1 Introduction

It is generally agreed that experience plays the role of mediating between the mind and the world. The challenge, according to McDowell (1996), is to come up with a conception of experience that meets the following two desiderata:

  • (EJ) Our empirical thinking is subject to rational constraint from the world by virtue of our perceptual experiences;Footnote 1

  • (SR) Our empirical thinking, including beliefs and judgments, belongs in the logical space of reasons, irreducible to the logical space of nature;

where “EJ” and “SR” are shorthand terms for experiential justification and the space of reasons, respectively. The seeming tension between (EJ) and (SR) has been polarizing modern philosophers—they tend to overweigh one while underplaying the other. One pole is to accept (EJ) but reject (SR), thereby committing oneself to the Myth of the Given: the idea that experiences, conceived of as standing outside of the space of reasons, can nonetheless serve as independent grounds that justify our beliefs. Unfortunately, “the idea of the Given offers us exculpations where we wanted justification.” (McDowell, 1996, p. 8). The other pole is to adopt a view that maintains (SR) while renouncing (EJ), like the Davidsonian coherentism. However, such a view makes experiences epistemically inert and hence detaches our system of beliefs from the external world entirely—turning it into “a frictionless spinning in a void.” (Ibid., p. 11).

McDowell’s diagnosis is that the two opposing views both presuppose an isolation of experience from the space of reasons. The way to remove the gap, as McDowell proposes, is to endorse conceptualism:

  • (Conceptualism): The content of experience is totally conceptual.

In his view, it is only because the content of experience is conceptual that it can stand in rational relations of justification to beliefs and judgments. For the same content of experience can also be the content of a belief or judgment or be exploited by a belief or judgment at least.

Conceptualism has spawned a lot of controversies. While the view seems to do a pretty good job in explaining the rational linkage between experience and belief, critics charge it with putting too stringent a constraint on the content of experience. Tony Cheng, in his book, John McDowell on Worldly Subjectivity (2021), explores the possibility of a McDowellian outlook without conceptualism. He urges that to carry out McDowell’s project of resolving the tension between (EJ) and (SR), what is vital is not to insist on conceptualism but only to keep with the rationality of experience. In Chap. 7, Cheng gives a critical review of McDowell’s thesis on the co-extensiveness of the space of reasons and the space of concepts and suggests that conceptualism is dispensable.

I have doubts about Cheng’s suggestion. I do not think that conceptualism can be dispensed with for the reasons he gives. In this commentary, first I will examine his dialectical approach to the thesis; doing so requires a clarification on McDowell’s conception of concept along the way, which in turn paves the way for my discussions of Cheng’s contentions in the subsequent part of this essay. Due to the limit of space, I shall only focus on two points he addresses that seem to me particularly critical.

2 The co-extensiveness thesis: stipulative vs. substantive

Let us first have an overview of Cheng’s argumentative strategy. He formulates McDowell's co-extensiveness thesis of the rational and the conceptual as the biconditional:

  • (CoX) Something is conceptual if and only if it is rational, (2021, p. 134)

or “conceptuality ↔ rationality” for short. Now, the crucial move he makes is to hold that the biconditional cannot all be a matter of “stipulation.” Otherwise if McDowell could stipulate whatever he wishes the conceptual to be, which would “trivialize the co-extensiveness thesis” and “make the debate between conceptualism and non-conceptualism unintelligible” (p. 139), Cheng does, however, grant one particular direction, “the conceptual → rationality,” to be a stipulation, and yet maintains that the other direction, “rationality → the conceptual,” must not be.

Whereas most of the effort he spends is on undermining on the latter conditional, I think it is crucial to get a clearer sense of what it means for the former conditional to be a “stipulation.” Cheng grants that “conceptuality → rationality” is made true by stipulation. In particular, he cites a passage of McDowell’s as textual evidence:

It is important that the connection between conceptual capacities and rationality is a stipulation. It is not that there is a universally shared idea of conceptual capacities, which determines a subject matter about whose properties people disagree. The notion of the conceptual can be used in a variety of ways, for a variety of purposes. (2008a, p. 132)

However, Cheng thinks that there is “a trouble with this”:

[T]his is at odds with many philosophers’ hope that talks about concepts in philosophy can make contact with theories of concept in empirical sciences. (2021, p. 134)

Similarly, when considering whether “Kant’s conception of concept fits McDowell’s stipulation” (p. 131), Cheng remarks,

