The pincer movement of The Idea of a Social Science : Winch,Collingwood,and philosophy as a human science

Abstract

This article argues that, in order to understand Peter Winch's view of philosophy, it is profitable to read him together with R. G. Collingwood's philosophy of history. Collingwood was both an important source for Winch and a thinker engaged in a closely parallel philosophical pursuit. Collingwood and Winch shared the view that philosophy is an effort to understand the various ways in which human beings make reality intelligible. For both, this called for rapprochement between philosophy and the humanities. Like Collingwood, Winch wanted to reformulate philosophy as a form of human science. Both thinkers advanced a conception of logic where the validity of judgements, propositions, and thought are dependent on their function as instruments in human dialogue. In their treatments of logic, Winch and Collingwood were fleshing out their idea that questions concerning human meaningful behaviour also tie back to the question of what philosophical analysis is about. There is a deep connection between two main issues in both Collingwood's and Winch's writings: on the one hand, the need for ‘internal’ understanding of how human beings relate to reality, and on the other hand, their critique of the idea of logic as a self-sufficient system, external to historically embedded forms of life. At the core of their shared vision there was a comprehensive critique of metaphysical realism.

Introduction

In the beginning of The Idea of a Social Science (ISS), Peter Winch announced a ‘pincer movement’, aiming to show that ‘any worthwhile study of society must be philosophical in character and any worthwhile philosophy must be concerned with the nature of human society’ (Winch, 1990[1958]: 3). The first part of Winch's ‘movement’ was one that his readers immediately noted and debated. Winch argued that social studies were more intimately related to conceptual inquiry than was usually acknowledged. But the argument cuts both ways. In this article we wish to focus on his idea that philosophy, too, needs to reconsider its relation with the human sciences; that a philosophy that ignores the socially and culturally embedded character of thinking is deficient. After ISS, Winch mostly worked on traditional core areas of philosophy, not on the philosophy of the human or social sciences. However, we would maintain that, in much of that later work, he was spelling out the implications of his pincer movement to the practice of philosophy. Conversely, attending to his general take on philosophy helps us to understand the aims and methods of ISS.

Secondly, we hope to highlight some crucial similarities between Winch and R. G. Collingwood. Collingwood was both an important source for Winch and a thinker engaged in what in many ways was a closely parallel philosophical pursuit. These points of connection were not noted in the initial reception of ISS – the reception that, until today, has defined the place of ISS in philosophical and methodological debate. Collingwood was seen as an outsider, not an obvious point of reference in late 20th-century analytic philosophy. Our current situation is different. On the one hand, thanks to new scholarship partly based on manuscript sources published only from the 1990s onwards, a more nuanced understanding of Collingwood is available. On the other hand, we now have access to Winch's later work, both published and unpublished.

The overarching similarity between Collingwood and Winch was their view on philosophy as conceptual analysis. Philosophy makes explicit what lies implicit in thinking as its conceptual conditions. As Winch put it, philosophy is the study of ‘man's [sic] relation to reality’, of ‘what difference this will make to his life’ in various realms of inquiry (Winch, 1990[1958]: 9). It is neither an investigation of reality – of what kinds of entities exist in the world – nor a mere empirical description of how people think of reality. Winch also formulates the task of philosophy in the ‘Kantian’ question, ‘How is such an understanding (or indeed any understanding) possible?’ (ibid.: 22). For Winch as well as Collingwood, to describe human thinking already faces us with questions about how to distinguish between good and bad thinking, because that distinction in itself is part of the activity of thinking. But thinking belongs to a context of action or inquiry, and its implications can been seen only in light of that.

In the philosophy of the human sciences, their shared approach meant outlining the kind of epistemic interest that characterised explanations of action and distinguished them from explanations of natural events. In the philosophy of logic, it involved attention to linguistic meaning and its connections with practices of argumentation. Both thinkers engaged in criticism of a formalist take on logic (dubbed ‘Aristotelian Logic’) in favour of a context-sensitive approach (dubbed ‘Socratic Logic’). An important implication of their view on philosophy as conceptual analysis was their rejection of ontology as a meaningful pursuit – ontology being understood as the idea of a context-free ‘description of the world and what it's like and what is in it’ (Winch, 1995: 212). ¹ More specifically, it implied criticism of metaphysical realism and the context-free epistemology associated with it.

For these reasons, it will be fruitful to read Winch and Collingwood together. The connection between the critique of formalist logic and a critique of realism is especially explicit in Collingwood.

