Does a Thought Exist?
One way of interpreting Descartes’s cogito argument is as an attempt to prove the existence of a thought. Begin with Descartes’s hyper-sceptical assumption, that everything you believe is false. Now formulate the thought ‘this thought exists’. If that is false, then it exists. And then it is true. So it must be true. So it must exist.
Descartes is often taken to have proven first the existence of himself, whereas the argument above works (if it does) to prove the existence of a thought. But Descartes believed that he was his thought; between himself as a thinking substance and the thoughts had by that thinking substance there was, so the Principles attest, only a distinction of reason.
When the radical Cartesian Arnold Geulincx discusses Descartes’s ‘hyberbolic’ doubt he refers to a structurally similar move, classic in the tradition of medieval obligationes (see Catarina Dutilh Novaes and Sara Uckelman on these), known as dilemma:
Suppositio haec, qua sic omnia falsa esse supposuimus, facit etiam ad clarissime demonstrandum primam veritatem. Praeclarius enim demonstrari non potest propositio aliqua, quam per dilemma, in quo ex falsitate propositionis demonstrandae infertur per necessariam consequentiam veritas ejusdem. Propositioni enim, quae sic demonstrata est, falsum subesse non potest, adeoque necessario vera est: nam sive vera sit, vera est, sive etiam falsa sit, vera est. V. g. si dicam: aliqua propositio vera est, dico adversario: Pone hoc esse verum vel falsum, perinde est, nam nihilominus demonstrabo esse verum. Si enim verum sit, aliquam propositionem esse veram, verum est. Si vero falsum sit, aliquam propositionem esse veram, ergo saltem verum est, nullam propositionem esse veram; sic equidem aliqua propositio est vera. (Antverpiensis Opera Philosophica, 142)
I translate:
This supposition, by which we have supposed everything to be false, also demonstrates Primary Truth most clearly. A proposition can be demonstrated no more excellently than by dilemma, in which, from the falsity of the proposition to be demonstrated, its truth is inferred as a necessary consequence. For the proposition thus demonstrated cannot be false and so is necessarily true: for it is either true and thus true, or false and thus true! E.g., if I were to say ‘some proposition is true’, I say to my adversary: suppose this to be true or false — either way I will yet demonstrate it to be true. For if it is true, then ‘some proposition is true’ is true. Or if ‘some proposition is true’ is false, then at least this is true: ‘no proposition is true’, and so again some proposition is true.
This, however, doesn’t issue in an existence claim. I recently came across a translation by Martin Wilson that renders ‘aliqua propositio vera est’ as ‘some true proposition exists’. That would strengthen the affinity between Geulincx’s dilemma and Descartes’s cogito. Then they would both prove existence. But to me it seems an unmotivated translation. If Geulincx had meant that, he could have expressed it unambiguously, for instance by using ‘existere’ or ‘subsistere’ which he often does elsewhere.
The issue of translation throws up a philosophical question. A.N. Prior regarded as provable the theorem that if you think that there is something you are thinking, then there is something you are thinking. But, in the collection of his later papers on logic, posthumously published as Objects of Thought, he warned against taking this as committing to too much. To concede that there is something that I am thinking is to allow that the closure of ‘I am thinking ___’ by some sentence in the gap yields truth. But we need not take the sentence in the gap to be the name for some abstract object such as a proposition or a fact — or what I am here calling a thought.
In the cogito argument as interpreted here, ‘I am thinking ___’ would yield truth when closed by ‘the closure of “I am thinking ___” by some sentence yields truth’. And the latter will then be true. There is an odd appearance of recursivity here, but, Prior argues, nothing fatally paradoxical.
The cogito thus interpreted serves as no existence proof. There being something that I am thinking doesn’t prove the existence of any object of thought. Nor does it prove my existence as the subject of thought. Certainly if I am thinking something then I exist. But that I am thinking something is the premise of the argument, not the conclusion, so here we have an assumption of my existence rather than a proof.
To be a nominalist about objects of thought, however, one must be a realist about something. The truth of ‘I am thinking something’ must depend on the being of something — presumably of me as thinking. And this could be a material object, if a person is a sort of thinking animal and an animal is a sort of material object. This is perhaps what Prior wanted to stress, so that the cogito could be safe for materialism.
But of course it isn’t what Descartes wanted. The purpose of the cogito isn’t just to show that something exists but also that thought is logically and thus metaphysically independent of everything else. It can bubble up from the whirlpool of doubt even while everything else remains sunk.
The way to ensure this, I think, is to cut the proof off from any reference to a thinker, who might be a material thing. The identification of thought with thinker is crucial and it should be read, as Joan Robinson might put it, ‘right-handed’ — once you have thought you have the thinker ipso facto.
What I mean is: treat the cogito as akin to Geulincx’s dilemma. ‘This thought exists’ must be true, since it would be so even on the assumption of its falsity. This is for reasons entirely independent of any revealing the existence of a thinker distinct from the thought. All the same, it is a thought being considered — that is vital to the argument. Otherwise you could say that ‘this thought exists’ is neither true nor false, because there is just nothing to be true or false. But you can’t say this, because it is patent that something is under consideration. It is here that Descartes’s ‘proof’ involves an indispensable intellectual intuition and no mere syllogism or Euclidean proof.
There must, then, be at least a thought, so long as it is considered. And this is possible even if there is nothing material or ‘psychological’ to consider it. Elizabeth Anscombe once proposed that it isn’t logically impossible for the snow to fall in well-formed propositions; likewise if there were no material objects and even no concrete objects, it is not logically impossible that some company of abstracta could somehow spell out a thought. Then the thought would have to be there to be expressed; perhaps it could even exist as ‘thought thinking itself’ (Aristotle, Metaphysics, 1072b).
The ability of the thought to be is to that extent metaphysically independent of anything else. It has, in Descartes’s terms, substantiality — existence per se and not in, e.g., the brain of an animal. I don’t think that Descartes treats thought in the Fregean way, as a purely abstract object. But I do mean to suggest that he doesn’t quite take it as a concrete object in our sense either. Geulincx seems to be onto something in hinting at an affinity between the cogito and the medieval dilemma.