McDowell says that in his context “spontaneity” can be simply a label for the involvement of conceptual capacities (McDowell, 1996, p. 9), but this is tendentious. Maybe Kant can do that, but that is not so for us, since in our era the notion of ‘concept’ can mean various things, and to stipulate all of our spontaneous capacities are conceptual is not helpful for discussions. (p. 132, my italic)

However, so remarking on McDowell’s conception of concept as a stipulation makes it look as arbitrary as stipulating the denotation of a logical constant to be whatever object in the domain we please. But apparently, “conceptuality → rationality” is far from arbitrary. As I diagnose it, Cheng seems to conflate the semantic question and the substantive question of McDowell’s conception of concept.Footnote 2 The semantic question concerns the meaning of the word “concept” as McDowell uses it. In contrast, the substantive question concerns whether (CoX) is true, whether the space of concepts is co-extensive with the space of reasons. In the substantial sense, neither direction of the biconditional, “conceptuality ↔ rationality,” is a statement about meaning.

With the distinction in mind, let us first clarify the meaning of “concept” as McDowell uses it. I do not think the particular passage of McDowell’s which Cheng’s quotes does enough justice to his construal. Rather, to better articulate what his is up to, it is instructive to look a few pages ahead, where he first uses the term “stipulation”:

I find it helpful to connect the idea of conceptual capacities with this notion of rationality. I use the idea of conceptual capacities in a way that is governed by this stipulation: conceptual capacities in the relevant sense belong essentially to their possessor’s rationality in the sense I am working with, responsiveness to reasons as such. (2008a, p. 129, my italics)

This stipulation is not trivial. McDowell explicates the notion of the conceptual by referring to the idea of responsiveness to reasons as such. The idea is that whereas both rational animals and non-rational animals are responsive to reasons, only the former are responsive to reasons as such, in the sense that only the former are capable of reflecting upon the reasons for them to act or believe the way they do, and that enables them to self-determine their beliefs and actions. Thus, the idea of responsiveness to reasons as such amounts to reasoning: “an activity in which someone explicitly considers what to believe and what to do, and takes reasons into account in determining her belief or her action” (p. 130). Apparently, conceptual capacities are such capacities that enable animals like us to reason.

In fact, early since Mind and World, McDowell has articulated his stipulative sense of “concept” in developing his conceptualism:

The way I am exploiting the Kantian idea of spontaneity commits me to a demanding interpretation for words like “concept” and “conceptual”. It is essential to conceptual capacities, in the demanding sense, that they can be exploited in active thinking, thinking that is open to reflection about its own rational credentials. When I say the content of experience is conceptual, that is what I mean by “conceptual”. (1996, p. 47)

At this point, I think it is fair to say that by regarding his conception as stipulative what McDowell means is that “concept” is his technical term in the context. He is aware that by using the word “concept” he does not cash out the cognitive significance of intentionality in the way cognitive scientists do. Instead, the way he uses “concept” is rooted in the Kantian tradition, where he finds his conception best spelt out when associating it with a cluster of closely related concepts in the Kantian framework: rationality, spontaneity, understanding, and freedom.

So, according to the above quote, the word “concept” refers to that which can be exploited in active thinking, thinking that is open to reflection about its rational credentials (i.e., responsiveness to reasons as such). Since we are clear about the distinction between the semantic and the substantive question, it is innocuous, and even necessary for McDowell’s solution to the tension between (EJ) and (SR), to semantically descend and write:

  • A concept is that which can be exploited in active thinking, which is open to reflection about its own rational credentials.

or, more generally,

  • (C) X is conceptual if and only if X is that which can be exploited in active thinking, which is open to reflection about its own rational credentials.

With (C), we are well-placed to examine the substantive question about McDowell’s co-extensiveness thesis. I hope that by clarifying McDowell’s conception of the conceptual, conceptualism does not look as overly demanding as it initially seems. The clarification also helps remove some misguided misgiving about conceptualism. Anyone who agrees that there is some kind of item accessible to the mind which can be exploited in such kind of active thinking and who agrees that the content of experience must be of this kind is in effect committed to conceptualism.

Now, for one direction of (CoX), “conceptuality → rationality,” it follows from (C) that it is true. But that does not mean that its truth is as trivial as “All vixens are female foxes.” For “conceptual” is not a synonym for “rational” as “vixen” is a synonym for “female foxes.” “Conceptuality → rationality” is no more trivial than a mathematical truth such as “Any triangle has as many angles as its sides,” where the number of angles is co-extensive with the number of sides. Both “conceptuality → rationality” and mathematical truths are a priori, established by careful reflections on the relevant concepts and their relationships.