Winch and the (un)recognised influence of Collingwood

Winch was very explicit about his indebtedness to Wittgenstein, to the extent that, in the ensuing debates on ISS, readers often took him simply to be presenting what would have been Wittgenstein's views on the themes he was addressing (see Pleasants, 1999: 33–4). In contrast, while Collingwood was a frequent reference in Winch's work, systematic attempts to place that work in dialogue with Collingwood's philosophy are almost completely lacking. ²

The title of Winch's first book is an unmistakable allusion to R. G. Collingwood and The Idea of History (1994[1946]). The title was a suggestion by R. F. Holland, the series editor. ³ Winch cites Collingwood, ‘that under-estimated philosopher’ (Winch, 1990[1958]: 90), for support of some of his key arguments (ibid.: 90–1, 103, 113–14, 126, 129, 131–3), leading to ‘a new appreciation of Collingwood's conception of all human history as the history of thought’ (ibid.: 131). In addition to The Idea of History, The Principles of Art (1938) is among the Collingwood references in ISS, presenting a critique of the idea of magic as ‘pseudo-science’ (Collingwood, 1938: 58–61). In his other early work, Winch cites Principles of Art in ‘Understanding a Primitive Society’ in 1964 (Winch, 1972: 8–49). Collingwood probably read Evans-Pritchard's book on the Azande already at the manuscript stage (James, 2005: lxx). In ‘Human Nature’ (published in 1971), Winch cites Collingwood's Autobiography, a central presentation of his philosophical ideas (Winch, 1972: 85–6).

Manuscript material in the Winch Nachlaß at King's College London contains early notes by Winch on Collingwood, including a detailed resume of Part V (‘Epilegomena’) of The Idea of History. ⁴ The notes are undated, but appear to be written before or right after the completion of ISS. In his late years, Winch re-engaged with questions about the status of logic. That work remains partially unpublished. It does not include references to Collingwood, but his theme, the critique of ‘Aristotelian Logic’, is similar to Collingwood's critique of ‘Aristotelian Logic’.

On some points, however, it appears that Winch was not quite aware of how close he was to Collingwood. In ISS, Winch repeats some of the then current criticism levelled at Collingwood's idea of re-enactment (Winch, 1990[1958]: 132). He points out that complete re-enactment of past states of mind is impossible. On this score, Winch buys into the then current misconception of re-enactment as psychological Einfühlung and not, as Collingwood would have it, a critical reconstruction of past reasoning along with the historically contingent conditions of its meaningfulness (D’Oro and Connelly, 2015). Our present reconstruction (or re-enactment) of Winch's reasoning goes beyond his self-understanding, in the spirit of Collingwood's idea of re-enactment as critical engagement with the historical material.

Winch's dominant take on Collingwood's philosophy of history, however, was that Collingwood was highlighting the form of inquiry characteristic of explanations of action. Later Collingwood scholarship, partly based on previously unpublished material – such as the additions included in the new edition of The Idea of History – has laid to rest some moot questions of interpretation. Today it is obvious that Collingwood's idea of re-enactment was a description of the epistemic form of historical explanation, not a methodology based on the psychology of Einfühlung (van der Dussen, 1994: xxviii). This means that Winch was correct in his general perception of Collingwood, and that the parallel between Winch and Collingwood is closer than was understood when ISS was originally published.

The epistemic interest in explaining action

The main task for Winch in ISS was similar to the one Collingwood undertook to solve in The Idea of History: to describe the epistemic interest involved in explaining action. The idea was not so much to propose a new methodology (for instance, qualitative as opposed to quantitative) as to explicate a form of understanding already implicit in the human sciences (see Ahlskog, 2022). A difference was that, for Collingwood, historians were already more or less practising the forms of inquiry he believed to be essential in the research of action; the task was simply to make them explicit and to defend them. Winch believed, in contrast, that the social sciences were going seriously astray because they had misunderstood key assumptions implicit in their research questions.

The most obvious aim for both writers was to emphasise the difference between explaining action in terms of reasons and explaining natural occurrences in terms of causes. Both authors agreed that the word ‘cause’ can also be applied in explanations of action, but as Winch puts it, the ‘form’ of the explanation was different (Winch, 1990[1958]: xii; cf. Collingwood, 1998[1940]: 285). The specific point about the difference between reasons and causes became a focal point of the subsequent debate on ISS.

A central question in that debate was whether reasons are a form of causal influence on behaviour, as Donald Davidson maintained in his classic article ‘Actions, Reasons, and Causes’ (Davidson, 1963). Davidson's question concerned the ontological status of reasons, i.e. whether reasons were ontologically distinct from causes or, instead, ultimately a species of causes. In analytical philosophy of mind and action, the position was subsequently entrenched that actions allow of causal explanation (D’Oro, 2012). The opposing position, sometimes described as rationalist, was represented by writers such as G. H. von Wright, G. E. M. Anscombe, and William Dray. For them, to ask for the reasons of an action was not to ask for its causally efficacious mental antecedents. Rather, it involved, roughly, looking for a justification of some kind. We must assume an internal (or ‘logical’) relation between action and justification. They maintained that the connection between actions and reasons was best captured in the ‘practical syllogism’ formulated by Anscombe (1957: §33), where a combination of beliefs and desires logically entails action as a practical conclusion.

It is true that Winch and Collingwood were closer to the rationalist position, as they emphasised the role of reasons in explaining action. However, in an important respect, their chief concern was quite elsewhere. The reasons versus causes debate focused on the question what brings about the behaviour we understand as ‘action’. Is it legitimate to say that reasons bring about action, the same way as causes bring about events in the natural world? The question that interested Winch and Collingwood was, instead, what it means to understand human behaviour as action. To say that reasons ‘lie behind’ action is to say that the right way to understand action is to understand its reasons. The thing to look for is intelligibility, not causal antecedents, whatever they are. The epistemic interest involved in the human sciences was to understand behaviour as the expression of meaning – or as Collingwood put it, as the expression of ‘thought’ (Collingwood, 1994[1946]: 217; cf. Winch, 1990[1958]: 131). This strand of their thinking was not completely appreciated in the debate.

