termed ‘aesthetic’.

Secondly, Smith sees this process as universal, so that all forms of human interaction with nature can be explained by the same epistemological model, which is also reflected in the content and structure of language. Science is an extension of this process; scientific models are formalised systems of interpretation which seek to impose coherence on an external reality. Scientific con­structs and hypotheses thus take their meaning relative to a specific context, and can only be fully comprehended within this framework. The move from one system of interpretation to another is, therefore, a qualitative transition from one context of meaning to another. Each system of interpretation consists of a (small) set of metascientific principles, which are fundamental to the model, which generate peripheral hypotheses aimed at linking together observations of appearances. These fundamental principles also specify the types of modification that will be aesthetically acceptable in the peripheral area of the model. Rejection of one system in favour of another is also on aesthetic grounds, the criteria being familiarity, simplicity, generality and coherence.

Thirdly, it will be argued below that the same epistemological perspective is to be found in the way Smith perceives the evolution of social institutions. These too are seen as being appropriate to, and deriving their full meaning from, a specific (socio-historical) context, defined in terms of the degree of specialisation attained. Qualitative changes occur when there is a transition from one socio-historical context to the next. During such a transition the relevance and meaning of a given institution — such as age as ‘the sole foundation of rank and precedency’ {WN, V.l.b.6) — may well change, so that what is appropriate in the age of hunters, for example, may be inappropriate in commercial society.

We shall see that the model Smith uses to depict this process of social evolution is itself a ‘system’ of the type we have been examining, and conforms to the same epistemological approach. The task of the next two chapters is to provide a detailed reconstruction of this model. Chapter 4 will concentrate on the fundamental principle which Smith employs, which is that of sympathetic interaction within a socio-economic context, and Chapter 5 will examine the nature of that context, and the dynamics of its qualitative development.

Notes

1. Cf. A.S. Skinner, A System of Social Science (Clarendon Press, Oxford, 1979), p. 14 et seq., for a discussion of this issue.

2. ‘To explain the nature and to account for the origin of general Ideas, is, even at this day the greatest difficulty in abstract philosophy.’ (EPS: Logics, p. 125.)

3. Despite the dangers of reification implicit in the use of such general labels to cover what was in fact a somewhat heterogeneous collection of views, it remains true that, if used with care, they form a convenient ‘shorthand’. Some further justification of the use of this label can be found in the fact that those who held and worked with the doctrine of the association ofideas recognised each other as a clearly defined ‘school’, sharing many common beliefs. Unpacking the term would involve us in a discussion of the relative influence of (at least) Hume and Locke upon Smith’s work, which is beyond the scope of the present work. However, for what it is worth, I would wish to maintain that, despite the immediate influence of Hume, it is to some of the ambiguities present in the work of Locke that we must turn, if we wish to trace the origins of many of Smith’s concepts.

4. H.J. Bitterman, ‘Adam Smith’s empiricism and the law of nature’, Journal of Political Economy (1940), p. 497. Cf. also J.R. Becker, ‘Adam Smith’s theory of social science’, Southern Economic Journal (1961).

5. That is to say, there is an unbridgeable gap between perception and the object represented, so that external reality is essentially unknowable. As Kant was to put it (1781): ‘The true correlate of sensibility, the thing in itself, is not known and cannot be known, through these representa­tions; and in experience no question is ever asked in regard to it.’ I. Kant, Critique of Pure Reason (Macmillan, London, 1950), p. 74.

6. ^l... the thing which presses or resists I feel as something altogether different from those affections, as external to my hand, and as altogether independent of it’ (EPS: External Senses, p. 135). Smith’s belief that touch is somehow more fundamental than the other senses is somewhat curious, but no doubt has its roots both in Locke’s epistemological distinction between primary and secondary qualities, and Newton’s doctrine of the impenetrability of matter. Smith in fact traces the origins of this latter view back to, ‘Lecippus, Democritus and Epicurus’ (EPS: External Senses, p. 140).

7. This interpretation is to some extent supported by the evidence that Smith does in fact regard the study of the use of language as sufficiently important to warrant a major study. Cf. his letter of 1 November 1785 to La Rochefoucauld (Letter 248, Corr).

8. Cf. also Essays on Philosophical Subjects (Astronomy), p. 37: ¹... custom and frequent repetition of any object comes at last to form and bend the mind or organ to that habitual mood and disposition which fits them to receive its impression without undergoing any violent change.’

