Inference or performance: what Hintikka argued in 1962, what his critics say he argued, and why the field settled on a hybrid
The chat was set up to answer one question. Is Hintikka’s reading of the cogito as existential self-verification faithful to Descartes and philosophically better than the inferential and intuitional readings? Or does it fail, and if so, exactly how and according to whom? On that reading, “I do not exist” defeats itself in the act of being thought, so “I exist” verifies itself. The chat’s verdict: Hintikka’s insight survives and his exclusivity does not. The cogito is first-person, momentary and act-bound, as he says. But the Second Replies deny only that it is a syllogism, not that it is an inference (Frankfurt). Once made rigorous, the performance machinery either collapses into the inference “If I think, I exist” or leaves the link between thinking and existence unexplained (Wilson, Feldman, Dicker). The field has settled on an “intuitively grasped inference” (Kenny, Curley, Markie, Newman), because only that reading lets the cogito serve as the model clear and distinct perception from which the truth rule is drawn.
What this review adds. Reading the 1962 article itself changes the shape of the dispute. Hintikka did not claim exclusivity in 1962. He argued that Descartes’ cogito is ambiguous between two interpretations, an inferential one that generalises to any act of consciousness (volo, ergo sum) and a performatory one that does not, and that Descartes “does not distinguish” them (pp. 25, 31). Dicker’s first objection, that Descartes held any thought establishes existence, is Hintikka’s own interpretation (i). His second, that the performance reduces to an inference, is answered by the 1963 reply the chat quotes: there is “no incompatibility whatsoever.” The critics are right about the systematic-role problem. They are less right about what the 1962 paper claimed.
“Is Hintikka’s reconstruction exegetically faithful to Descartes and philosophically superior to inferential and intuitional readings — or does it fail, and if so, exactly how and according to whom?”
Asked: the reconstruction with article page references; its primary-text basis; a catalogue of criticisms that separates exegetical objections (faithful?) from philosophical ones (a good account regardless?); and the systematic-role question.
Refused: treating “performative” as self-explanatory. The brief required it to be defined against Austin, and asked whether Descartes’ concern with silent thought undercuts the analogy with public speech acts.
The chat’s report is the spine; Hintikka’s four texts are the object. Everything below is in the folder.
| Work | What it contributes here | Enters at |
|---|---|---|
| Chat report, “Hintikka’s Performative Cogito: A Critical Survey” (15 July 2026). The folder has two PDF copies and a fact-checked merge, Hintikka_Performative_Cogito_MERGED_factchecked.md. | The reconstruction, the taxonomy of critics, the systematic-role section, and the recommendations. | throughout |
| Jaakko Hintikka, “Cogito, Ergo Sum: Inference or Performance?” Philosophical Review 71 (1962), 3–32 | The object of the whole debate, read in full here: the Hamlet dilemma (p. 8), existential inconsistency (p. 15), process and product (p. 16), non-generalisability (p. 21), and the two interpretations (pp. 25, 31). | §§01–03, 07 |
| Hintikka, “A Discourse on Descartes’s Method,” in M. Hooker (ed.), Descartes: Critical and Interpretive Essays (1978), 74–88 | The later position; the method of analysis. | §09 |
| J. and M. B. Hintikka, “The Cartesian Cogito, Epistemic Logic and Neuroscience,” in The Logic of Epistemology and the Epistemology of Logic (1989), 113–136 | The Mark Twain skit (p. 113); the reply to Feldman, “a radical distortion” (p. 133, n. 4). | §§07, 09 |
| Georges Dicker, Descartes: An Analytical and Historical Introduction (2nd ed., 2013), ch. 2, pp. 48–49 | The two reasons the performatory reading “has been generally rejected.” | §06 |
| E. M. Curley, Descartes Against the Skeptics (1978), ch. 4 · L. Nolan (ed.), The Cambridge Descartes Lexicon (2016), “Cogito Ergo Sum,” 128–135 | Frankfurt’s constructive reading as Curley quotes it (p. 83); the three-way menu and the “Intuitively Grasped Inference” (pp. 129–131). | §§04, 08 |
Read this once; everything after it is elaboration.