Winch's position came under scrutiny in subsequent debates on social ontology. Both critics and sympathisers have tended to understand the question he addressed as one of ontology – i.e. a question of the ways in which ‘meanings’ figure as building blocks of social reality. Importantly, this is also an idea that underpins relativist interpretations of Winch. In relativism of that kind, social worlds are seen as self-enclosed entities, with languages of their own, with set meanings, unavailable to scrutiny from ‘outside’ (see Ahlskog and Lagerspetz, 2015; Gunnell, 2016; Hutchinson, Read, and Sharrock, 2008).

It is common, especially among naturalists, to construe Winch as relying on a hypothesis about meanings as hidden mental entities. For example, Paul Roth derides, with obvious reference to Winch, philosophers of social science ‘who believe that there exist conceptual models lurking in mental space awaiting discovery’ (Roth, 1987: 138). He believes the assumption of definite meaning entities is a hypothesis completely undermined by Quine's argument for the indeterminacy of translation (see Ahlskog, 2022).

The emphasis on ontology, or on what kinds of entity exist in the social world, is quite alien to the concerns Winch was addressing. In the opening chapter of ISS, he indeed singles out the concept of ‘reality’ as a central concern for philosophy. However, philosophy is not about reality in itself, but about ‘the force of the concept of reality’: the difference that the concept of reality makes in human life (Winch, 1990[1958]: 9; emphasis in original). Nigel Pleasants (1999: 39) misrepresents the relevant passage as stating that Winch ‘seeks an account of being’, that is, an ‘ontology’. He further maintains that Winch, later in the book, supports a form of realist ontology of institutions, which according to Pleasants stands in contrast with Winch's ‘idealist’ take on epistemology. As proof, Pleasants cites Winch's opposition to Popper's methodological individualism (Ahlskog, 2022: 161–2; Pleasants, 1999: 45–6; Winch, 1990[1958]: 127–8). For Winch, institutions have a real presence in shaping the actions and thinking of individual agents. That is the view that Pleasants identifies as Winch's ‘realist’ ontology of institutions.

More recently, Alice Crary construes Winch as proposing ‘a distinctive social ontology … on which objective features of the social world are irreducibly ethical’ (Crary, 2018: 31; emphasis in original; cf. Tsilipakos, 2018: 82). Winch, on Crary's reading, tells us that values are just (or almost?) as real as are physical objects (Crary, 2018: 36). However, it seems to us rather that Winch was not making an attempt to determine what kinds of thing generally exist in the social world (or elsewhere). Instead, he asks: given specific needs of social explanation, what is the reality that needs to be taken into account? In other words, he defines reasons and meanings not as social or mental entities giving rise to behaviour, but as that which sociologists and historians must look for in order to render behaviour intelligible as action.

Against rationalism

The causalist take on social explanation meant that to explain a particular case of human behaviour (or indeed any event, natural, mental, or social) was to show that it was an instance of a general regularity. This view was described by Hempel as ‘the Covering Law Model’. As G. H. von Wright (1971) points out, explanation in terms of the Covering Law Model meant that explanation and prediction logically have the same structure. To explain is to show that the explanandum could have been predicted, because both explanation and prediction rely on one's knowledge of general laws together with prevailing initial conditions. This general view, drawing on Hume and exemplified by J. S. Mill, was of course one of the targets of the criticism of ISS. However, this does not imply that Winch chose instead to subscribe to the rationalist view. As von Wright also argued, the rationalist view – entailing the reconstruction of behaviour in terms of a practical syllogism – has the same structure as the Covering Law Model (von Wright, 1971: Chapter 1, Section 9). In both cases, to explain is to bring the explanandum under a general law or rule – which once more means that the resulting behaviour could have been predicted. On the rationalist reconstruction model, provided that the agent acts consistently, his or her action necessarily follows from the considerations included in the practical syllogism.

However, to think that Winch adhered to the rationalist reconstruction model is to ignore one of the central arguments of his book. Winch pointed out that any account of a relation – causal or conceptual – holding between two individual phenomena presupposes criteria of classification (judgements of identity) where individual phenomena are classified as either the same or not the same. To claim that a logical entailment relation holds between specific cases of, say, desire, belief, and action is to presuppose that one has understood the relevant judgements of identity. The further point here is that those judgements are embedded in the specific intellectual culture and cannot be assumed a priori as a rationalist armchair exercise. As Lars Hertzberg puts it, ‘Rationalism to the contrary, then, a person's practical reasoning cannot be used as a clue to understanding his way of life, for we must already have an understanding of his life if we are to understand his reasons’ (Hertzberg, 1980: 154). ⁵

The question of what, for a given person, counts as an ‘appropriate’ response to a situation can be solved only by attending to how this specific agent makes sense of the situation (see Winch, 1987: 30–1; see also Collingwood, 1994[1946]: 475, 494–5). In many cases, several responses may count as appropriate; but to recognise their appropriateness requires, in any case, engaging with culturally specific ideas of what counts as what (Winch, 1990[1958]: 86–8).

