Difficulties with Saying What One Thinks

Saying What One Thinks, by Léa Salje. Oxford: Oxford University Press, 2025, 160 pp. ISBN: 9780198914181. How do we know our minds – our thoughts, attitudes, and the like? Familiar sorts of accounts focus on, among other things, introspection understood as a quasi-perceptual faculty or process, reliable second-order ascriptive mechanisms, our ability to express our mental states by self-ascribing them, and various versions of the idea that we can gain first-personal knowledge of our own mental states by considering the world (so-called “transparency”). In these discussions the focus is generally on the cases in which self-knowledge comes easily. Few philosophers have considered cases in which we have to struggle to know even what we are thinking. Léa Salje's intriguing recent book, Saying What One Thinks, attempts to account for one sort of struggle. It explores the question, “why we sometimes have to put our thoughts into words before we can know what they are” (2025, 71) by focusing on what Salje calls “articulatory self-knowledge.” These are cases in which we first have a sense that we have a thought but can't quite identify what it is. We then try out various formulations to capture it, paying attention to our emotional and other reactions as we do so. Finally, we hit upon a formulation that feels right and that we settle on as capturing the thought's content, thereby coming to know our own minds in a way that (at the end point) does not seem to be alienated or based on evidence or inference, but rather as fully non-alienated as the ordinary case in which you can say right off the bat what you want for dinner. These are cases, she claims, which do not fall neatly into the familiar categories in the self-knowledge literature (3–4). The burden of the book is to argue that these cases reveal unexpected insights into non-propositional thought and our conscious mental life (64–67), the role of language in stabilizing and enabling person-level metacognition of our thoughts (68–74), the ways in which we can acquire creative understanding of deep structures in the world (82ff), the value of conversation and interpersonal interaction (94–116), and the possibilities for art as an expressive vehicle for thought (117–138). The core of the book is an explanatory account of the target phenomenon, built around extended exploration of one particular example. The example is meant to be representative, of course, and the account developed on its basis is meant to apply to the target phenomenon quite generally. Salje rejects views on which there is a “complete and determinate thought” (25) at the outset of the process (chapters 2 and 3). She likewise rejects views at the opposite extreme, which hold that the initial stage involves no content-bearing mental state at all (59). She proposes that the initial stage involves something in between these two options: “a non-propositional pattern of contents, which neither completely nor determinately determines a set of truth conditions” (68). As she elaborates this proposal using her central example, what is present in consciousness at the initial stage is a “non-propositionally structured pattern of mental contents” (64) – “a set of contents that does not have an internal structure of a sentence, and is not of the right form to be the straightforward content of a sentence” (65). It is rather an associative grouping or pattern of contents – in her central example, it involves ideas of her children, an abstract notion of power dynamics, a kinasthetic representation of back and forth motion, an emotionally resonant sense of pleasure (63–4). She proposes that such patterns or groupings have their source in the mind's tendency to shuffle or cycle in a non-directed way through mental representations (“monkey mind,” as she calls it, following a Chinese Buddhist tradition (64)). Unity is imposed on a group that appears in this shuffling process, she claims, by “a person-level act of pattern recognition” (65): “spotting a pattern (or seeming to spot a pattern) imposes unity on a collection of mental contents; by singling out a group of contents as interrelated in an interesting way, those contents just do coalesce into a non-propositional thought of the kind we are interested in” (65, italics in original). Salje's rough suggestion, then, is that the process described in stage (2) and culminating in stage (3) is a sort of “translational” process (pp. 68–9) that attempts to capture the initial non-propositional content in a linguistic form. Crucial here is the idea that the initial content places constraints on acceptable “translations” but “allows for a range of adequate stopping points, some of which may be better than others at verbally capturing the initial flash from Insight” (69). In this way, Salje wants to allow that there is an aspect of change in content involved in the passage from stage 1 to Stage 3, even while also holding to the idea that the result at Stage 3 (when everything is successful) captures (well enough) the non-propositional thought that was present at stage 1. For any two representations R1 and R2, R2 is a paraphrase or approximate re-expression of R1 iff: A given representation is true iff it represents things as being-thus-and-so and things are thus-and-so (120). … we finally have an answer to the starting question of the book, the question of why we