Where our notion of ‘causation’ comes from



This paper is a collection of quotations, about half from Collingwood and half from various other philosophers (Reid, Macmurray, Gasking, Austin and Strawson). Their common theme is the idea that the concept of causation, of events ‘causing’ other events, thought by some philosophers to be the concept that natural science is founded on, is actually an anthropomorphic metaphor derived from certain features of personal action. The earliest clear opinion that our notions of ‘cause’ (and ‘event’) derive from the concept of personal action, is that of Reid (1788).

It is very probable, that the very conception or idea of active power, and of efficient causes, is derived from our voluntary exertions in producing effects; and that, if we were not conscious of such exertions, we should have no conception at all of a cause, or of active power, and consequently no conviction of the necessity of a cause of every change which we observe in nature (Reid 1788, p.278).

When we turn our attention to external objects, and begin to exercise our rational faculties about them, we find, that there are some motions and changes in them, which we have power to produce, and that they have many which must have some other cause. Either the objects must have life and active power, as we have, or they must be moved or changed by something that has life and active power, as external objects are moved by us...

When a few of superior intellectual abilities find leisure for speculation, they begin to philosophise, and soon discover, that many of those objects which, at first, they believed to be intelligent and active, are really lifeless and passive…As philosophy advances, life and activity in natural objects retires, and leaves them dead and inactive. Instead of moving voluntarily, we find them to be moved necessarily; instead of acting, we find them to be acted upon; and nature appears as one great machine, where one wheel is turned by another, that by a third; and how far this necessary succession may reach, the Philosopher does not know...

It must have been by the observation and reasoning of the speculative few, that those objects were discovered to be inanimate and inactive, to which the many ascribed life and activity. But while the few are convinced of this, they must speak the language of the many in order to be understood. So we see, that when the Ptolemaic system of astronomy, which agrees with vulgar prejudice and vulgar language, has been universally rejected by Philosophers, they continue to use the phraseology that is grounded upon it, not only in speaking to the vulgar, but in speaking to one another. They say, The sun rises and sets, and moves annually through all the signs of the zodiac, while they believe that he never leaves his place.

In like manner, those active verbs and participles, which were applied to the inanimate objects of nature, when they were believed to be really active, continue to be applied to them after they are discovered to be passive. (Ibid., pp. 281-284.)

Macmurray (1938) thinks the same way. As well, he draws implications relating to the possibility of a causal explanation of personal actions.

The distinction between acts and events gives rise to the effort to account for events. If an observed change is recognised as an actum, it is thereby accounted for, and no further question arises. But if it is recognised as an event, it is not accounted for. The recognition amounts to the negative judgement, ‘This change is not the deed of an agent’. An alternative source of the change is then required. In modern thought the name for such an alternative source of change is the term “cause”. The idea of a cause is the idea of a source of change which is not an agent, and does not involve intention.

The recognition of the distinction between act and event thus has a direct bearing on the problem of causality. In the first place it shows that the proposition ‘Every event has a cause’ is perfectly compatible with the proposition, ‘No action has a cause’. Indeed, if ‘no action is an event’ and ‘no event is an action’, and if the term “cause” means a source of change which is not actio, then the two propositions imply one another. And it follows from this that there is a definite contradiction in talking of the cause of an action. To do so is simply to imply that it is not an action.

...the idea of an event having a cause is an attempt to think an event on the analogy of an act, while at the same time denying that it can be referred to an agent. The cause is that which is responsible for the production of the event. It is also something which is not an agent, and therefore cannot be responsible for anything, or produce anything. Thus the cause is at once an agent and not an agent. As a result, whatever we assign as the cause appears itself to be an effect which demands a cause to account for it. The negation of agency is not a pure invention to account for the idea of causality. It is what happens when I accept as an accident what I had first taken to be a deliberate act. In that case I simply agree that there was no actio involved, and that therefore the agent is not responsible for the change under consideration. If I am not satisfied to leave the matter there, I shall proceed to ask, in effect, ‘If actio is not the source of the change, what is?’ I am then committed to the effort to discover something which has the same functional relation to the observed change as actio would have had if it had been an actum, but which is not actio. I look for something which functions as if it were an agent, but is not an agent, which behaves in a fashion analogous to acting, but without acting. Yet if we could find anything of the kind, it would clearly be an agent, and the change would be his act.

