1.1 Ayer’s Verificationism
In this chapter, I want first to consider briefly the connection that some have seen as holding between R. G. Collingwood’s Essay on Metaphysics and A. J. Ayer’s Language, Truth and Logic.Footnote 1 Both T. M. Knox and Alan Donagan believed that “between 1936 and 1938 Collingwood radically changed his mind about the relation of philosophy to history.”Footnote 2 Donagan contends, further, that this break stemmed from Collingwood’s having read Ayer’s book, with the result that “he had come to endorse Ayer’s view that the propositions of traditional metaphysics are unverifiable.”Footnote 3 Recently, several important studies of Collingwood’s thought have claimed, going beyond Donagan, that Collingwood in Metaphysics endorsed Ayer’s verificationism, or at least yielded unduly to it. Here I have in mind papers by Guido Vanheeswijck and Michael Beaney.
Ayer’s verificationism makes three central claims: (1) there are only two kinds of meaningful statements; (2) such statements are either analytic statements – tautologies – or they are verifiable synthetic statements; (3) all synthetic statements must be empirically verifiable (at least in principle) or they are meaningless (and not propositions or statements at all). But Collingwood denied, discredited, or had grounds for denying all three of these claims.
1.2 The First Verificationist Claim
In his “Notes for an Essay on Logic,” in a passage that has not been discussed elsewhere, Collingwood challenged the first of these claims. He asks: “Query: what becomes of the lex exclusi medii [the law of excluded middle] on the log. pos. [the logical positivist] theory?”Footnote 4 In answering this query, Collingwood first suggests that where a non-analytic statement is not itself an empirically verifiable statement, it is not, on positivist grounds, a meaningful statement at all. It would be a piece of nonsense and so would its contradictory. It follows, then, Collingwood continues, that either a non-verifiable statement or its contradiction is a proposition – which is what the law of excluded middle would lead us to expect but which positivism would deny – or else the law of excluded middle, insofar as we conform to positivism on the point at issue, is in effect abrogated.Footnote 5
In sum, positivism and the law of excluded middle cannot both be held consistently. This conclusion, which Collingwood has forced on the logical positivists, would be a damaging one for them to have to draw. For the positivists’ first claim, that there are only two kinds of meaningful statements, itself presupposes the law of excluded middle; but the positivists’ third claim, that all synthetic statements are empirically verifiable (at least in principle) or they are meaningless, undercuts (in the way just shown) the very law they are committed to.
The argument here by Collingwood is a dialectical argument; it is concerned with the “consupponibility” of two of positivism’s central claims (that is, claims (1) and (3)). Of these two claims, the third – that all synthetic statements must be empirically verifiable (at least in principle) or they are meaningless (and not propositions or statements at all) – is the crucial one. We need, accordingly, to take up this claim directly and on its own, and to consider Collingwood’s response to it. I propose that we do so in two steps, by looking first at Collingwood’s idea of absolute presuppositions and then by seeing how Collingwood assesses that idea in the light of the central verificationist claim.
1.3 Absolute Presuppositions
In his Essay on Metaphysics, Collingwood put forward the view that every system of inquiry – that is, every science – has some foundation, which is from the standpoint of the practices in that inquiry logically ultimate. It is these background foundations that Collingwood referred to as absolute or fundamental presuppositions.
The things Collingwood called fundamental presuppositions of science were necessarily conceived by him to be meaningful. For if absolute presuppositions were not meaningful, then they could not be supposed (by anyone who thinks about them) or be presupposed (in some sort of logical relationship) by inferior propositions, or be reasoned to and assessed by metaphysicians in their attempts to determine the grounds of scientific inquiry and explanation.
But there is, Collingwood’s argument continued, no clear way in which presuppositions such as these could be determined to be true or false. It is difficult to see what fact of the matter or state of the world they could be referred to in order to determine their truth-value. Further, as absolute, they presuppose nothing in the universe of inquiry they govern, whereas all of our network of relevant knowledge claims, by which we do determine truth or falsity in that domain – for example, by going to the facts of the world – already presuppose them. So, clearly, we cannot take the way in which we could determine, to be true or false, a statement within the universe of discourse organized by a fundamental presupposition to be the same as the way in which we could then determine that presupposition itself to be true or false.