| Hintikka’s claim | Textual anchor | Main objection | Verdict (chat, with this review’s amendment) |
|---|---|---|---|
| The syllogism fails as logic | Hamlet (1962 p. 8); Gassendi’s ambulo (p. 6) | None serious | Holds. Widely accepted. |
| “I exist” is existentially self-verifying | “whenever it is put forward by me” (CSM 2:17) | Feldman: “I think” is superfluous in “I think but do not exist” | Holds as an insight |
| The cogito is not an inference | Second Replies (CSM 2:100) | Frankfurt: the text denies only the syllogism | Fails if claimed exclusively. 1962 does not claim it exclusively (§07). |
| Momentary, first-person, not generalisable | CSM 2:17–18; 1962 p. 21 | Dicker: Descartes held any thought suffices | Holds for interpretation (ii); Hintikka grants (i) |
| The cogito is sui generis | 1962 p. 16: “not strictly speaking perceived by means of thinking” | The systematic-role problem: it cannot then be the model for the truth rule (CSM 2:24) | The decisive objection. Hence the hybrid. |
The deepest objection is not exegetical but architectural. Descartes draws the general rule, “whatever I perceive very clearly and distinctly is true,” from the cogito (CSM 2:24). If the cogito is certain for a reason unique to first-person existence claims, it is precisely not a representative clear and distinct perception, and the rule has no model. In the chat’s words, a performatively grounded cogito “looks like a sui generis fixed point that cannot underwrite a general rule about clarity and distinctness.”
That is why the field moved to the hybrid. It is also why the exegetical quarrel over the Second Replies matters less than it seems. Even a reader who grants Hintikka the text still has to explain how Meditation III gets its rule.
The brief’s first demand: “Steelman Hintikka in his own terms before criticizing.”
As logic, the syllogism is either question-begging or invalid. Hintikka’s alternative makes “I exist” the product of an act. Four features of the text then fall into place.
Hintikka’s paper is famous for its conclusion and less read for its argument. The argument begins with logic, not with Descartes.
Read the dictum as an enthymeme: whatever thinks exists; I think; therefore I exist. Hintikka shows this is a dilemma. If classical quantification theory’s existential presuppositions are kept, the step is valid but idle, because the existence of the referent of “I” was built into treating “I think” as a predication. If they are dropped, the step is invalid, and the counterexample is literary:
Hamlet did think a great many things; does it follow that he existed? Hintikka 1962 p. 8 checked
Dicker, reconstructing the same point, quotes Hintikka on the first horn: in logics where the pattern is valid, “the term which replaces a … must not be empty,” so “we in fact decided that the sentence ‘I exist’ is true when we decided that the sentence ‘I think’ is of the form B(a)” Dicker, ch. 2 n. checked.
The older objection comes from Gassendi, as Hintikka notes: “Gassendi already claimed that ambulo, ergo sum, ‘I walk, therefore I am,’ is as good an inference as cogito, ergo sum” p. 6 checked. Descartes’ reply locates the difference in awareness, not in logic:
“I may not, for example, make the inference ‘I am walking, therefore I exist’, except in so far as the awareness of walking is a thought. The inference is certain only if applied to this awareness…”
AT 7:352 · CSM 2:244 checked
A sentence p is existentially inconsistent for a person if “p and that person exists” is inconsistent. “I do not exist” is the paradigm. Its negation is then existentially self-verifying:
…as existentially inconsistent sentences defeat themselves when they are uttered or thought of, their negations verify themselves when they are expressly uttered or otherwise professed. Hintikka 1962 p. 15 checked
“Hence the indubitability of this sentence is not strictly speaking perceived by means of thinking (in the way the indubitability of a demonstrable truth may be said to be); rather, it is indubitable because and in so far as it is actively thought of. In Descartes’s argument the relation of cogito to sum is not that of a premise to a conclusion. Their relation is rather comparable with that of a process to its product. The indubitability of my own existence results from my thinking of it almost as the sound of music results from playing it…”
Hintikka 1962 p. 16 checked
The first sentence is the one the systematic-role objection seizes on (§08). If the cogito’s certainty is “not strictly speaking perceived,” it is not a clear and distinct perception at all.
And the word cogito names the act, not a premise: its function is “to express the performatory character of Descartes’s insight; it refers to the ‘performance’ (to the act of thinking) through which the sentence ‘I exist’ may be said to verify itself” p. 17 checked.
A fourth consequence completes the picture: Descartes as his own audience. He “arrives at his first and foremost insight by playing for a moment a double role: he appears as his own audience” p. 13, n. 22 checked. The chat quotes the later, more vivid form of the same idea: “Descartes’s little skit is analogous to someone’s, say Mark Twain’s, proving his existence to a skeptic by confronting the doubter and confirming his existence to him by saying: ‘I exist.’” Hintikka & Hintikka 1989, p. 113 checked
The chat puts non-generalisability at “1962, pp. 19–20.” The discussion begins on p. 20, but the sentence “Descartes’s insight is not generalizable” is on p. 21.