sometimes have to put our thoughts into words before we can know what they are. The answer is that sometimes, our thoughts consist in dynamic non-propositional sets of diverse mental contents, and it is only by transforming such a dynamic non-propositional thought into a stable verbal structure that the thinker is able to metacognise it, and so to gain complete and determinate knowledge of what thought it is (71). I do not find it easy to understand how these pieces are supposed to fit together. I have questions about Salje's version of “non-propositional thought”, about her appeals to paraphrase and truth-conditions in this context, and about whether her account actually generalizes in the way she would like. [It] says nothing about how to specify things being thus-and-so, instead leaving the specification inherently ostensive; things are thus-and-so! Such an ostensive, rather than a (dis)quotational mechanism is extremely freeing, both in terms of what sorts of formats can have truth conditions of this kind, and what sorts of properties can be represented” (120). It won't do to reply that perhaps one can thus ostend the relevant truth condition without understanding the thought. Perhaps that would sometimes be possible with Kotarbinskian truth conditions, but it seems that on Salje's own view, we must be able to recognize the truth-conditions of these “non-propositional thoughts”: how else are we supposed to be able to recognize whether the verbal formulation we arrive at is a good enough approximation of them? At this point you might be thinking, “Wait a minute! How can pointing at that pattern/grouping (image of son, image of daughter, abstract notion of power dynamics, kinasthetic representation of back and forth motion/vacillation, feeling of alluring happiness) actually give you any insight into truth conditions at all?” That's certainly my reaction. And it raises in stark fashion the question of why we should think that that group of contents or representations even has truth conditions. Why think it's even the right sort of thing to have truth conditions, however incomplete and indeterminate? To focus our attention, and to pave the way for some points that I want to make later, it will be helpful to sketch Salje's central example more fully. On Salje's account, what is there at the beginning is just an associative grouping. (I don't know what she means to be adding by using the word “dynamic” here, as she so often does.) Here is her initial description of the grouping that appeared early in the morning: “something to do with power, or people,” “an abstract feeling of motion, a sort of back and forth, back and forth,” a feeling of happiness (63). As she describes things, that is all that is there at the beginning. Much later, as her reflections go on through the day, she finds herself thinking about her children. She then begins thinking about their relationship, and about changes in their interactions as the younger one becomes more capable of playing with the older one. Then she has the verbalized thought, “I can't control how they relate to each other, but amazingly they each seem to really flourish when in each other's company,” which seems important to her but also leads to a feeling of tension, and she adds to this the further thought, “My hope that this will last throughout their lives is intense, but it grates against my knowledge that relationships are ever-changing things” (63). (Salje's suggestion, of course, is that these verbalized thoughts are “paraphrases” of what appeared in her mind at the starting point of the process.) Something about power relations, or people, and back-and-forth motion, and a happy feeling. (In fact, the “something about” might be adding an element that isn't really there at the beginning either.) It can be tempting to think that the materials in the grouping at the beginning are the manifestations or upshots of some underlying attitude that is there from the beginning and that likewise gives rise to the later thoughts of the children – and that it is that underlying attitude that warrants treating this as a matter of the ascription of some properties to a particular portion of reality. But that is precisely not how Salje is understanding the matter. We consequently have to be very careful not to import that picture into our thinking about what is going on. On Salje's view, all that happens is that some representations flow through the person's mind in “monkey mind” fashion, and the person has a feeling that there is significance in the pattern, and then later some additional representations flow through their mind (the representations of the children), and they add those to the description of the pattern. There is no underlying attitude of which any of this is the expression or manifestation, on her view. But precisely because of that, it is hard to see why we should regard the pattern – however resonant the person might find it – as having truth-conditions (as saying things are thus-and-so). Salje owes us a much fuller discussion of how these sorts of groupings could plausibly be thought to have truth-conditions, given that there is not supposed to be any attitude underlying them. Salje doesn't give us any reason not to understand her central example in either of these plausible ways. Not only is there a question about whether Salje's account is the best fit for her example; there is also