In Collingwood’s view, there simply is no way, no plausible method, whereby we could verify (that is, confirm or disconfirm) that fundamental principles of scientific inquiry are true.Footnote 6 For verification itself always rests on presuppositions and some presuppositions, then, must necessarily escape its reach. Indeed, if we were concerned to verify a basic presupposition, it would be presupposed in any relevant process of verification, including any process of verifying it.Footnote 7
To speak of verification here then is senseless. Empirical verification in such a case simply cannot do its job. It is out of the question.
1.4 Absolute Presuppositions and Positivism
It should be clear that Collingwood was not using the notion of verification here in the same way that the logical positivists of his day did. For them, it followed that such statements, when not empirically verifiable either in practice or in principle, would be meaningless; they would be statements that lacked literal significance altogether.
If, according to the positivist, someone said that all events have a cause, that person would in effect be asserting a well-confirmed (or probable) empirical hypothesis. And were that person or someone else to find an event that did not have a cause, then that observation statement – though it would not on its own falsify the empirical hypothesis in question – would have something of a disconfirming effect on it.Footnote 8 Enough such observations and the hypothesis that all events have a cause would cease to be well-confirmed. One could even imagine the case in which the hypothesis would be withdrawn, as discredited, after having repeatedly encountered such empirical failures.
But the actual assignment of truth values in given cases is not the principal issue with verificationism. Rather, the point is that the hypothesis about causes, if it is not true by definition, must be one that is subject to empirical tests.Footnote 9 And if people could conceive no way to verify this initial statement about causes (that is, could neither find nor imagine any way to empirically confirm or disconfirm the statement that all events have a cause), then it really was not a meaningful statement at all; it was a pseudo-statement, nothing but a meaningless string of sounds.Footnote 10
Collingwood, however, was neither looking for an empirical way to verify fundamental presuppositions, nor denying they were meaningful because he could not find one. He was doing something radically different from what was called for in the positivist program. Collingwood was looking at basic presuppositions systemically, taking account of the role they played within systems of inquiry, and arguing that, given this role, they could not be verified at all – certainly not in the way that the statements governed by these basic presuppositions could be verified.Footnote 11
Metaphysical principles – that is, absolute presuppositions – are distinct from historical statements about absolute presuppositions. The historical statements can be true or false; one can assert that a certain presupposition governed scientific inquiry at a given time, and this assertion can be true or false (see EM 54–55). But metaphysical principles, the presuppositions themselves – were they to be separated out from the complex of questions, propositions, and subordinate presuppositions they organize, were they to be formulated, and then asserted on their own – are not confirmable as true or false. The job of metaphysical principles is not to be true; their main function is to guide inquiry, and, in the favored case, to guide it successfully.
This is the consistent line Collingwood took, in his An Essay on Metaphysics and elsewhere. And the clearest evidence that he had not bought into logical positivism was that he affirmed throughout that basic presuppositions were both useful and meaningful in their role of guiding scientific inquiry, even though they were not and could not be verified, empirically or otherwise (EM 42–43).Footnote 12
Absolute presuppositions are capable in principle (logically capable) of having truth values. But as presuppositions (when functioning that way) they do not need to take such values, and if they were to be asserted simply on their own, they could not plausibly be given such values as true or false.
Let me be more precise. Absolute presuppositions could be assigned such values. One could assume as true a presupposition, as part of a proof or in a give-and-take argument. One could assert that someone (at another time, for example) believes a certain fundamental presupposition to be true. Or someone could think that such a presupposition was true, on the ground that it actually governed ordinary thinking or was actually accepted in the science of their day.Footnote 13 But what one could not do is determine or establish that they are true; one could not “prove” them, as Collingwood says in the Idea of History, to be true or false. Indeed, it becomes difficult to identify even a sense in which they could be true or false when they are disengaged from their moorings within a system of inquiry.
The account I am giving here is stronger than the claim (frequently made) that absolute presuppositions cannot be empirically confirmed as true or disconfirmed as false (a claim I agree with), and thus that they cannot be assigned truth values on that basis. Rather, the claim Collingwood makes is that there is no accredited way to establish that they are true.
The fact that Collingwood said that absolute presuppositions were not susceptible of being shown to be true or false does not indicate that he endorsed positivism. Rather, to have endorsed positivism, he would have to have made this claim about fundamental presuppositions within the confines of verificationism. He did not; he denied verificationism in believing that such presuppositions, while they could not be shown to be true (or false), empirically or otherwise, nonetheless had literal meaning. Rather than endorsing positivism, he was showing the way round it.Footnote 14 The place to look in order to find models or sympathetic examples of Collingwood’s idea that basic presuppositions could not be verified would be, then, not logical positivism, which he criticized and totally rejected, but rather the works of strong philosophical critics of that doctrine, like Wittgenstein or Quine.