The exposition is again in the language of Associationism (this might almost be Hume on causality) but, as before, the explanation proceeds well beyond that perspective.

9. For a discussion which is relevant to this issue, see R. Olsen, Scottish Philosophy and British Physics 1750-1880 (Oxford University Press, Oxford, 1975), p. 123 et seq.

10. We can find some support for the present interpretation in Smith’s letters. He wrote to George Baird, on 7 February 1763 (Letter 69, Corr) concerning a proposed system of rational grammar. ‘I am’, he says ‘convinced that a work of this kind... may prove not only the best System of Grammar, but the best System of Logic in any language, as well as the best History of the natural progress of the Human mind in forming the most important abstractions upon which all reasoning depends.’

11. See, for example, Essays on Philosophical Subjects: Rhetoric (p. 100), History (p. 110), Theatre (p. 122) and Painting (p.123).

12. ⁱ... it should be remembered, that to make a thing of one kind resemble another thing of a very different kind, is the very circumstance which, in all the Imitative Arts constitutes the merits of imitation.’ (EPS: Imitative Arts, p. 191.) Smith explicitly includes science in this category (EPS: Astronomy, p. 46).

13. J.R. Lindgren, ‘Adam Smith’s theory of inquiry’, Journal of Political Economy (1969).

14. In other words, science is a matter of postulating relational concepts, by a process of ‘abstract reasoning’. This is, I think what Haakonssen means when he says of Smith, ‘... the basic structure of his moral philosophy and indeed his philosophy generally...’ (is that) ‘... things have to be dealt with relationally, in their coherence with other things.’ K. Haakonssen, The Science of a Legislator (Cambridge University Press, Cambridge, 1981), p. 120.

15. J. Viner, ‘Adam Smith and laissez-faire’, Journal of Political Economy (1927).

16. LA. Schumpeter, History of Economic Analysis (Allen & Unwin, London, 1954).

17. J. Cropsey, Polity and Economy (Greenwood, London, 1977).

18. E.G. West, ‘The political economy of alienation’ Oxford Economic Papers (1969).

19. D.D. Raphael and A.L. Macfie (eds), Introduction to the Theory of Moral Sentiments (Clarendon Press, Oxford, 1976), p. 24.

20. D. Winch, Adam Smith’s Politics (Cambridge University Press, Cambridge, 1978), p. 10. He goes on, however, to warn against regarding the TMS as being ‘logically or philosophically prior in all respects...’ and against using it as ‘an ad hoc source to fill in gaps in the opinions presented in the WN...’ — a position which is not incompatible with the present interpretation which sees the TMS and the WN as specialised presentations of the same fundamental model.

21. The sympathetic/egoistic dichotomy referred to here was first postulated by German commentators on Smith, in the mid-nineteenth century: Hilderbrand (1848), Kries (1853) and Skarzynski (1878).

22. ‘Allow his principle, the universality of gravitation; and that it decreases as the squares of the distance increase, and all the other appearances which he joins together by it, necessarily follow’ (EPS: Astronomy, p. 104).

A Lakatosian ‘protected core’ par excellence.

23. Some support for this interpretation is to be found in A.S. Skinner, ‘From one point of view this is the classic pattern of cultural history —■ human activity released within a given environment ultimately causing a qualitative change in that environment —■ as illustrated, say, by the development of language or the transition from feudalism to the commercial stage.’ (p. 26.)

24. Ronald Meek has argued that the ‘four stages of theory’ is a fundamental organising principle which underlies much of Smith’s work. This view gained some support with the publication (in 1978) of the second set of lecture notes on jurisprudence, which placed greater emphasis on the ‘four stages’. However, on the present interpretation, the ‘theory’ is better seen as a heuristic device which displays aspects of a more fundamental methodological approach. (Cf. R.L. Meek, ‘Smith, Turgot and the four stages theory’, History of Political Economy (1971).

25. T.D. Campbell, Adam Smith’s Science of Morals (Allen & Unwin, London, 1971), p. 39.

26. J.R. Lindgren, ‘Adam Smith’s theory of inquiry’, Journal of Political Economy (1969), p. 899.

27. He goes on to qualify this, saying that ^l... the advantage does not equally correspond. The simpler the machine the better, but the simpler the language... the less it will be capable of various arrangements...’. (Ibid.)