Text published in the journal Ratio 14, 2001. A revised version of this text will be published in the book Lines of Thought: Rethinking Philosophical Assumptions (Cambridge Scholars Publishing, 2014).
– 11 –
‘I AM THINKING’
My first purpose in this paper is to examine what people might mean when judging or saying that they are thinking. This examination will reveal some unexpected features of this kind of judgment that have destructive consequences for our evaluation of the supposedly self-verifying character of the statement ‘I’m thinking’ and, further, for our standard analysis of the claims of certainty for the Cartesian cogito. I begin by considering some concrete situations in which one could say ‘I’m thinking’.
I. An ordinary language approach to ‘I’m Thinking’
Consider the following dialogues:
(i) Meg and her husband Carl are planning to take a short bicycle trip. As Meg prepares to go outside, Carl leans out of the window and lingers there. Seeing her husband’s passivity, Meg asks him, somewhat bored: ‘What are you doing now?’ His answer is, ‘I’m thinking...’. Hearing this, Meg asks: ‘And what are you thinking about?’ He replies: ‘I’ve been thinking that it would be better to go by car, since it seems that the weather will change’.
(ii) A theoretical physicist presents a problem to a colleague. At first, his colleague says nothing, his head bowed, visibly thinking about the problem. After some moments, as if to justify his silence, he says: ‘I’m thinking… I believe I can find the right answer’. After a while, he smiles and gives his colleague the solution he was seeking.
Examples in which a person says to herself that she is thinking are less common, but there are situations in which this occurs. Consider the following:
(iii) Mary is a student preparing an oral presentation on medieval arguments for the existence of God. She is pondering Gaunilon’s first objection to Anselm’s argument, when she is distracted by the sound of a telephone ringing in her neighbor’s room. Then, trying to recover the threads of her thoughts, she asks herself, ‘Now, what was I just thinking about?’ When she at last remembers, she says to herself, ‘Yes, I was thinking that…’ and takes up her mental argument where she left off.
What I wish to point out is that these three examples, along with other concrete examples of the self-attribution of thinking, generally share a common pattern, which requires closer examination.
The first thing to be noted is that when a person thinks that she is thinking, she can always ask, ‘What am I thinking about?’ and, in principle and more reasonably, another person would also be entitled to ask her, ‘What are you thinking about?’ In the first of our examples, this question was posed by Meg, and in the third, Mary asked herself the question; in the second example, the question was not asked, but obviously could have been asked. Also noteworthy is the kinds of answers we give to such questions. In the first example, the answer consisted in pointing to a thought that Carl had been thinking a moment before, namely, that the weather could change. In the second example, if the first physicist had asked his colleague, ‘What are you thinking about?’, he would probably have answered: ‘Wait a moment; let me finish my reasoning…’, And later he could have said: ‘I was thinking that…’, followed by a description of his thoughts. In the third case, a more detailed answer would be, ‘I’m thinking that… and I am trying to come to the conclusion that…’. In what follows, I will analyze the common pattern in the self-attribution of thinking, using such answers as heuristic clues.
Considering such answers, I will argue that the thought ‘I’m thinking’ can usually be restated as ‘I was thinking that p’, a point that for expositive reasons I subdivide into the two following claims:
(a) ‘I’m thinking’ generally means ‘I’m thinking that p,
(b) ‘I’m thinking that p’ generally means ‘I was thinking that p’.
Let us begin with (a). At least in standard cases (since we will later see that there is a conceivable exception), the question of what a person is thinking leads, when completely answered, to the presentation of another thought or thoughts, which may complete the occurrence of ‘I’m thinking’ by describing what the person is thinking. In other words, in concrete situations, with the question ‘What are you (or what am I) thinking?,’ the answers usually have the form ‘I’m thinking that p’, where p stands for one or more thought-contents (or propositions) that the person is thinking. In some cases, p stands for only one thought. An example would be if Carl had said, ‘I’m thinking that the garden needs to be watered’. But p often stands for a complete chain of thought. Thus, if the physicist in the second example were later asked to state what he was thinking, he would say something in the form: ‘I was thinking that p1, p2, p3… which has led me to the conclusion pn’.