What about the other direction, namely, “rationality → conceptuality”? For the remaining part of this essay, I shall focus on two critical points Cheng makes about it.

3 Conceptualism—an additional thesis?

Cheng briefly mentions a point several times which is made by Peacocke about formulating the Myth of the Given. He seems convinced by the point and cites it as textual evidence against the indispensability of conceptualism. I think the point is worth unpacking. According to Peacocke, a leading non-conceptualist, the core idea shared by those who reject the Myth is this:

[T]here cannot be a state that both justifies you in making a judgment with a certain content, and yet is also a state that lacks representational content. (2008. p. 191)

The Myth therefore is committed to the existence of such a state. Such a state purports to justify beliefs without itself having any representational content. However, apart from the core idea, Peacocke thinks that it is an additional thesis that “all representational content is conceptual content” (Ibid.) Peacocke’s so formulating the Myth seems to shift the burden of proof to the conceptualist. Cheng accepts the point and asks, “What do we need if we want to maintain this additional thesis?” (p.139) The answer, he thinks, is to establish the truth of “rationality→ conceptuality,” namely, that rationality requires conceptual capacities.

Is the conceptualist committed to the additional thesis, which is independent of the “core idea”? Pace Peacocke, I think the conceptualist can reject his formulation of the Myth as inaccurate. Sellars (1956) argues against the Given to the effect that non-conceptual content, if such there be, has no epistemic efficacy. That is because for anything to play a justificatory role, it must have propositional form, yet non-conceptual content does not have propositional form.Footnote 3 So, to say that a non-conceptual content justifies a belief or judgment is incoherent—regardless of whether it is representational. The conceptualist needs not deny that the experiences of animals other than humans have representational contents. What matters is whether the content in question has propositional form in order to qualify for epistemic or discursive activities. So, the conceptualist is not committed to the additional thesis. Instead, conceptualism is committed to this:

  • (J) All representational content which can play a justificatory role is conceptual content.

Meanwhile, conceptual content is propositional if anything is. So, Sellars’ argument against the Given entails (J). Conceptualism enjoys the default position against the Myth. The burden is rather on the non-conceptualist to show how non-conceptual content, despite being non-propositional, can provide justifications. Thus, if conceptualism is not an additional thesis, Cheng’s question, “What do we need if we want to maintain this additional thesis?” does not even get off the ground.Footnote 4

So why does Peacocke think that the conceptualist has to endorse that “additional” thesis? Well, he must have had some candidate non-conceptualist account in mind which he thinks is defensible and at least equally plausible as any conceptualist one. So, what kind of non-conceptual mental item does he regards as standing in rational relations to, providing reasons for, beliefs and judgments?

In general, the non-conceptual can be divided into three sorts:

  1. (1)

    Allegedly existing mental items such as sense-data or qualia;

  2. (2)

    Mental representations at the sub-personal level posited by cognitive science;

  3. (3)

    Ways in which features of external objects are presented in experience.

The prospects of (1) and (2) are dim: (1) is non-representational and (2) is not personal. Rather, Peacocke’s non-conceptualist account (2001) is built upon (3). The idea, generally put, is that the ways, despite being non-conceptual, are representational and hence have correctness conditions. Then, insofar as empirical concepts are concerned, the correctness of conceptual contents depends on the correctness of non-conceptual representational contents because the individuation condition of the former depends on that of the latter.

A full assessment of Peacocke’s account would take us too far afield, but given what we have discussed so far, I have two critical remarks. First, if we suspend non-conceptualism in reading (3), the way so described by (3) seems compatible with the McDowellian definition of concept (C), for it resonates with the idea of demonstrative concept (McDowell, 1996, p. 58). Second, I suspect McDowell may well say that the very idea that non-conceptual representations have correctness conditions presupposes a picture of intentional state attribution that is parasitic on the correctness condition of the conceptual. If so, then either the ways turn out to be conceptual after all, or it is a non sequitur that the ways non-conceptually construed stand in rational relations to beliefs and judgments.

