Monthly Archives: June 2013

Descartes’ Cogito and Hintikka



It may be surprising for general readers to hear that Descartes’ famous “cogito” argument, which runs, “I think, therefore I am”, has been the subject of controversy among academics for the past few decades.  “What is there to dispute?”, one might wonder.  “Certainly”, it might be said, “Descartes must be right in saying that, if I am thinking, I must exist.  The one seems to be sufficient evidence for the other”.

Against this line of thought has arisen the following joke which seems to have proliferated in intro philosophy courses, and runs something like this:

Descartes walks into a bar.  The waiter asks, “Would you like a glass of Burgundy?”  Descartes answers, “I think not”, and vanishes in a puff of logical dust.

The line of thinking here is that if “Cogito, ergo sum” entails that whenever Descartes is thinking he should exist, it should also entail that if Descartes is not thinking, he does not exist.

The latter inference conceals a fallacy.  To see this clearly, consider what happens when the same type of reasoning is applied to the example red–>color.  Suppose the argument is that whenever I have red I can infer that I have a color. According to the argument pattern in the joke, it should also be the case that whenever I do not have red I do not have a color.  But this is obviously a mistake.  Moreover, it is an example of an invalid argument form (P->Q, ~P/~Q) which is an involves the fallacy of denying the antecedent.

Here are the steps, for convenience:

Consider this-

(1) Red –> Color

(2) ~Red

(3) ~Color  [Fallacy of denying the antecedent]

And now compare the same pattern to the one found in the Descartes joke-

(1) I think –> I exist

(2) I am not thinking

(3) Therefore, I do not exist [again, clearly a fallacy]

By contrast, note that when the argument is put into the form of modus tollens so that it runs, “I think -> I exist, I do not exist/I am not thinking” it comes out as a perfectly reasonable inference.  However, the inference, “if I am not thinking, then I do not exist” cannot be logically derived from  “I think, therefore I am”, except by introducing a fallacy.



We have an example above of an argument in which the cogito is treated as an inference.  At first glance it seems obvious that it should be.  This assumption has been challenged, however.  In a 1962 article titled, “Cogito, ergo sum: Inference or performance?” Jaakko Hintikka argued that the argument at least need not be considered as an inference and may be interpreted as a performance.  For example, Hintikka proposed the following reformulations of the cogito to bring out its performative aspect:

  1. “I am in that I think.”
  2.  “By thinking I perceive my existence.”

By claiming that the cogito can be treated as a performance, Hintikka is aligning it with a category of other statements like “I pronounce you man and wife” or “Let there be light”.  In each case, the sentence announces the action being performed.  In the case of the cogito, or so Hintikka would have it, in the performance of the action announced in 1 & 2, one will become aware that one exists.

When viewed as a performance, the argument should not be seen (so Austin stated) as having a truth value (i.e. being true or false) but should be evaluated in term of whether it is successful or not successful.  In other words, the success of 1 or 2 depends simply upon whether Descartes (or anyone else) is actually thinking.

Hintikka’s strategy becomes evident here: by recasting the cogito as a performance, the question of its cogency becomes less a matter of searching out the linkage between thinking and existing and more a question of simply performing a certain activity.


 Hintikka’s distinction between treating the cogito as either an inference or a performance generates a false dilemma.  I would like to argue that there is a way in which, by treating “I think, therefore I am” as an entailment the cogito can be viewed as both an inference and a performance without inconsistency.

[section explaining Hintikka’s motivation to be inserted here]

But how is this possible?  It was stated above that inferences have truth conditions while performatives are either successful or unsuccessful.  How then can they wind up in the same category?  The answer is that the cogito ought to be viewed as a statement that is true or false when it is uttered by someone.  This brings in both the performative and inferential aspects as follows: the statement as an inference is either true or false, but it is such that its truth or falsity depends upon whether the act of uttering it is actually (successfully) carried out.  Whenever this is done, whenever someone utters “I think, therefore I am”, the statement automatically becomes true if the person uttering it is using the statement in a self-referential way.  The fact that no one can deny the cogito, when uttered self referentially, makes it appear as though it were a statement (or proposition) and has independent standing as a true sentence; but, in fact, it is true whenever someone utters it.  The fact that the Latin word “cogito” may also be translated as “I am thinking” may be brought in as further support for this view.

This is not a trivial point in the context of assessing Hintikka’s argument.  First and foremost, it absolves scholars of the need to debate the relative merits of a performative vs. inferential view of the cogito, since they may be viewed, as explained above as working together synchonically.  I would argue that the cogito is an example of a case where the two mutually exclusive categories may overlap and actually reinforce one another.  By viewing the statement as an utterance we do not cancel out it inferential aspect.  Indeed, it is difficult to see how that might be possible with the word “ergo” present in the very formulation of the cogito.  Hintikka can only call the cogito a performance by eliminating the word; but arguably it is present even in his reformulations (see 1 and 2 above).  Certainly, one can reason, however banal it might now seem, that if one is thinking, one, in fact, must exist; but, moreover, it seems that Descartes must have had the discovery of the linkage between the two concepts in mind when he made the cogito the starting point of his attempt at constructing a purely deductive philosophical system.




Filed under Early modern philosophy corner