Somewhat ironically, Ernest Gellner, Winch's would-be sociological nemesis, notes the implication that any analysis of conceptual connections in social life must take account of the facts of their role in that life. However, Gellner takes that to be an argument against Winch's views, while it should rather count as one of the central points Winch was pressing. Gellner writes:

The Wittgensteinian idea of the correlativeness of activities (or institutions) and concepts – even assuming that Wittgenstein had solved the problems involved – cuts both ways: it implies not merely that in order to understand outer facts we have to know the ideas which give them life, but also that, in order to understand those ideas, we have to look at the outer goings-on which give them substance. And this is precisely what very many social scientists are doing anyway. (Gellner, 2003[1973]: 49)

The culturally or historically embedded character of logic motivated both Collingwood and Winch to consider the status of logic as part of thinking. That question was not fully explored in ISS or in The Idea of History, but it is addressed in their other work.

On the one hand, logic is applicable to all thinking, just because ‘thinking’, by definition, implies requirements of consistency. Collingwood put this by saying that the study of thought is a ‘criteriological’ activity (1999: 84–5, 108). When Winch describes meaningful behaviour as rule governed, his point is, similarly, that meaningful behaviour implies the application of a ‘criterion’ for right and wrong ways of doing things (Winch, 1990[1958]: 58). On the other hand, what ‘consistency’ amounts to in a given case can be seen only in the actual context – seeing in what ways specific connections obtain between different concepts.

Both Collingwood and Winch focused on questions about the status of logic as a representation of thinking, not on specific questions about formalisation. Their inquiries are therefore largely independent of any subsequent developments of formal logic. They asked what is implied when formalisations are presented as models to which thinking must adhere. Winch put the central idea this way:

The notion of logic is the notion of what is and what is not intelligible in human behaviour and it can be applied to anything men do. If it is abstracted from the ways in which men live it loses its significance as logic even as applied to relations between statements, for a statement is essentially something which men may make in the course of their lives. (Winch, 1972: 56)

This implied a criticism of attempts to present context-free applications of logical rules, which Winch and Collingwood dubbed ‘Aristotelian Logic’.

Critique of ‘Aristotelian logic’: Collingwood

In his Autobiography, Collingwood (2013[1939]) gives pride of place to a conception of a ‘logic of question and answer’, today often seen as one of his most central and original ideas. His suggestion is that truth, properly speaking, does not pertain to single statements but rather to ‘complexes’ of questions and answers. A statement, seen in isolation, might mean anything or nothing. What you might think of as two instances of the same statement would mean quite different things (and hence could have different truth-values) depending on how you identify the questions to which they are offered as answers.

Collingwood says that he got his idea of a logic of question and answer from his work in archaeology. Even today, history is (at least popularly) often described as collective memory – a repository of (preferably sacred) testimonies, handed down from a succession of authoritative sources. However, archaeology cannot raise such pretensions, since it is dependent on non-verbal material. Potsherds are normally not placed on a site in order to bequeath a testimony to future generations. Archaeologists must state their own questions and find their answers through targeted inquiries, directed at objects not originally intended to convey any kind of narrative. According to Collingwood, the special nature of its material made archaeology the launching pad of a methodological revolution in the historical disciplines.

In Collingwood's The Idea of History, the idea of questions and answers enters in two ways. On the one hand, historical scholarship makes progress because historians pose new questions to the material, rather than just repeating testimonies. On the other hand, understanding the actions, testimonies, and statements of historical agents is itself based on viewing those actions, testimonies, and statements as answers to questions, that is, as responses to challenges the agents encountered at the time.

The opposite of this conception was ‘Aristotelian logic’. Collingwood's polemic against it is included in The Principles of Art (1938: 259–69), An Autobiography (2013[1939]: 30–43), and The Idea of History (1994[1946]: 253–5). Collingwood's description of ‘Aristotelian Logic’ was his reconstruction of what he took to be the underlying, unifying assumption behind a number of superficially unrelated philosophical projects. It was the assumption that we can assess the validity of any actual case of reasoning by merely attending to its abstract form (Collingwood, 1938: 259; 1994[1946]: 253–4). Collingwood was not specifically targeting Aristotle, but also contemporary propositional logic, which of course rather saw itself as making a significant departure from Aristotle. Nevertheless, he thought the overarching aim had not changed. It was ‘to make language into a perfect vehicle for the expression of thought’ (Collingwood, 1938: 259). A ‘frightful offspring’ of this ‘propositional logic out of illiteracy’ was the project of ‘reducing’ propositions ‘to logical form’, Collingwood said, ‘ending, for the present, in the typographical jargon of Principia Mathematica’ (2013[1939]: 35–6, n. 1). The basic unit of the analysis was the ‘thought’ or ‘proposition’, the best linguistic expression of which would be the indicative sentence in the present tense. In the ideal case, according to ‘Aristotelian’ logic, ‘there is, or ought to be’ a ‘one-one correspondence between propositions and indicative sentences, every indicative sentence expressing a proposition, and a proposition being defined as the unit of thought, or that which is true or false’ (Collingwood, 2013[1939]: 35–6; cf. Russell, 1905).