1.5 The Notion that Collingwood Endorsed Positivism
Of course, we have not yet taken adequate account of the view that Collingwood in fact endorsed positivism, which is contrary to the view I have just expressed. Let us turn to that task now.
Both Michael Beaney and Guido Vanheeswijck have expressed some such view. Beaney has said both that Collingwood, in his Metaphysics, “adopts” logical positivist verificationismFootnote 15 and that he “endorses” it.Footnote 16 And Guido Vanheeswijck says that Collingwood, in his Metaphysics, made “exaggerated concessions” to the positivist attack.Footnote 17 In this section I will consider an argument, primarily by Michael Beaney, in support of the view that Collingwood endorsed the logical positivist doctrine of verificationism.
Near the beginning of his chapter on logical positivism (EM 162–63), Collingwood lays out a syllogism that he thinks captures Ayer’s main argument:
(1) Any proposition which cannot be verified by appeal to facts is a pseudo-proposition.
(2) Metaphysical propositions cannot be verified by appeal to observed facts.
(3) Therefore, metaphysical propositions are pseudo-propositions, and therefore nonsense.
Collingwood continues, then, by critiquing the “minor premise” (2). It contains a major blunder, he alleges; it mistakes a presupposition (for that is what metaphysical statements are, or are aimed at) for a proposition. Under the influence of this mistaken view, Ayer thinks that “logical efficacy, or the power of causing questions to arise, belongs exclusively to propositions, or things which are either true of false.” In fact, Ayer, as a positivist, “does not [even] possess the idea of supposing and a fortiori not the idea of an absolute presupposition” (EM 163).
Beaney’s charge here is that Collingwood has rejected the wrong premise. He has rejected premise (2); he should have rejected premise (1), but instead he accepts it (and this despite the fact that it is one of the central claims of positivism).Footnote 18
I am not sure that Collingwood has rejected premise (2). He thinks it confusingly and ineptly put. His objection to it is in part verbal, but it could be rephrased in a way that he could accept. Let me briefly fill in this claim. Collingwood objected to thinking of fundamental metaphysical principles as propositions. He insisted instead that such principles are correctly understood to be presuppositions. The job of such presuppositions is not to state facts but to guide scientific inquiry; as such, absolute presuppositions are not verifiable. We cannot determine them to be true or false empirically, or justify their use on those grounds.
Accordingly, we should take premise (2) to be saying that fundamental metaphysical presuppositions cannot be empirically verified (see EM 165). On this understanding (and interpreting the basic syllogism to accord with this understanding), premise (2) would say that “metaphysical propositions [understood as absolute presuppositions] cannot be verified by appeal to observed facts.”
In short, Collingwood expressed some misgivings about premise (2), as stated, but he did not reject it flat out. But even after the suggested Collingwoodian readjustment or gloss on premise (2) is taken on board, there is still something a bit off about the basic syllogism.
There is an important term (“nonsense”) that appears in the conclusion, even though it is not in either of the premises. This will not do; we do not yet have a valid syllogism. We can remedy this by recalling what Ayer meant by a pseudo-proposition (or a “pseudo-statement”) and by recalling what Collingwood took Ayer to mean by it: a pseudo-proposition was something – a synthetic (or non-analytic) statement – that looked like a proposition and that should be meaningful but was not; a pseudo-proposition had no literal meaning; it was nonsense.Footnote 19 So if we insert the phrase “and therefore nonsense” at the end of premise (1), we have the makings of a valid syllogism.
And this is exactly the point on which Collingwood’s argument finally settles:
[T]he doctrine of the “logical positivists” that metaphysical propositions are nonsensical will involve the bankruptcy of all thinking in which any use is made of absolute presuppositions; that is to say, the bankruptcy of all science. Any attack on metaphysics is an attack on the foundations of science; any attack on the foundations of science is an attack on science itself.
We know (from Section 1.4) that Collingwood regarded absolute presuppositions as both meaningful – as having literal significance – and as not being empirically verifiable, as not being determinably true or false by “appeal to observed fact.” So, he would have regarded the conclusion to the positivist syllogism – the conclusion that “metaphysical propositions are pseudo-propositions, and therefore nonsense” – to be false, and false precisely in the claim that absolute presuppositions are “therefore nonsense.”