The search for causes, then, depends upon a preoccupation with action, even if it is only half-conscious. But the effort to establish a pure science is an effort to exclude the reference to action and the agent and to describe the world merely as Object for a Subject. This involves the exclusion of all references to action, as well as the suppression of all interest in action. The search for causes must therefore disappear, since that involves both a reference to and an interest in action. All changes then appear as events, even if in fact they are acta and accepted as such. (Macmurray 1938, p.81-83.)

Collingwood (1940) distinguishes three senses of the verb cause. In sense I, what is caused is an action of a person, and to ‘cause’ the agent to perform that action is to give him or her a motive for doing so. Collingwood suggests that for causing in this sense we might equally substitute making, inducing, persuading, urging, forcing, and compelling. He gives the example of a 1936 headline ‘Mr Baldwin’s speech causes adjournment of House’, and comments as follows.

This does not mean that Mr Baldwin’s speech compelled the Speaker to adjourn the House whether or no that event conformed with his own ideas and intentions; it meant that on hearing Mr Baldwin’s speech the Speaker freely made up his mind to adjourn. In the same sense we say that a solicitor’s letter causes a man to pay a debt or that bad weather causes him to return from an expedition. (Collingwood 1940, p.290.)

Collingwood cites the OED in claiming that the English use is traceable back to the Middle Ages. And he says an identical sense was well-established in Latin and Greek.

Sense II of cause can also be traced well back — through Middle English back through familiar Latin uses of causa, to Greek of the fifth century BC. In this sense, both what is caused and what does the causing are natural phenomena (rather than actions), but with the proviso that what does the causing, the cause II, is at least in principle subject to human control. This second sense of cause is always to this extent related back to human practical activity. It is at home in what Collingwood calls ‘practical’ (vs applied or theoretical) science — disciplines such as medicine in which people develop abilities to intervene in and control otherwise natural events. This species of cause-effect terminology would perhaps be more accurately rendered in means-to-ends terminology. Collingwood provides the following definition of ‘cause’ in sense II: “A cause is an event or state of things which it is in our power to produce or prevent, and by producing or preventing which we can produce or prevent that whose cause it is said to be” (ibid, pp.296-7, italics in original). He further explains as follows.

This usage, representing sense II of the word cause, can be recognised by two criteria: the thing described as a cause is always conceived as something in the world of nature or physical world, and it is always something conceived as capable of being produced or prevented by human agency. Here are some examples. The cause of malaria is the bite of a mosquito; the cause of a ship’s sinking is her being overloaded; the cause of books going mouldy is their being in a damp room; the cause of a man’s sweating is a dose of aspirin; the cause of a furnace going out in the night is that the draught-door was insufficiently open; the cause of seedlings dying is that nobody watered them. (Ibid, p.299.)

Causes in the second sense are never by themselves sufficient causes, but are merely contributive, albeit perhaps crucially so. The cause in question is just one of a full complement of factors required to necessitate the final outcome. Collingwood qualifies causes II as causes relative ‘to a given agent’. Thus, “..for any given person the cause in sense II of a given thing is that one of its conditions which he is able to produce or prevent” (ibid, p.304, italics in original). For example, for a car driver, the cause of a skid was cornering too fast; for the road-designer, insufficient camber on the road; for the car manufacturer, an inadequate lateral suspension mechanism or a too-high centre of gravity; for the tyre-maker, poor tread pattern, etc. It follows that, for a person with no power to affect any of the conditions contributing to an event, the event has no cause, in sense II, at all. Collingwood embraces this inference.