It appears necessary that what is thought must at least be a complete thought-content: It makes no sense to say that one thinks parts of a thought or a mental event that is not a thought. Certainly, I can say, for example, that I was thinking about the
But the word ‘about’ already makes it clear that I’m not referring to the
meaning of this name, and still less to an image of this bridge, but rather to
the thoughts that accompanied this image or idea. Finally, it is valuable to
note that when asked, ‘What are you thinking about?’, one could also give the
unsatisfactory answer: ‘I’m thinking something’ (or ‘I’m thinking x’, where x means ‘some
thought’), leaving p unspecified. In other words: ‘I’m thinking that p’
can be shortened to ‘I’m thinking something’, which can be further shortened to
‘I’m thinking’. Although they have different linguistic meanings, what is meant
by all three phrases is better expressed by ‘I’m thinking that p’. Golden Gate Bridge
Now suppose that when Meg asked, ‘What are you thinking about?’, Carl had answered, ‘I’m thinking about nothing’ or ‘I’m thinking about this thought’. In such cases, Meg, being no philosopher, would have good reason to be angry, since these answers are devoid of sense. One cannot meaningfully say ‘I’m thinking’ without assuming that this expression can be complemented by a thought-content. If I say to myself, ‘I’m thinking’, without reference to a thought-content, then I’ve not thought, but only spelt these words in my mind. I’m not proposing that in such cases there is no mental occurrence, or that I’m not conscious of one, but rather that it is not the real experience of a thought. When I silently spell out in my mind the combination of words ‘I-am-thinking’, I’m not having an awareness of thinking, but only an awareness of what we could call a syntactic occurrence in my mind, namely, of my inwardly spelling ‘I-am-thinking’, recognizable as a linguistically correct sentence, etc. This syntactic occurrence is semantically empty, since it is devoid of content, expressing an impossible thought like ‘This square is round’. In sum: it is not possible to judge that I am thinking without being able to refer to a complementary thought-content that provides a basis for this judgment.
This point becomes more compelling when we compare ‘I’m thinking’ with other phrases including verbs of propositional attitudes, such as ‘I hope’, ‘I wish’, ‘I believe’ or ‘I doubt’. They can’t really express wishes, hopes, beliefs or doubts without their objects. If, for example, I say to myself ‘I’m hoping’ without having something to hope, it is not true that I am hoping. These are mere mental occurrences of symbols, and I am obviously aware of them, but they are not accompanied by a real occurrence of hoping, etc. And the same applies when I say to myself, ‘I’m thinking’ without intending a certain thought. Or suppose that Carl says to Meg, ‘I promise you...’, and when Meg asks, ‘Well, what are you promising?’, Carl answers, ‘I promise you, and that is all’. In this case, Carl’s utterance fails to make sense, since it does not refer to the propositional content of a promise and thus is a void speech-act and not really a promise. Something similar can be said of ‘I’m thinking’. Just as saying ‘I’m promising’ without referring to the content of the promise does not amount to a real promise, ‘I’m thinking’ without reference to a thought-content does not amount to a real thought.
Suppose now that Meg asks Carl, ‘What are you thinking about?’, and Carl responds, ‘I’m thinking that I’m thinking, and that’s all’, or ‘I’m thinking my own thought that I’m thinking’. Does this make any sense? Suppose that Carl says to Meg, ‘I promise that I promise and that’s all’, or ‘I promise my own act of promising’, or that he says to himself ‘I hope my own hope’ without intending to hope anything. Clearly, just as no promise is made and no hope is hoped, no thought is thought. This shows that in ‘I’m thinking that p’, p can’t be replaced by ‘I’m thinking’ to complement the sentence. Expressing this in a more general form: It is not possible to judge ‘I am thinking that p’, where p is constituted only by ‘I am thinking’ or by iterations of ‘I’m thinking’. (The adverb ‘only’ is important here, since if p were replaced by some other thought containing ‘I’m thinking’, the whole answer would acquire a sense. If Carl had answered, ‘I’m thinking that I’m thinking that the weather will change...’, the whole sentence would not be meaningless.)