Now we have in philosophy a template for valid arguments (called modus tollens) that goes like this: (i) if p, then q, (ii) not q, therefore (iii), not p. In other words, if one has an argument in which a conclusion follows from the premises, and that conclusion is deemed to be false, then the premises are, in whole or in part, false.
Collingwood would not have thought the second premise – “metaphysical propositions [understood as absolute presuppositions] cannot be verified by appeal to observed facts” – to be false, so he must have regarded the first premise, as amended above, to be the problematic one – problematic, that is, in view of the modus tollens argument.Footnote 20 The first premise so amended would read (1*), “Any proposition which cannot be verified by appeal to facts is a pseudo-proposition and therefore nonsense.” Collingwood had never explicitly accepted this premise (either in its original or in its amended form). Instead, he merely let it stand. Then, after introducing a corrected Collingwoodian understanding of premise (2) into the syllogism, he allowed the positivist conclusion to be drawn and then declared that conclusion to be false.
Given Collingwood’s rejection of the conclusion to the positivist syllogism, he had grounds for denying the first premise and for declaring it to be, in his view, false. Accordingly, we must reject Beaney’s contention that Collingwood accepted the first premise (for this he could not have done). Thus, Collingwood did not endorse logical positivist verificationism. Rather, he condemned it as an attack upon science, both natural and historical, and upon rational thinking. This is how the chapter on logical positivism ends – remember that the chapter is entitled “Suicide of Positivistic Metaphysics” – and is how the concluding chapter of the entire book ends.Footnote 21
1.6 Conclusion
There is a considerable gulf, then, between my view of what Collingwood thought about logical positivism and the view that Vanheeswijck and Beaney attribute to him. In my view, Collingwood denied logical positivism flat out; in their view he was seduced by it.
In this chapter, I have attempted to do two things. I have tried to lay out and assess the main points that Vanheeswijck and Beaney used to support their view, and I have attempted to develop what I think is a viable alternative to their views, one that takes account of Collingwood’s treatment of absolute presuppositions in An Essay on Metaphysics (in particular, on the vexed question of whether they can be proven or otherwise determined to be true or false), and at the same time avoids the conclusion that Collingwood had, mistakenly, bought into logical positivism in his discussion of absolute presuppositions.Footnote 22
It might be useful (and fair), as I end the chapter, to consider Ayer’s response to Collingwood’s criticisms. At the very end of Ayer’s introduction to the second edition Language, Truth and Logic, he addresses one feature of Collingwood’s critique of the argument in that book. Without mentioning Collingwood by name, Ayer says,
the reduction of philosophy to analysis need not be incompatible with the view that its function is to bring to light “the presuppositions of science.” For if there are such presuppositions, they can no doubt be shown to be logically involved in the applications of scientific method, or in the use of certain scientific terms.
He continues a few lines later, saying, “[I] now think it is incorrect to say that there are no philosophical propositions. For, whether they are true or false, the propositions that are expressed in such a book as this do fall into a special category … [P]hilosophical propositions, if they are true, are usually analytic.”Footnote 23
I think Ayer makes two important concessions to Collingwood. Ayer is open to the possibility that there are, as Collingwood claimed, absolute presuppositions of science and is open to the claim that there are genuinely philosophical theses (formulations, statements), of the sort Collingwood had in view with such presuppositions, and that these “fall into a special category.” However, if we stick just with Collingwood’s favored case, that there are fundamental presuppositions of science, and if we want to go beyond making merely historical statements about them, Ayer’s view is that the philosophical identification and correct formulation of any of these presuppositions is in effect a true analytic statement.
Collingwood, of course, did not regard these formulations as analytic statements in Ayer’s sense; rather, statements of presuppositions express something that is contentful and substantive but not provable to be true (or to be false). So, if we were to continue this dispute, we would have to design a new verificationist syllogism and have at it on that basis. But one thing we could be certain of is that Ayer and Collingwood (or those who adhere closely to their views) would continue to advance genuinely distinctive and differing philosophical positions.
1.7 A Final Word
Ayer concludes the chapter on “Language, Truth and Logic” in his autobiography with an amusing anecdote. Collingwood, Ayer says, took the book seriously enough to devote a part of his lectures to refuting it. Then Ayer adds: “He ended one such lecture by saying, ‘If I thought that Mr. Ayer was right, I would give up philosophy.’ When the audience arrived for the next lecture, they were startled to find that it had been cancelled. [But] the story ends lamely: he had been stricken with influenza.”Footnote 24