The brief’s second demand: the primary-text basis, and its contestation.
The Second Meditation favours Hintikka on momentariness. The Second Replies favour Frankfurt on inference. The Principles and the anticipations show that both structures were available to Descartes.
“The textual base is genuinely double-edged.” The chat’s summary, and the right one.
“…he does not deduce existence from thought by means of a syllogism, but recognizes it as something self-evident by a simple intuition of the mind. This is clear from the fact that if he were deducing it by means of a syllogism, he would have to have had previous knowledge of the major premiss ‘Everything which thinks is, or exists’; yet in fact he learns it from experiencing in his own case that it is impossible that he should think without existing.”
AT 7:140 · CSM 2:100 checked
The Lexicon entry, which the chat treats as the reference consensus, adds a textual point in Frankfurt’s favour from the Meditation itself. Even the discursive version “is, after all, inferential: ‘I must finally conclude’ (enfin il faut conclure) is an explicit expression of inference to the conclusion, I exist” Cambridge Descartes Lexicon, p. 131 checked.
Principles I.7 calls “I am thinking, therefore I exist” “the first and most certain of all to occur to anyone who philosophizes in an orderly way” (AT 8A:7, CSM 1:195). Markie stresses that Descartes repeatedly insists on the first person (ego cogito), which tells against Hintikka’s willingness to let the content be impersonal.
| Anticipation | Text | Bearing |
|---|---|---|
| Augustine | Si fallor, sum (De civitate Dei XI.26) | Reads most naturally as a deduction: if I err, I must exist to err. It sides with the inferential camp. |
| Descartes on Augustine | “…it is such a simple and natural thing to infer that one exists from the fact that one is doubting that it could have occurred to any writer. But I am very glad to find myself in agreement with St Augustine…” | To Colvius, 14 November 1640, CSMK 3:159 checked. Descartes himself calls it an inference. |
| Aquinas | “Nullus potest cogitare se non esse cum assensu: in hoc enim quod cogitat aliquid, percipit se esse” (De veritate X.12 ad 7) | Hintikka quotes it himself as having “closely anticipated” one of his formulations (1962 p. 16 checked). The structure is performatory. |
The Augustine connection is followed up in Chat 9, on Menn’s Descartes and Augustine. The chat’s conclusion on the anticipations is balanced: they “corroborate Hintikka’s insight without vindicating his exclusivity thesis.”
The catalogue of criticisms, split as the brief required into exegetical and philosophical objections.
Frankfurt and Dicker win the textual point. Feldman and Wilson press the philosophical one. But Hintikka’s paper anticipates Dicker’s first objection almost word for word.
“Widely regarded as damaging, though not annihilating.” The chat’s own grading of the strongest objections.
Dicker gives the textbook statement of why the reading lost:
This novel interpretation of the cogito, however, has been generally rejected by Descartes scholars, for two different reasons. First, it makes the certainty that I exist depend narrowly on only one specific thought; namely the attempt to doubt my own existence. But, as we have seen, Descartes believed that any of his thoughts, regardless of its content, established his existence. Second, when one asks exactly why trying to doubt my existence causes me to be certain that I exist, the only clear answer seems to be that I accept the argument: “If I try to doubt my existence, then I exist; I am trying to doubt my existence; therefore I exist.” But then the performatory cogito reduces to an inference or argument after all. Dicker, Descartes p. 49 checked
Frankfurt (1966; Demons, Dreamers, and Madmen, 1970) supplies the textual rejoinder of §04. Markie (1992) adds the first-person point of §05. The chat reports all three as the exegetical case against Hintikka.