Now we turn to the next claim: (b) What we mean by ‘I’m thinking that p’ can generally be restated as ‘I was thinking that p’, because whenever a person thinks ‘I’m thinking that p’, what she really means is ‘I was thinking that p’. Consider our examples again. Carl says to Meg: ‘I’m thinking that the weather will change and that it would be better to go by car’. But what Carl really means is not that he is thinking those things at the same time that he is answering Meg’s question, as the present-tense of the verb ‘to think’ misleadingly suggests. What he means is, more precisely stated, ‘I was thinking that the weather would be bad and…’. Again, when the physicist says, ‘I’m thinking… I believe I can find the right answer’, he doesn’t mean that the contents of his thoughts are occurring in his mind at the exact same time that he is uttering this sentence. What he means, more precisely stated, is ‘I have been thinking some thoughts..., and I expect that once I have completed thinking my thoughts, I will have the solution’. Finally, when Mary says to herself ‘Yes, I’m thinking that...’, what she means is ‘Yes, I have been thinking that...’. Therefore, it is clear that in the usual cases, ‘I’m thinking…’ is an imprecise way to say ‘I just thought…’ or ‘I was just thinking…’. Indeed, the thought or thoughts referred to by ‘I’m thinking’ usually occurred in the past. If someone makes references in a referred-to chain of thoughts to future thoughts, to the conclusions of a process of reasoning, his intended thoughts were in fact already thought in the past. It is because those thoughts were thought in a very recent past that language misleadingly allows us to say ‘I am thinking’, instead of ‘I just thought’ or ‘I was just thinking’. Were ordinary language fastidiously accurate, we would not say ‘I’m thinking’. I conclude, therefore, that the most precise restatement of ‘I’m thinking’ is typically ‘I just thought (I was just thinking) that p’.
One could ask: ‘Why should this be the case?’ The answer is easy to find: our consciousness is incapable of entertaining or focusing on more than one cognitive event at the same time. I can’t consciously think a thought at the same time that I consciously think that I’m thinking that thought. I must first finish the thought I’m thinking in order to think that I have thought this thought, which demands that this thinking must necessarily refer to a thought already focused on in my consciousness. One could object that it is possible for a person to say to himself, ‘I’m thinking that 12 x 12 =
144’, without reference to the
past thought that 12 x 12 = 144, simply by thinking this. This sounds
linguistically awkward, different from the usual cases we have considered, but
it becomes meaningful when paraphrased by the statement, ‘Now, I will think
something; here it is: 12 x 12 = 144’.
However, this shows again that the two occurrences – the thought and the
awareness that it will be thought – are not simultaneous, but only in reverse
Contrary to the claim that the two mental occurrences – ‘I think’, as an awareness of a thought, and the thought that is the object of that awareness – can’t be focused on simultaneously in consciousness, one could reply that in some cases one can indeed think the thought ‘I’m thinking p’, while still maintaining the thought of p in one’s consciousness. However, this is somewhat misleading. It leaves out the difference between focus and fringes in the diachronic (temporal) unity of consciousness. To think 370 +
person could think 300 + 500,
in order to reach 800, and then add 70 plus 10 to 800
and thus obtain 880. When a person is thinking the thought that 70 + 10 = 80,
she retains in her short-term memory the result 80 and how she achieved it.
This does not, however, mean that the two mental acts are being jointly focused
on in her consciousness, or that they can be thought at the same time. While a
person’s consciousness is focusing on its present thought, it retains in
short-term memory a thought that was already finished, in order to bind the
process of thought into a continuous unity. What is not being focused on is not
in our consciousness in a relevant way, and is likewise by no means certain, as
our frequent arithmetic mistakes make clear.