In sense II of the word cause, only a person who is concerned with producing or preventing a certain kind of event can form an opinion about its cause. For a mere spectator there are no causes. When Hume tried to explain how the mere act of spectation could in time generate the idea of a cause, where cause meant the cause of empirical science, that is, the cause in sense II, he was trying to explain how something happens which in fact does not happen. (Ibid, p.307.)

The third sense of cause Collingwood distinguishes is the sense in which we think of scientific theories as explaining the causes of natural phenomena. He introduces it as follows.

Sense III refers to a type of case in which an attempt is made to consider natural events not practically, as things to be produced or prevented by human agency, but theoretically, as things that happen independently of human will but not independently of each other: causation being the name by which this dependence is designated. This is the sense the word has traditionally borne in physics and chemistry and , in general, the theoretical sciences of nature. (Ibid, p.287.)

What cause III apparently lacks in social relevance it makes up for in the intimacy of the cause-effect association. There is no ‘relativity’ such as there is with causes II, nor sets of contributing causes, nor effects exerted across a spatial or temporal distance. Causes III and their effects are very closely conjoined.

Here that which is ‘caused’ is an event or state of things, and its ‘cause’ is another event or state of things standing to it in a one-one relation of causal priority: i.e., a relation of such a kind that (a) if the cause happens or exists the effect also must happen or exist even if no further conditions are fulfilled, (b) the effect cannot happen or exist unless the cause happens or exists, (c) in some sense which remains to be defined, the cause is prior to the effect; for without such priority there would be no telling which is which. If C and E were connected merely by a one-one relation such as is described in the sentences (a) and (b) above, there would be no reason why C would be the cause of E, and E the effect of C, rather than vice versa. But whether causal priority is temporal priority, or priority of some other kind, is another question. (Ibid, pp.285-6.)

Collingwood does not see anything problematic about senses I and II. They are relatively straightforward, easy to understand, and give rise to no perplexities. Sense III, however, he sees as inherently contradictory. He argues (ibid, pp.314-315) that if there is a ‘tight’ one-one relation between C and E such that C is in itself necessary and sufficient for E, then there can be no spatial or temporal distance between C and E, and thus C cannot be prior to E (as required in the original definition of sense III). Collingwood claims that Hume’s notion of causation is based on sense II of cause and is a ‘loose’ one, according to which causes can precede their effects, and need not in themselves be necessary and sufficient to bring about those effects. Collingwood says that Kant’s definition of causation — at least the definition offered in the Critique of Pure Reason as a basis for natural science — requires both that every event have a cause and (from Hume) that causes are temporally prior to the events they cause. Collingwood argues (ibid, pp.329 and 331) that ‘every event has a cause’ logically requires the one-one relation of sense III, and thus is not consupponible with the temporal priority condition.

Another difficulty Collingwood sees with sense III is that people cannot help but read some notion of causal ‘necessity’ or ‘compulsion’ into it. The sense of compulsion must attach to the (unconsupponible) ‘C temporally precedes E’ feature of sense III — the feature which is carried over from sense II, and Hume — because, with only a ‘tight’ connection between C and E, there would be no ‘room’ for any compulsion to be exerted. Collingwood does not think this necessity can be explained either in the traditional rationalist way, as a kind of logical necessity, or in the traditional empiricist way, as a function of induction, or of the ‘momentum’ produced by constant conjunction.

Most people think that when we use the word causation in sense III we mean to express by it something different from logical implication, and something more than uniformity of conjunction, whether observed only, or observed in the past and also expected in the future; and this ‘something different’ and ‘something more’ is in the nature of compulsion. On the historical issue of what has actually been meant when [these] words have actually been used, this is correct. (Ibid, p.321.)

The obvious source of the necessity or compulsion which we cannot help seeing in cause III is, Collingwood thinks, the persistence in sense III of the earlier sense II. And the connotation of compulsion in sense II is in turn traceable back to sense I.