4 Structurality as a substitute?

Cheng concurs with McDowell that “bare presence cannot have rational significance” (2021, p. 139), but he denies that non-conceptual content is nothing but bare presence:

Bare presence’ is used by philosophers to mean, for example, unstructured things; the only thing we can say about them is that they are present. But why should we accept that the conceptual things are the only structural stuffs? To be sure, non-conceptualists need to spell out what kinds of structure are possessed by various non-conceptual contents, and why those structures can provide justification as opposed to exculpation, but this project seems initially feasible, in the sense that it is not impossible in principle. (p. 136)

I would like to invite Cheng to clarify what he means by “structure” here. Sub-personal representations, arguably, are “structured,” e.g., ones at the early stages of visual processing, but they are not in the space of reasons. In addition, it is not clear why being bare presence implies being unstructured. Surely McDowell would say that bare presence is not “conceptually structured” but that would not commit him to that bare presence is not structured in any sense.

Nevertheless, Cheng thinks that there is textual evidence in McDowell’s later writings about his modified account of experience, which replaces propositional content with “intuitional content” that indicates his tacit concession to the idea that the content of experience is non-conceptual and yet structural. For example, McDowell writes,

[E]very aspect of the content of an intuition is present in a form in which it is already suitable to be the content associated with a discursive activity, if it is not — at least not yet — actually so associated. (2008b, p. 264)

Cheng takes this passage, especially the terms “in a form” and “not yet—actually so associated,” as suggesting that the contents of experiences are “not already conceptual, but they are in a form in which it can become conceptual later.” (2021, p. 139, the author’s italics) Cheng’s interpretation seems to be further bolstered by another passage of McDowell’s:

If it is to become the content of a conceptual capacity of hers, she needs to determine it to be the content of a conceptual capacity of hers. That requires her to carve it out from the categorically unified but as yet, in this respect, unarticulated experiential content of which it is an aspect, so that thought can focus on it by itself. (2007, p. 347)

In Cheng’s view, McDowell’s use of the terms “carve it out” is in effect on a par with the idea of conceptualizing the non-conceptual. Meanwhile, he seems to take McDowell’s “in a form” as parallel to his own structural constraint for the rational significance of experience:

[W]e need to insist only that experiential contents are structural, or in a suitable form; the thought that they are conceptual is an additional thesis. (2021, p. 139)

I have raised my doubt about the second half of the quote. Nonetheless, the textual evidence seems to give Cheng an opening for suggesting that “conceptualization, just as such, is innocent” (2021, p. 140), provided that non-conceptual contents, in virtue of being structural, are not bare presence.

For Cheng’s idea of conceptualization to work, he has to flesh out his structural requirement. He seems to gesture toward it (pp. 145–147) by identifying non-conceptual contents with the subject’s phenomenology (for participants cannot report most of the features) and taking conceptualization to be the subject’s attention’s “carving out” the phenomenology (and hence reporting the attended feature). Perhaps phenomenology is structured due to its features present in it, but it is not clear whether that amounts to rational significance. Nor does structurality alone explain why attention can select one feature among others, if attention is a faculty of spontaneity.

In contrast, the limitation of verbal reportability is not a problem for conceptualism, since conceptual capacity is not supposed to be identified with reportability. Just as McDowell distinguishes between “the act of thinking” and “the content of a piece of thinking” (1996, p. 28) so can the conceptualist distinguish between the act of experiencing and the content of a piece of experiencing. Thus, just as limitation is on thinking rather than on thinkable contents, (Ibid.) so is limitation on accessing rather than on accessible contents in experience.

I do not think that at this point the disagreement between Cheng and McDowell is merely verbal. The “conceptual” is meant to capture the sense in which there is a certain component of experience that can be exploited in active thinking that is open to reflection about its rational credentials. Indeed, intuitional content must take some form, but whatever form it takes, it must be in such a way that makes itself exploitable for active thinking and other discursive activities. Discursive activities, which belong in the space of reasons, impose rational constraints on what form intuitional contents can take—how categories are to be unified (McDowell, 2008b, p. 265). In this substantive sense, intuitional content is all conceptual.

Recall that any adequate conception of experience must meet the two desiderata: (EJ) and (SR). The point of insistence on the conceptuality of intuitional content is to secure its place in the logical space of reasons so that not just (EJ) but also (SR) can be met. The latter desideratum allows the faculty of spontaneity to extend all the way out to the content of experience. Meanwhile, it requires that intuitional contents figure in rational activities and hence qualify for the definition (C). Therefore, McDowell’s modified account is still conceptualism.