We should also consider the possibility that the thought of p is simultaneously accompanied by a nearly conscious, pre-conscious or unconscious (higher-order) thought that one is thinking p in the synchronic (contemporaneous) unity of consciousness, and even by other non-conscious thought processes. I believe that this can be actually the case, but since the prior thought is not being focused on or entertained in consciousness, it is not the kind of thing we are considering here and thus has no relevance for our conclusions.
Some might still object that, independently of this, we still have a self-reflexive consciousness that we are thinking. This suggests something like: (i) when I think p, I simultaneously have an equally conscious second-order thought about it, namely, ‘I’m thinking p’, or (ii) when I think p, p always appears as part of the thought ‘I’m thinking that p’. However, our analysis rejects (i), and (ii) not only lacks sufficient intuitive evidence, it is refuted by the existence of unconscious thoughts.
Nevertheless, there is an obvious way in which it is plainly possible to have second-order thoughts and to think that we are thinking, namely by retrospection. I can think that I thought p in the same sense that Carl thinks he was thinking that the weather would change. ‘I’m thinking’ can occur as part of p in ‘I’m thinking that p’, and in this way I really can have thoughts about past occurrences of thinking: by remembering that a process of thinking something occurred, that it went on in my mind.
Until now, we have analyzed the standard case of ‘I’m thinking’. However, there is at least one conceivable non-standard case that deserves consideration:
(iv) During a conversation with some friends, it occurs to Mary to add a new idea to a paper she is writing. At this moment, someone interrupts her thoughts with a question. After she answers the question, Mary asks herself: ‘What was I just thinking about?’, but no answer occurs to her. She knows that she had a new idea for the paper, but she is unable to remember it. She can only say to herself, ‘I was thinking about something, but I don’t know what’.
This is a rather common experience: often one can remember that one was thinking, but not what one was thinking. However, this kind of episode is ordinarily verbalized as ‘I was thinking’, and not as ‘I am thinking’. The explicit use of ‘was’ in this example shows that it concerns thoughts that were thought shortly before, rather than thoughts we have just finished thinking before we say this. A delay results from something external or internal that distracts us from our own chain of thought. Nevertheless, since I don’t wish to discard any reasonable possibility, I will also accept this case as a conceivable analysis of a (linguistically awkward) use of ‘I’m thinking’. In this case, the form of ‘I’m thinking’ will be not the usual ‘I was thinking p’, but rather ‘I was thinking something (a thought or thoughts), although I don’t remember what’. More concisely, ‘I was thinking x’, where x doesn’t mean a particular thought like p, but refers to some proposition in an indeterminate manner.
It is worthwhile to remark that this is a derivative case, dependent on the standard one. If Mary answers her question in the form ‘I was thinking x’, one is still entitled to ask: ‘Why are you so sure that you really were thinking any thoughts, if you can’t remember them? How do you know that it was not an illusion?’ A natural answer could be, ‘Because I have had other experiences like this and later remembered what I had been thinking’. If she could never remember the thought that she has the impression of having been thinking, she couldn’t be sure that she had really been thinking some thought, instead of having the mere illusory impression of having been thinking one.
Now I believe that we have learned enough about the phenomenological grammar of ‘I am thinking’ to consider how it has been misunderstood in philosophy.
II. Some Philosophical Consequences of the Proposed Analysis
The first consequence of our conclusions is that ‘I am thinking’ is no longer a self-verifying judgment that cannot be denied without inconsistency. Turning now to the Cartesian cogito, there is general agreement that ‘I am’ or ‘I exist’ is a self-verifying, certain, incorrigible judgment, since one must exist in order to be conscious of one’s existence. This is shown by the fact that the negative of ‘I exist’ – ‘I don’t exist’ – is, if not senseless, at least a necessarily false judgment. And authors like A. J. Ayer and Jaakko Hintikka have quite naturally extended this conclusion to ‘I’m thinking’. The latter is also a self-verifying judgment, since its denial – ‘I’m not thinking’ – is at least a necessarily false judgment, for if I judge that I’m thinking, how could I possibly not be thinking?