In his alternative view, Collingwood emphasised that a sentence that stands alone is not a carrier of meaning or truth. In a statement that Winch also included in ISS, Collingwood compares a grammarian with a butcher, but preferably a butcher that, according to reported African custom, can ‘cut out a steak of a living animal, and cook it for dinner, the animal not being much the worse’ (Collingwood, 1938: 259; Winch, 1990[1958]: 126). The grammarian should inspect the sentence without killing the context.

A proposition has a recognisable meaning only as an answer to some possible (but not necessarily explicit) question. Conversely, to inquire into the meaning of a proposition is to try to see the question to which it is meant to be an answer (Collingwood, 2013[1939]: 29–43). This is an ‘historical’ exercise in the sense that the investigator must get an idea of what questions had ‘arisen’ when the proposition was put forward. To judge the truth of a statement is also to judge the previous questions and answers that have led up to it. The same thing, of course, is true of questions as well: in order to understand the meaning of a question you can ask why anyone would raise it – and you may question the meaningfulness of an interrogatory sentence if no context is provided. In other words, assertions and questions are equally context-dependent. One might think that this gives rise to an infinite regress: an assertion is dependent on a question, which is dependent on other questions and so on without end. But in practice that is not the case, for a question can be legitimately dismissed when it does not ‘arise’ (ibid.: 38).

Collingwood argued for what he called a ‘Socratic’ conception (1998[1940]: 156–8; 2005[1933]: 10–11; 2013[1939]: 35), in which the validity of judgements, propositions, and thought are dependent on their function as instruments in reasoning. Our assessment of the formal validity of a given instance of reasoning is parasitic upon our grasp of the sense of the relevant propositions and thoughts, as they unfold in lived experience:

A poet will say at one time that his lady is the paragon of all the virtues; at another time that she has a heart as black as hell. At one time he will say that the world is a fine place; at another, that it is a dust-heap and a dunghill and a pestilent conglomeration of vapours. To the intellect, these are inconsistencies. A lady, we are told, cannot be a paragon of virtue at one time and as black as hell at another; therefore a person who says it must be making it up.… On the poet's behalf it may be replied, to some one who argues that a lady cannot be both adorably virtuous and repellently vicious, or that the world cannot be both a paradise and a dust-heap, that the arguer seems to know more about logic than he does about ladies, or about the world. (Collingwood, 1938: 287–8)

One might believe Collingwood is just making the trivial point that you can mean words and sentences in different ways. The woman is virtuous in one sense and vicious in another. The relevant statements could, then, be disambiguated and formalised after all. Collingwood would no doubt have been happy to admit that such formalisations are available. The deeper point is that the very process that leads to the disambiguation – where we assess the different ways ‘virtuous’ and ‘vicious’ might be understood in the context – is where the actual work is done. Once that assessment is complete, there is nothing left for the formalisation to achieve. There may be an additional point: it may be important, especially in the context of a poem, to keep the sense of paradox. The analysis, ‘x is virtuous in one sense (V₁) and non-virtuous in another (∼V₂)’, will solve the contradiction but it will also dissolve the sense of bewilderment that the poet wanted to express.

Collingwood's ‘revolution’ in logic signals the detection of what Bernard Williams would call an ‘impurity’ at the heart of philosophical analysis. As Williams (1995: 148) writes, if philosophy is to have anything important to say, it must ‘address a lot more than philosophy’. Meaningful philosophy cannot consist of the mere analysis of pure logical form. It will forever be contaminated by the fact that philosophical analysis depends on our pre-existent, independent understanding of our forms of knowledge and experience. It relies on an understanding external to any a priori self-definition of the proper aims and methods of philosophy (see Moran, 2016: 317). The dependence of philosophy on history does not simply mean that history provides raw material for logical analysis. Collingwood argues that the authority and sense of compulsion associated with logical argument is itself is a function of the historically specific reactions and responses of human beings in actual arguments. To make sense of logical relations, one must look at their instantiations in vivo, not in vitro.

Critique of ‘Aristotelian logic’: Winch

Winch was mainly interested in the question of the status of logic as a formal science of thinking. A central point of reference for him was Wittgenstein's Philosophical Investigations (1953: I, §81), where Wittgenstein raised the question in what way logic was ‘a normative science’. On Winch's view (which he takes also to be Wittgenstein's), logic is normative because it presents an abstract object of comparison, applicable as a corrective to any reasoning. However, its application is context-dependent and often contested – which, in a sense, gives the formalisations of an actual argument a tentative and descriptive aspect.