a question about whether her account generalizes to cover the other cases that she wants to cover. For instance, her account doesn't apply in any obvious way to another of her examples of the phenomenon she is attempting to elucidate. She describes a woman who is struggling to explain to her cynical friend why she takes steps to reduce her carbon usage, given the negligible effect of her individual actions in the face of global climate change (103–104). The woman makes several attempts and ultimately hits on a formulation that captures the psychological complexities of her position. The example does seem to be one in which a person is trying to articulate a thought that she initially grasps only dimly. But there is nothing here of representations cycling through the mind in “monkey mind” fashion and a pattern striking the person as resonant, and it doesn't look as though the person is trying to arrive at a paraphrase or approximate restatement. Rather, the example looks much more like the second sort of alternative described above: a case in which a person takes an initially inchoate smudge of thought and/or feeling and develops it into something sharply formulated. Of course, Salje could say that the monkey mind cycling and attention to patterns is happening in this case too, but below the level of consciousness. But now it looks like the tail is wagging the dog. Why, other than a commitment to Salje's account, should we thus assimilate this example to her central one? If I were trying to investigate the phenomenon that is of interest to Salje, my first question would be: what exactly is the phenomenon? And here I would want to look at a range of cases to try to sort out relevant similarities and differences. It isn't at all clear to me what should be clustered together for a unified account. I don't doubt that Salje experiences phenomena like her central example, so we would want to include such experiences on our list of things to be understood. As just noted, another sort of example is the one in which the woman is trying to explain her motivations and feelings about climate change to her cynical friend. It isn't at all clear that we should seek a single account to cover both of these cases. My own sense of things, pending further research, is that these are importantly different cases, and that they also differ significantly from the central example on which Salje builds her account. In fact, the second one would seem to be of a piece in certain important respects with an example that Salje opens her discussion with (pp. 13–14), in which a person has a vague sense of an objection to an argument and then works it out in detail. This seems to be a case of development, not a case of paraphrase or approximate restatement, and it doesn't seem to get initiated in the same way as Salje's central example. Yet another sort of example is captured in a delightful quote from Collingwood (1958) with which Salje motivates her discussion (12), in which Collingwood highlights the movement from an inarticulate experience of emotion, where one can only say something like “I feel … I don't know what I feel”, to the ability to express the emotion in verbally articulate form. It is not at all clear whether this is relevantly the same phenomenon as any of the others I've sketched, and it is hardly clear that cases of this sort are properly accounted for in the terms provided by Salje's account. Of course, in highlighting these various examples I don't claim to have catalogued all of the relevant sorts of cases. I have highlighted just a few examples of the wide range of cases in which we seem to need to struggle to put something into words before we can know our own minds. Given this variety, it is not clear what the relevant psychological “kind” is, nor how explanations of what goes on in one sort of case should relate to explanations of what goes on in others. I consequently think that before setting out to develop a general account, it would be fruitful to more fully consider the range of cases. I wouldn't be at all surprised if we will need to develop a range of answers to Salje's question, “Why do we sometimes have to put our thoughts into words in order to know what they are?” At the same time, I want to acknowledge that Salje's own account points in a very interesting direction. Let's leave aside the notion of paraphrase and think for a moment about the following idea that is fundamental to psychoanalysis and related forms of psychotherapy: when a person is presently unable to know their own mind but has a sense that there is already something there to be known, it can often be helpful to encourage them to think associatively – to encourage them not to look for rational connections, but to just see what comes up and to work with it and the resulting emotional responses. This looks very much like the process that Salje describes in her example. In therapy and psychoanalysis this process can sometimes result in an articulation of something that was plausibly there before, other times in a verbalized development of an initial inchoate element, but either way, provided all goes well, the end result is not that the person has arrived at some form of inferential self-knowledge; rather, the person is in better first-personal contact with their own mind. How that happens, across a wide variety of types of cases, is an interesting and important question.1