| Critic | Objection, as the chat reports it | Held? |
|---|---|---|
| Feldman 1973 | (a) Existential inconsistency is under-explained. (b) In “I think, but I do not exist,” the first conjunct is superfluous, so thinking drops out. (c) Hintikka “falls prey to the same objections he levelled against the traditional theory” (p. 363). | not held |
| Wilson 1978 | “I will try to show that the so-called ‘performative interpretation’ does not provide a philosophically superior reading of the cogito” (p. 48). Against the Second Replies argument: “the claim that the cogito is an inference—even that it is a logically valid inference—is not equivalent to the claim that it is a syllogism. Modus ponens is a valid form of inference” (p. 50; the chat gives p. 56). The audience objection: “The assertion ‘I do not exist’ is self-defeating only if it is taken to reach its intended audience at about the same time that the intention originating it occurs” (p. 61). | held checked |
| Ayer 1956 | “I exist” as a degenerate case like “this exists”: performative in flavour but not substantive knowledge. | not held |
| Nakhnikian 1969; Vendler 1972 | “Pragmatic inconsistency” formalised; the grammar of silent thought versus public utterance. | not held |
| Kenny 1968; Curley 1978 | The “intuited inference” repair. Curley: Descartes “consistently blurs the distinction between inferences and propositions” (p. 79 checked). | Curley held |
The chat’s TL;DR says Hintikka “overreaches when it denies the cogito is any inference at all,” and its recommendations tell the reader to “reject his exclusivity.” Section VIII of the 1962 paper claims something different. The cogito, Hintikka argues, admits two interpretations, and Descartes held both without distinguishing them:
(ii) Descartes realized, however, that there is more to the Cogito than interpretation (i). … On this interpretation the peculiarity of the sentence ego sum is of performatory character. The verb cogitare now has to be interpreted rather narrowly. The word cogito may still be replaced by such “verbs of intellection” as dubito (or profero) but not any longer by verbs referring to arbitrary mental acts, such as volo or sentio. Hintikka 1962 p. 25 checked — interpretation (i), the inferential one on which “the word cogito may thus be replaced by any other word which refers to one of my acts of consciousness,” is on the same page
Two consequences follow. Dicker’s first objection is Hintikka’s interpretation (i). Hintikka agrees that Descartes held any thought establishes existence, and asks why cogitare is nonetheless privileged: “If Descartes could argue volo, ergo sum and sentio, ergo sum as well as cogito, ergo sum, why did he refuse to infer that his nature consists of ‘Wille und Vorstellung’?” (p. 31 checked). Dicker’s second objection meets the 1963 reply the chat itself quotes: “there is no incompatibility whatsoever between saying that cogito, ergo sum is a performance and that it is an inference.” The critics’ target is a stronger thesis than the one on the page.
The brief’s fourth question, “treat in depth”: can a performatively certain cogito be the model for the truth rule?
No. That is the real reason for the hybrid, and it survives the correction in §07. What Hintikka says of interpretation (ii) is exactly what makes it unusable as a model.
“This is where the performative reading is most exposed.”
“…So I now seem to be able to lay it down as a general rule that whatever I perceive very clearly and distinctly is true.”
AT 7:35 · CSM 2:24 checked
The rule is extracted from “this first item of knowledge.” It needs the cogito to be a perception, and a clear and distinct one, of the same kind as “2+3=5.” T2 says the cogito’s indubitability is “not strictly speaking perceived.”
Newman’s SEP formulation, quoted by the chat, is the same point: “There is no inconsistency in claiming a self-evident grasp of a proposition that has inferential structure.” not held The chat adds that Gewirth’s reading of the Third Meditation doubt as “unrestricted and distributive” presses from the other side.
Only partly. If the cogito carries both interpretations, interpretation (i) could serve as the model for the rule and interpretation (ii) could supply the special certainty. But the truth rule is drawn from the cogito’s certainty, and on Hintikka’s account that certainty belongs to (ii). The systematic-role objection therefore stands against the part of Hintikka that is distinctive.
Hintikka’s note gives Feldman’s paper as Philosophical Review 83; the chat’s vol. 82 (1973), pp. 345–363 is the correct citation.
An Austinian performative (“I promise”) is a public utterance that does something in the saying. The cogito is, for Descartes, silent thought. Hintikka accommodates this by letting a thought-act stand in for a speech-act and casting Descartes as his own audience (p. 13, n. 22). Wilson on the audience requirement and Vendler on the grammar of thinking argue that the analogy strains, because the features that make public performatives self-guaranteeing are exactly what silent introspection lacks. The same point is at the centre of Derrida’s critique in Chat 2, from the opposite side: for Derrida, the silence of hearing-oneself-think is the illusion.
What was checked, and what the check changes.
The chat’s bottom line holds: keep the insight, default to the hybrid for systematic work. Its characterisation of the 1962 paper as exclusive needs to be withdrawn.
“Adopt Hintikka’s insight, reject his exclusivity.” Good advice, applied to a thesis Hintikka did not hold.
Hintikka showed that the certainty of “I exist” lives in the act. His critics showed that a certainty living only in the act cannot be the model for anything else. Both are right, and the hybrid is the price of keeping both.
The next chat takes the act and its limit together. The certainty of the occurrence (tier one) and the certainty of a self (tier two) may not be the same certainty at all. Chat 5.
Click a node to collapse or expand its branch. Drag to pan; hold Ctrl or ⌘ and scroll to zoom.
The interactive argument map is available only in the web version; it reproduces, as a collapsible tree, the structure of the sections above.