Logic, as a science of thinking, is different from psychology because its subject matter is the distinction between valid and invalid arguments. In his Lectures on Logic, Winch says that a psychological survey might concern ‘e.g., the extent to which tiredness, noise, emotional disturbances etc. may interfere with people's reasoning capacities’. ⁶

As far as the psychologist's empirical survey is concerned, the fact that people reason incorrectly is just as important as the fact that they reason correctly (in some contexts it may be more so). The distinction between valid and invalid arguments is something that a psychologist making the sort of investigation I have given as an example would have to understand, but it would be presupposed by his investigation; it would not itself be the subject-matter of his investigation. ⁷

Without the distinction between sound and unsound reasoning, the notions of reasoning (or reason, rationality, thinking) ‘would have no sense’. ⁸

To say that someone was thinking or reasoning, and then reject as inadmissible the question whether he was reasoning soundly or not would be to be inconsistent. (Contrast, in this very important respect, the statement that somebody is, e.g., day-dreaming: here the question of the soundness or otherwise of his mental processes just does not arise). ⁹

A special trait in logic is that pointing out logical implication can be a corrective to someone's reasoning. From certain premises, a conclusion must follow. That in turn invites the question ‘What is the authority of logic?’, or ‘What is the authority of reason?’. Winch repeatedly quotes Lewis Carroll's short story ‘Achilles and the Tortoise’ (Winch, 1990[1958]: 55–7). ¹⁰ The story is a reductio ad absurdum, the point of which is to demonstrate that logic by itself cannot ‘take’ anybody ‘by the throat’ and force them to accept a conclusion. The ‘authority’ of logic derives from the fact that it depicts an instance of reasoning going on between people. You cannot reject the conclusion of an argument by pain of having to withdraw from further discussion. ¹¹

In the 1990s, Winch used the expression ‘Aristotelian logic’ to describe the sort of formalist approach that he opposed (Lagerspetz, 2019). This is apparently a term he got not from Collingwood, but from Wittgenstein. Winch pointed out that ‘Wittgenstein, so [Rush] Rhees says, used the term “Aristotelian logic” (with which he contrasted his own way of treating logic) to include Frege and Russell!’. ¹² Winch wanted to apply and vindicate (later) Wittgenstein's idea of logic.

For the ‘Aristotelian’, logic is a formal system implemented on actual cases of reasoning, to be used as a standard for judging their correctness. Logic is a kind of machine, in one end of which you feed premises and out of the other come conclusions. The machine never makes mistakes, thus whenever your premises are true, the conclusions will also be true. In Wittgenstein's words, the machine is ‘ideally rigid’, for the possibilities of its future movements ‘are already there in it in some mysterious way’ (Wittgenstein, 1953: I, §194). This is not an empirical discovery but a conclusion from the abstract structure, which knows no wear and tear and no distortion of the parts. Moreover, you may feed the machine profitably with premises even if you have no idea of what they mean. The system itself guarantees the correct solutions. According to Winch's summary, the ‘Aristotelian’ conception of logic has ‘two aspects’:

A conversion of what we say into ‘canonical form’ – which is supposed to express the real form of our thought, a form that is supposed to take its authority from its mirroring the structure of reality.

The idea that this structure exercises a special sort of constraint on our thinking. ¹³

For instance, the proposition ‘It is raining and it is not raining’ can be formalised as ‘p & ∼p’. The logical form (the typographical form) itself supposedly indicates that it is self-contradictory. Winch's response to this is that no two propositions per se contradict each other. Contradiction arises when people use propositions to say contradictory things. We find out whether two sentences contradict when we see how people react to them – in other words, what they mean by them. For instance, ‘It is raining and it is not raining’ may involve logical contradiction, but it may also be a good description of Swansea weather. Of course, if someone says, ‘It's raining and not raining’ and obviously means (something by) it, our first reaction is typically not to dismiss it but to give it a reading that makes sense. The typographic image of ‘p & ∼p’, on the other hand, is of course neither self-contradictory nor coherent in itself, for it might just be a design for a wallpaper pattern. ¹⁴ The expression ‘form’ papers over a certain ambiguity. On the one hand, ‘form’ is ‘the geometry of the written pattern’ (Carnap, 1963: 28), and logic may be seen as the art of converting one pattern to another according to rules of syntax. On the other hand, what makes the written pattern a logical form is its connection with thinking – a connection that, in any given case, must be demonstrated.

A consequence of this view is that you can never know whether two statements contradict each other unless you see how they are used. This may mean that you must see how you would use them. In his paper ‘Darwin, Genesis and Contradiction’, Winch takes issue with the idea that two statements ‘as such’ may be contradictory. Discussing the contrast between Genesis and evolutionary theory, he argues that you can settle the question of whether the two ideas about the origins of humankind contradict each other only by considering what kind of use you can think of giving to them – ‘whether, and if so how, [you] can live with both of them’ (Winch, 1987: 137). Thus it is a mistake to think that the work of defining their relation ‘has, as it were, already been done in a hidden realm of logic and simply needs to be revealed to view’ (ibid.: 139).

Questions about logical relations must be addressed from inside the various ways in which the world is intelligible to us. It is not as if we have, on the one hand, our ways of making sense of the world and then, separately, a cut-and-dried system called Logic. It is not as if the latter is a mind-independent ontological structure, somehow running through and regulating the former (cf. Diamond, 2003: 52). The project of weeding out incoherent or logically contradictory ideas must wait until we understand what, in the given case, would count as contradiction.

If logic is a formalisation of how human beings think it must, in other words, be logic-in-use. On this view, logic is an aspect of a social life where human beings agree or disagree. In ISS, Winch writes, ‘Criteria of logic are not a direct gift of God, but arise out of, and are intelligible only in the context of, ways of living or modes of social life’ (Winch, 1990[1958]: 100).