The cause-effect terminology conveys an idea not only of one thing’s leading to another but of one thing’s forcing another to happen or exist; an idea of power or compulsion or constraint. From what impression, as Hume asks, is this idea derived? I answer, from impressions received in our social life, in the practical relations of man to man; specifically, from the impression of causing (in sense I) some other man to do something when, by argument or command or threat or the like, we place him in a situation in which he can only carry out his intentions by doing that thing; and conversely, from the impression of being caused to do something. (Ibid, p.309.)

Collingwood attempts to specify the nature of the derivation of sense II from sense I. He talks of animistic theories, literal beliefs, and (ibid, p.311) of ‘linguistic survivals’ and ‘thought-forms not being as dead as they are supposed to be’, but he denies that metaphor has a hand in the derivation.

In Plato’s Timaeus, and in the Renaissance Platonists whose part in the formation of modern natural science was so decisive, the constant use of language with animistic implications is neither an accident nor a metaphor; these expressions are meant to be taken literally and to imply what they seem to imply, namely that the way in which men use what we nowadays call inorganic nature as means to our ends is not in principle different from the way in which we use other men. We use other men by assuming them to be free agents with wills of their own, and influencing them in such a way that they shall decide to do what is in conformity with their plans. This is ‘causing’ them to act in sense I of the word cause. (Ibid, pp.309-10.)

In my own opinion, the fact that the Greeks took their animism literally shows not that the linguistic vehicles of their animism were literal referring expressions but rather that, if these expressions were metaphors, they were unacknowledged and were, naively, taken literally. Whether or not they were metaphors can be decided on independent rhetorical grounds. If we have an expression which overcomes manifest gross differences between the general referent and the apparent referent to achieve a subtle and difficult reference to some feature of the general referent, then we should infer that metaphor is being employed. No matter who does it, and with whatever naive ignorance of or sophisticated regard to the rhetorical niceties, the applying of human action verbs to natural phenomena can only be metaphorical.

However, later on, Collingwood does confront the problem of explaining the compulsion connotation — why we cannot but think of causes ‘producing’ or ‘necessitating’ their effects — as a problem of the use of language as much as of ‘the persistence of thought-forms’. He says that,

When similar language is used of senses I and II we know what it means. In sense I it means that X affords somebody a motive for doing Y; in sense II, that X is somebody’s means of bringing Y about. But what (since it cannot [literally] mean either of these) does it mean in sense III? (Ibid, pp.315-316.)

From this point on, Collingwood’s opposition to explaining the derivation of one ‘thought-form’ from another in terms of figurative — metaphorical or analogical — extension of the earlier term seems to soften. He is apparently content in the following extracts to have at least the ‘compulsion’ component in sense III derived from senses I and II by anthropomorphic metaphor or analogy.

The idea of compulsion, as applied to events in nature, is derived from our experience of occasions on which we have compelled others to act in certain ways by placing them in situations (or calling their attention to the fact that they are in situations) of such a kind that only by so acting can they realise the intentions we know or rightly assume them to entertain: and conversely, occasions in which we have ourselves been thus compelled. Compulsion is an idea derived from our social experience, and applied in what is called a ‘metaphorical’ way not only to our relations with things in nature (sense II of the word cause) but also to the relations which these things have among themselves (sense III). Causal propositions in sense III are descriptions of relations between natural events in anthropomorphic terms. (Ibid, p.322.)

Apparently, not even Newton was immune to the blandishments of anthropomorphic explanations.

Here [in the Scholium appended to his Definitions] and throughout his treatment of the subject it is perfectly clear that for him [Newton] the idea of causation is the idea of force, compulsion, constraint, exercised by something powerful over another thing which if not under this constraint would behave differently...