I agree with the interpretation of ‘I exist’ as a self-verifying, incorrigible judgment, since the thought ‘I don’t exist’ (in the sense of ‘I do not presently exist’) is unavoidably false. But the extension of this conclusion to ‘I’m thinking’ is a mistake. Although it is a very certain judgment, what we mean by ‘I’m thinking’ is not a self-verifying, incorrigible, undeniable certainty. Since the thought or judgment that I’m thinking really means ‘I was thinking that p’ or ‘I was thinking x’, it is in principle possible that the judgment that I’m thinking is mistaken.
To grasp this, suppose that Carl is suffering from a serious kind of mental confusion, incorrectly identifying his own memories. When Meg asks him what he is doing, he answers, expressing the thought r, by saying that he is thinking (was thinking) that: ‘The flowers in the garden need to be watered’. Nevertheless, it could be the case that instead he thought s: ‘Someone should mow the lawn’. It could even be that Carl really was not thinking anything at all: he may have been in a state that neurologists call absence. In the first case, the judgment ‘I am (was) thinking’, in the sense of ‘I am (was) thinking something’ or ‘I am (was) thinking x’, is true. The more determined judgment ‘I am (was) thinking that r’ is false, however, since what he was thinking was s and not r. In the second case, not only is the whole judgment ‘I am (was) thinking that r’ is false, since r was not thought, but even the judgment ‘I am (was) thinking something’ is false, since there was no thought at all. For the judgment ‘I’m thinking’ to be true, a thought must be thought, even if it is a false one. (A similar point can be made in many other cases. When I say ‘I have promised to talk’ even though I have actually promised to remain silent, the statement ‘I have promised to talk’ is false, but not the statement ‘I have promised’. In contrast, if I say ‘I have promised to talk’, and I really have not promised anything, both statements ‘I have promised to talk’ and ‘I have promised’, are clearly false.)
The same applies to the adventitious case in which ‘I’m thinking’ means the same thing as ‘I was thinking a thought (or thoughts), but I can’t remember what’ or ‘I was thinking thought x’, where x means ‘some thought’. In this case, ‘I’m thinking’ will be true if there is (was) a certain thought to replace x, otherwise it will be false. Since it is possible that someone mistakenly thinks that he was thinking some thought, even though he or she actually wasn’t thinking anything, ‘I’m thinking’, in the sense of ‘I was thinking x’, can be false. And this means that it, like ‘I’m thinking that p’, is not self-verifying.
Now we will consider the consequences of our analysis for the Cartesian cogito, the well-known judgment usually expressed with the sentence ‘I think, therefore I am’. An orthodox way of interpreting the cogito is by reconstructing it as a kind of inference. Thus, if we use as a rule of inference the proposition ‘If a thing is identified as having an attribute, then this thing exists’, we can interpret the cogito as expressing the following immediate inference:
1. ‘I am (a thing having the attribute
of being) thinking’ (Rule applied to 1: if a thing is
_________________________ identified as having an attribute,
2. ‘I am (an existing thing)’ then this thing exists.)
Assuming that the premise can’t be false, it follows that the conclusion also can’t be false, which proves my existence as a thinking being. A reconstruction more or less similar to this is accepted by many interpreters as capable of producing the kind of certainty that Descartes was seeking, and there are good reasons to believe that this is the best way to make sense of Descartes’ own intentions, even if his texts show that he also had other possibilities in mind.