Winch contrasted ‘Aristotelian’ logic with what he called Socratic logic; in other words, the kind of dialectic that Socrates both exercises and describes in Plato's dialogues. The central feature of this dialectic is that it consists of interaction between specific people. The ‘binding’ character of an argument is in part dependent on the attitude of the person who puts it forward – whether she believes it, for what reason she presents it and what kinds of consequences she is prepared to accept. This dialectic has a moral aspect, which Socrates often stressed. In the Gorgias, as Winch pointed out, at crucial junctures Socrates makes his interlocutors feel ashamed for the positions to which they have committed themselves. ¹⁵ They are involved in self-contradiction, not because it is a contradiction per se to say disgraceful things, but because it is a disgrace for the speaker to do so. For instance, if Callicles took seriously his own insistence that there is no distinction between shameful and acceptable pleasures, he would come across as shameless – as a person of the kind that he does not want to be. Socrates was trying to call his interlocutors back to sobriety and the love of truth ‘through the candor on which he insisted’. ¹⁶

Critique of metaphysical realism

By focusing on the context-dependence of linguistic meaning, Collingwood and Winch also produced an argument against ontology as a philosophical pursuit, especially targeting metaphysical realism. Metaphysical realism is the doctrine that the existence of objects (at least some of them) is independent of any process of our coming to know them. Hence, for the realist, it must be meaningful to state that an object exists without assuming anything about our possible relations with that object. The problem here is that the nature of our possible relation to the object, in part, specifies the kind of object it is. For Collingwood and Winch, talk of the ‘reality’ of any kind of ‘thing’ – external objects, mathematical objects, God, the mind, the past, good and evil, and so on – must in each case be understood in relation to possible ways of relating to that ‘thing’. Lacking such connections, ontology is left without a defined subject matter; in other words, it is not a meaningful inquiry at all.

For Collingwood, resistance to realism was a life-long project. He gives it a central place in his Autobiography, connecting it with his proposal of a logic of question and answer. His criticism included Cook Wilson at Oxford as well as G. E. Moore and Bertrand Russell at Cambridge (Collingwood, 2005: 245). His main objection in the Autobiography was that realism treats questions of truth simply as a matter of matching claims about reality with reality in itself. The idea of simple comparison ignores the crucial function of thinking as problem solving.

The Oxford ‘realists’ talked as if knowing were a simple ‘intuiting’ or a simple ‘apprehending’ of some ‘reality’. At Cambridge, Moore expressed, as I thought, the same conception when he spoke of the ‘transparency’ of the act of knowing.… This doctrine, which was rendered plausible by choosing examples of knowledge statements like ‘this is a red rose’, ‘my hand is resting on the table’, where familiarity with the mental operations involved has bred not so much contempt as oblivion, was quite incompatible with what I had learned in my ‘laboratory’ of historical thought. (Collingwood, 2013[1939]: 25–6)

Realists thought of empirical knowledge in terms of a kind of presence; the best example being the experience of immediate eye contact or tactile contact, such as seeing a coloured patch or feeling one's own hand. In such ‘immediate knowledge’ (Moore, 1957[1910–11]: 123), no question about evidence or research method muddies the waters between the epistemic subject and the object of knowledge.

‘My hand is resting on the table’ was a kind of standard example, typical of philosophy seminar situations. In 1939, Moore used the example, slightly modified, in his (in)famous ‘Proof of an External World’ (Moore, 1959[1939]). Moore supposedly demonstrated the existence of objects external to the mind by producing his two hands. Collingwood's implicit objection in the Autobiography was that Moore had not specified a question to which his example was an answer. No one had doubted that Moore had two hands.

Winch was more lenient in his comment on Moore in ISS. Presumably drawing on interpretations by Norman Malcolm and Alice Ambrose (Ambrose, 1952[1942]; Malcolm, 1952[1942]), Winch presented a rather benevolent gloss on what Moore had been doing:

Moore was not making an experiment; he was reminding his audience of something, reminding them of the way in which the expression ‘external object’ is in fact used. And his reminder indicated that the issue in philosophy is not to prove or disprove the existence of a world of external objects but rather to elucidate the concept of externality. (Winch, 1990[1958]: 10)

On this view, Moore's paper was not a defence of realism, but an attempt to call us back from metaphysical to ordinary language. It was a reminder of how we ordinarily prove the existence of something or other: we may do it by showing up that thing. One problem with this interpretation is that Moore explicitly rejected it when it was presented to him (Moore, 1952[1942]: 668–74). The generally benevolent attitude shown by Winch and other Wittgensteinians towards Moore's ‘Proof’ might just be due to the known biographical fact that Wittgenstein personally liked Moore (but not necessarily his philosophy). Wittgenstein's own comment is more incisive. Why didn't Moore choose an example of something that he knew but the others did not – for instance, the distance between two stars? Moore could then have proceeded with an account of the inquiries he had made to find out, ‘pursuing some line of thought which, while it is open to me, I have not in fact pursued’ (Wittgenstein, 1972: §84). That would indeed have been a real example of how facts about the external world are established. However, it would not have been a case of what Moore wanted for his realist project, namely ‘immediate’ knowledge. The question that now arises (for Collingwood as well as Wittgenstein) is whether Moore's examples of ‘immediate knowledge’ are really cases of knowledge at all; and if so, how they would relate to the more complex cases that we would normally call knowledge (see Lagerspetz, 2021).