Taken au pied de la lettre, Newton is implying that a billiard-ball struck by another and set in motion would have liked to be left in peace; it is reluctant to move, and this reluctance, which is called inertia, has to be overcome by an effort on the part of the ball that strikes it. This effort costs the striker something, namely part of his own momentum, which it pays over to the sluggard ball as an inducement to move. I am not suggesting that this reduction of physics to social psychology is the doctrine Newton set out to teach; all I can say is that he expounded it, no doubt as a metaphor beneath which the truths of physics are concealed. (Ibid, p.326.)

The basic problem facing scientists is restated.

The anthropomorphic dilemma. The question here is whether the natural scientist in his detailed study of the world of nature presupposes that this world is animated by something like human mind, or at any rate human psyche, or whether he makes no such presupposition. It is not a pseudo-metaphysical question. It is not a question as to whether the world of nature is thus animated or not. It is a question as to the presuppositions which in fact underlie the natural scientist’s approach to that world. The alternatives are:

(a). The natural scientist is trying to construct a science of nature in terms of analogies drawn from the conscious life of man. It is only throught such analogies that nature becomes intelligible to man; a science of nature which renounced their use would accordingly be no science at all. When Darwin in the Origin of Species announces “the highly important fact that an organ constructed for one purpose may be converted into one for a widely different purpose” (Ch.VI), his use of frankly teleological language need bring no blush to the cheek of his disciples. Thus described, the facts of animal anatomy become intelligible. Described without appeal to the human activities of constructing and adapting, means and ends, they would be unintelligible.

(b). The natural scientist, in so far as he uses these analogies, is obscuring his own thought by saying what he does not mean. A well-devised vocabulary for use in natural science would avoid them. The natural scientist does not really believe that nature devises and adapts, invents means to bring about her ends; he thinks that this is a purely human type of behaviour, and that his business is to describe everything he can in terms of physical and chemical processes in which it has no place.

The orthodox view of natural science during the Kantian period [from Kant to Einstein] was (b). But once more the issue was not clearly defined. The natural science of the period regarded itself as a non-anthropomorphic natural science, and in attacking anthropomorphism pinned its faith to causation in sense III as its favourite weapon. It failed to realise that within this sense of the word there lay concealed an element of anthropomorphism, concealed because to discover it would have required the exercise of mataphysical analysis, and metaphysics was barred: and that the so-called ‘materialism’ which was the favourite metaphysical doctrine of these anti-metaphysicians was in consequence only in name a repudiation of anthropomorphism; really it was anthropomorphic at the core. (Ibid, pp.334-336.)

Collingwood believes that, most recently, “..physicists have been escaping from the damnosa haereditas of the Kantian confusion by the heroic measure of reconstructing their own science in such a way that the idea of causation no longer figures in it at all” (ibid). Doing without the idea of causation may be an option for physics. But biological sciences, such as physiology, seem to rely on causal explanation — if only as a medium into which to translate their more obviously figurative ‘functional’ and ‘mechanistic’ explanations.

Gasking (1955) is quite specific about the relation between personal action concepts and the concepts of events and causes. He says that causation talk is elliptical for talk about one special type of personal action — namely, technical, productive activity. Thus Gasking’s notion of causality is identical to that of Collingwood’s ‘cause’ in sense II. Gasking sums up his account as follows.

I have made two points:

First: that one says ‘A causes B’ in cases where one could produce an event or state of the A sort as a means to producing one of the B sort. I have, that is, explained the ‘cause-effect’ relation in terms of the ‘producing-by-means-of ’ relation.

Second: I have tried to give a general account of the producing-by-means-of relation itself: what it is to produce B by producing A. We learn by experience that whenever in certain conditions we manipulate object in a certain way a certain change, A, occurs. Performing this manipulation is then called ‘producing A’. We learn also that in certain special cases, or when certain additional conditions are also present, the manipulation in question also results in another sort of change, B. In these cases the manipulation is also called ‘producing B’, and, since it is in general the manipulation of producing A, in this case it is called ‘producing B by producing A’. For example, one makes iron glow by heating it. (Gasking 1955, pp.485-6.)