Nevertheless, if the conclusions derived from our analysis of the real self-attribution of thinking are correct, then all reconstructions of the cogito as an argument are, from a heuristic, systematic point of view, doomed to fail. For ‘I’m thinking’ to be a thought of something, it must be understood as ‘I was thinking that p’ (or, derivatively, as ‘I was thinking x’). But ‘I was thinking that p’ (like ‘I was thinking x’) is a corrigible, uncertain thought, which means that it provides an uncertain premise when included in an argument. But if the premise of the cogito-inference is uncertain, then the conclusion of the inference – ‘I exist’ – must be at least as uncertain as the premise. This is not a very useful result, since the conclusion – ‘I am’ – is already known to be certain by simple introspection. Even the reverse argument, ‘I exist, therefore I think’, would be more compelling, since to have a conscious awareness that I exist, I must be thinking, at least in some sense.
Descartes’ hypothesis of the malign genie only makes things worse. Although the malign genie can’t force me to think that I am when I am not, he can without difficulty make me think that I was thinking p when I wasn’t, or that I was thinking something, even if I had no thoughts at all. And this must be very easy for him, since he is capable of misleading us in the most simple reasoning demanding short-term memory, as in the case with the four arithmetic operations (see our previous remarks about the focus and fringes of consciousness).
Defending the inferential way of interpreting the cogito, one could argue that I am using the proposition ‘I think’ in a sense that Descartes never intended. He aimed to use this proposition in a self-reflexive, philosophical sense in which there is no need to refer to a complementary thought-content. Consequently, the cogito must have the form ‘I’m thinking this thought itself’. However, our more precise consideration of the structure of ‘I’m thinking’ in the first part of this paper already ruled out this suggestion as illusory. The reflexive use of ‘I’m thinking’ is a philosophical fancy without any intuitive support.
Yet, one could credibly argue that the statement ‘I’m thinking’ without a complement can be at least a syntactic mental occurrence, and that when it happens in my mind, I am immediately conscious of this occurrence. This is certainly true, as we have already seen, and it is also true that I achieve consciousness of my existence through having such experiences. But it is not enough to guarantee that I’m thinking, since the consciousness of a syntactic occurrence is not the consciousness of what is at stake, namely, what is to be thought by means of the syntactic occurrence.
Finally, one could consider what Descartes himself would probably answer when pressed by these considerations. Perhaps he would give an answer like the one he gave to Frans Burman’s comment that we are unable to focus our attention on more than one thought at one time while we are reasoning. Descartes’ answer was that although we can’t grasp very many things, we can grasp more than one: For example: ‘I now grasp and think that I am speaking and eating simultaneously’ (my italics). Transposed to ‘I’m thinking’, this kind of answer would mean: although it doesn’t seem to be the case, I’m really able to think ‘I’m thinking’ and the content ‘p’ simultaneously, and this guarantees the certainty of ‘I’m thinking’.
Our response to this is that the distinction between the synchronic and diachronic unity of consciousness, jointly with the distinction between its focus and fringes, shows that in his answer to Burman, Descartes takes an equivocal view, confounding mere awareness with conscious thought. While I’m writing these words, for example, I’m also aware of the light from the lamp beside me, hear the hum of the air-conditioner, feel my fingers touching the keyboard, am aware of my body position, etc. All these things are on the fringes of the synchronic unity of my consciousness. It is correct to say that I’m aware of all these things, that in a certain sense I’m grasping all of them (if the air-conditioner stops running, I will direct my attention to it). However, it is certainly not correct to say that I’m thinking all or even some of these things, as Descartes would be forced to suggest. The only things I’m consciously thinking are the words I’m writing, which are passing through the focus of my consciousness to lose themselves in the temporal extension of its diachronic unity. Consequently, Descartes gives us a bogus example, which would be correct only if he could show that we are in fact simultaneously able to think more than one thought in the sense of articulating these thoughts simultaneously in the center of our attention, when in fact we really aren’t able to do so. But if it is so, if there should be a thought p on the synchronic fringe of my consciousness, it could also be an illusion. As already noted, it is plainly possible that I have unconscious or pre-conscious thoughts accompanying the thought that is being focused on in my consciousness. However, this would be the kind of thing within the reach of the malign genie, for he would be able to confuse the relationship between what is focused on in consciousness (what is clearly and distinctively experienced by us) and what is not. What is not focused on is only an object of secondary awareness and can always be wrongly interpreted.