Winch gradually builds up his non-realist position in the chapters included in Trying to Make Sense (1987). The following example comes from Winch's essay ‘Im Anfang war die Tat’, included in that book (ibid.: 33–54). It concerns the question of what it is to assert something and how that is connected with methods of finding out.

Winch goes for a walk with you. Suddenly he points to an ordinary-looking building and says, ‘That house is made of papier-mâché’ (Winch, 1987: 40–1). He does not provide anything that you would normally accept as a reason. You go together and knock at the walls, which are completely solid, but nothing can convince him. In his essay, Winch raises the question of whether, in that situation, he would be making a recognisable assertion at all. It is not only that you wouldn't understand why he utters that sentence, but that its very status as assertion is blurred out. The intelligibility of expressing the belief that something is made of papier-mâché depends on an argumentative context where such claims are made and disputed or tested (ibid.: 42). The question is whether, in the situation Winch has described, one could make sense of describing the protagonist as claiming or believing anything at all.

One might perhaps reply that, by going to the house with Winch and knocking at the walls, it appears that you, his listener, have in fact gone along with the idea that Winch has made an assertion. Why else would you be checking how the walls sound? In response, however, Winch suggests it is more plausible in the situation to say you go with him to the house in order to test him. You will be trying to find out whether he really is making an assertion and what, if anything, that assertion might be.

A usual philosophical (‘realist’) reaction at this juncture is to respond as follows. Regardless of why Winch utters the sentence ‘That house is made of papier-mâché’, and regardless of how, if at all, its truth is tested – independently of all that, the content of the sentence ‘in itself’ is not in dispute. Its content corresponds to a specific thought (propositional content), which we may try to match with a certain part of reality, that is, the house. But Winch's claim is that, in the situation he has described, you would not be able to match the utterance confidently with anything at all. His utterance does not (at least yet, with the information that was given) amount to anything one would recognise as a claim and so we would not know what to make of it.

Winch argues that the realist is ‘obsessed with a certain picture’, ‘a mythological account of p's truth-conditions’. This mythology, roughly, presupposes an ‘ontological realm of facts’, with which the proposition p is compared (Winch, 1987: 39). Winch's point is that the real question here should be: What sense can we make of such comparisons in different circumstances? In other words, how do we actually go about it when we make comparisons between a proposition and the facts? Answering the question requires us to discuss criteria of verification, not as an afterthought once the truth claim itself has been made and understood, but as a way of spelling out what the truth claim amounts to. ¹⁷

In other words, the idea of ‘claims about reality’ is context-sensitive. Not only is our thinking influenced by context, but the very intelligibility of raising questions about what we ‘know’, ‘believe’, ‘assert’, or ‘must conclude’ presupposes an idea of the ways in which the propositions are ‘about’ reality. Philosophical realism assumes that that question has already been solved, whereas Winch saw it as one of the chief questions facing the philosopher (Winch, 1987: 195).

Conclusion: Reading Winch and Collingwood together

In making his ‘pincer movement’, Winch argued that understanding meaning in human life is a concern equally crucial to philosophy and to the interpretive human sciences. In their treatments of logic and metaphysical realism, Winch and Collingwood were fleshing out their idea that questions concerning meaningful behaviour also tie back to the question what philosophy is about.

By connecting Winch's work with Collingwood's ‘logic of question and answer’, we also see more clearly that Winch's concern did not lie with ontology. When Winch wrote, ‘What is real and unreal shows itself in the sense that language has’ (1972: 12; emphasis in original), he was not advancing the idea that ‘language’ comes before ‘experience’, which has a been major issue in structuralist and post-structuralist philosophy of language. He was voicing the point he also made in ISS about the ‘force’ of reality claims: the meaning of a given reality claim is something we must grasp, so to speak, from the inside of the relevant question-and-answer complexes. Conversely, if we were to read Winch as proposing an ontology, he would naturally come across as a relativist and a linguistic idealist. He would indeed have been a relativist if he had proposed that a specific set of logical rules, embodied in language, exists for each individual culture – rules that regulate all possible thinking, enclosing us in set views on what is real and unreal. ¹⁸ What he did say, instead, was that not only the force of a given assertion, but also the rules that hold for its relations with other assertions, need to be considered in the concrete give-and-take of human communication.

In sum, there is a deep connection between two main issues in both Collingwood's and Winch's writings: on the one hand, the need for ‘internal’ understanding of how human beings relate to reality, and on the other hand, their critique of the idea of logic as a self-sufficient system, external to historically embedded forms of life. This clarifies the general perspective that informed the approach to social explanation proposed in ISS. Winch did not advance a claim about the nature of human behaviour (determined by reasons rather than causes) but a claim about the kind of epistemic interest that guides our interest in action. The reverse side of this is his point about philosophical analysis: conceptual relations – which belong to the domain of philosophy – are intelligible only in the context of particular instances of social life.