Gasking goes on to claim that the notion of ‘cause’ he is thus elucidating is “the fundamental or primitive one” (ibid.), and that “A statement about causes in the sense here outlined comes very near to being a recipe for producing or preventing certain effects” (ibid.) — that is, statements about causes are shorthand for statements, or instructions, concerning (potential) practical actions by people.

The following brief passage from Austin (1961) is relatively well known.

‘Causing’, I suppose, was a notion taken from man’s own experience of doing simple actions, and by primitive man every event was construed in terms of this model: every event has a cause, that is, every event is an action done by somebody — if not by a man, then by a quasi-man, a spirit. When, later, events which are not actions are realised to be such, we still say that they must be ‘caused’, and the word snares us: we are struggling to ascribe to it a new, unanthropomorphic meaning, yet constantly, in searching for its analysis, we unearth and incorporate the lineaments of the ancient model. (Austin 1961, pp.150-151.)

The theme also has the imprimatur of Strawson (1986). Strawson returns to Reid’s idea that we can know causation only as we know our own actions. Reid says that “From the consciousness of our own activity, seems to be derived, not only the clearest, but the only conception we can form of activity, or the exertion of active power” (Reid 1788, p.36). In Strawson’s terms, we do this by ‘imaginatively identifying with’ either ‘agent’ or ‘patient’ of the action in question.

Hume tracked down to a subjective source what he took to be the distinctive feature of our confused conception of causality as a natural relation. That distinctive feature he usually referred to as the idea of a necessary connection. But he allowed that it bore other names of which he said that they were virtually or, as he put it, ‘nearly’ synonymous. His list of nearly synonymous terms includes efficacy, agency, power, force, energy, necessity, connexion and productive quality [Hume 1739 I. III. xiv.]; to which he might have added compulsion without straying far outside the bounds of his notion of near-synonymity. In tracking the idea down to its subjective source he of course followed, or claimed to follow, his leading principle: seek the impression from which the idea is derived... If we concentrate on the trio power, force and compulsion, and ask from what impression the idea discernible in them all is derived, the most obvious answer relates to the experience we have of exerting force on physical things or of having force exerted on us by physical things — including here the bodies of other people as physical things...

Of course, however, we do not limit the application of the idea of force to those mechanical transactions, those pushings or pullings, in which we ourselves, or our fellows, are engaged as agents or patients. We extend the idea to all such transactions. Is there, as Hume suggests in the footnote [An Enquiry Concerning Human Understanding, VII. ii, final footnote] referred to, an element of anthropomorphic projection in this extension? Perhaps so. In a great boulder rolling down the mountainside and flattening the wooden hut in its path we see an exemplary instance of force; and perhaps, in so seeing it, we are, in some barely coherent way, identifying with the hut (if we are one kind of person) or with the boulder (if we are another): putting ourselves imaginatively in the place of one or the other. (Strawson 1986, pp.122-3.)

In general, then, the search for causal theories is a search for modes of action and reaction which are not observable at the ordinary level (or not observable at all, but postulated or hypothesised) and which we find intelligible because we model them on, or think of them on analogy with, those various modes of action and reaction which experience presents to gross observation or which we are conscious of engaging in, or suffering, ourselves. (Ibid., p.125.)

— Derek Melser —


  • Austin, J.L. 1961. Philosophical Papers. Oxford: Oxford University Press.
  • Collingwood, R.G. 1940. An Essay on Metaphysics (a.k.a. Philosophical Papers Vol.II). Oxford: Oxford University Press.
  • Gasking, D. 1955. “Causation and Recipes”. Mind 64: pp.479-487.
  • Macmurray, J. 1938. “What is action?” Proceedings of the Aristotelian Society, supplementary volume 17: pp.69-85.
  • Strawson, P.F. 1986. “Causation and Explanation”, in Vermazen, B. and Hintikka, M.B., eds., Essays on Davidson — Actions and Events. Oxford: OUP.
  • Reid, T. 1788/1977. Essays on the Active Powers of Man. NewYork: Garland.