In conclusion, our analysis of ‘I’m thinking’ favors the unorthodox view of the Cartesian insight that concentrates analysis on the role of ‘I exist’ as a self-verifying, incorrigible, undeniable truth. Even if this view fails to give us the most adequate textual interpretation, it reflects, I believe, the truth about the matter.
 Similar distinctions are made in William James, The Principles of Psychology (New York: Dover (1890)), vol. 1, chap. 9 (distinction between focus and fringes of consciousness). See also J. R. Searle, Mind, Language and Society: Philosophy in the Real World (New York: Basic Books, 1998), pp. 74-5.
 This is, to my mind, the reason why my argument doesn’t render vacuous the theory that identifies an important sense of the word ‘consciousness’ with having suitable reflexive, simultaneous higher-order beliefs or thoughts about lower-order thoughts or other mental events. According to this view, I can still say that p is a conscious thought, because the thought of p is accompanied by a higher-order non-conscious thought of p. David Rosenthal, a defender of this view, also notes that a second-order thought cannot be conscious unless when it is the object of a third-order thought that remains unconscious… so that the thought that is on the top remains always unconscious. See David Rosenthal: Consciousness and Mind (
Oxford University Press, 2005), part I. Oxford
 See A. J. Ayer, ‘I Think, Therefore I Am’ in, W. Doney (ed.): Descartes: A Collection of Critical Essays (New York: Doubleday, 1967), p. 81.
 As A. J. Ayer writes: ‘The sense in which I cannot doubt the statement that I think is just that my doubting it entails its truth: and in the same sense I cannot doubt that I exist’ (op. cit., p. 81). Though arguing for distinctions, Jaakko Hintikka also accepts a similar identification. (See ‘Cogito ergo sum: Inference or Performance?’ in W. Doney (ed.), op. cit., pp. 133, 139.)
 See A. Kenny, Descartes: A Study of his Philosophy (
South Bend, IN:
Press 2010), cap. 3. See also Bernard Williams, ‘The Certainty of the Cogito’
in W. Doney (ed.), op. cit., pp. 88-107. St. Augustine
 Descartes’ argumentative path finds its synthesis in the formula: ‘To be doubting I must be thinking; to be thinking, I must exist’. Still, it is difficult to see how the last and essential claim could be made without some kind of inference. Descartes recognizes explicitly the necessity of inference in Principia Philosophiae (cf. V. R. Miller & R. P. Miller (eds.), Principles of Philosophy (Dordrecht: Reidel, 1983), Part I, § 10).
 If the ‘I’m thinking’ in the cogito is conceived as having the form ‘I was thinking x’, being in this form derivative, it could possibly be true only presupposing the previous truth of thoughts of the form ‘I was thinking that p’.
 Claudio Costa, ‘Über den Gewiβheitsanspruch im cartesischen cogito’, Prima Philosophia, Nr. 4, 1999, 1-20, p. 10.
 ‘Quod mens non posit nisi unan rem simul concipere, verum non est; non potest quidem simul multa concipere, sed potest tamen plura quam unum; e.g. jam ego concipio et cogito simul me loqui et me adere’. Adam & Tannery (eds.), Oeuvres de Descartes (Paris: Vrin/CNRS, 1964-1976), V, 148-149. Cf. also J. G. Cottingham (ed.), Descartes’ Conversation with Burman (Oxford: Clarendon 1976), pp. 6-7.
 This view is to be found in different versions in A. J. Ayer, see W. Doney (ed.), op. cit., pp. 80-87, Jaakko Hintikka, see W. Doney (ed.), op. cit. pp. 108-139, and in Harry Frankfurt’s book, Demons, Dreamers and Madmen (Indianapolis: Bobbs-Merrill, 1970), p.