John McDowell’s variant of naturalism, if discussed at all, is often regarded as not having the true credential of naturalism. Crispin Wright (2002), for example, argues that McDowell’s picture will collapse into rampant platonism, one of McDowell’s critical target, which has it that the mind is autonomous in a way that is entirely independent of the physical reality. One reason why McDowell’s “naturalism” is not taken seriously is that his picture crucially involves several dualisms, and they can seem at odd with standard versions of naturalism. Note that McDowell himself does not use the term “dualism” to label his own views. This term has become negative for a long time, and “dualisms” are often been invoked to name one’s critical targets, for example “scheme-content dualism” (Davidson, 1974, 1989; McDowell, 1999; also cf. his discussions of “dualism of norm and nature” in Wittgenstein’s context, 1996a, p. 93-94). However, it is actually apt to understand McDowell’s picture with a series of dualisms, as I will explain below. In the remainder of this section, I will introduce two such dualisms, and in the next section I will focus on a third one that has been ignored in the Anglo-Saxon literature. The main body of this paper will be from section 3, so the backgrounds provided in the first two sections will be rather minimal.
Let’s begin with the most famous one, the one between the realm of law and the space of reasons. “Law” here means natural laws, i.e., regularities identified by natural sciences. “Reason” is what philosophers are familiar with, i.e., rational relations that are exemplified (at least) by human minds. This is a dualism because McDowell insists that they are sui generis, i.e., different in kind (see also Sellars, 1956). Rational relations are not the kind of relations that are identified by natural sciences. This is at odds with most versions of naturalism, according to which the realm of nature includes only the realm of law. This brings us to the next dualism, between first nature and second nature. First nature is governed by natural laws, so is recognised by everyone as natural. Second nature is governed by rational rules, and whether it is part of nature is a matter of dispute. It is not supernature, but rather is full of semantic meaning and reason-bearing relations. One major difficulty is that it is hard to have a proper non-circular definition of nature. Humans, like other animals, are born with characteristics in first nature. But given our very special and complicated first nature, we can be initiated into second nature governed by rational relations. “Second nature” is a notion borrowed from Aristotle, and McDowell also invokes a German notion “Bildung” to explicate the process of transforming from the first nature to the second nature. One key point is that McDowell does not use any notion of causation to characterise the realm of law and the space of reasons, since he holds that causation appears in both. It is, to be sure, hard to make sense of causation in the space of reasons without substantive accounts of causation (Gaskin, 2006).
The above is an all-too-brief description of McDowell’s dualisms of the realm of law/the space of reasons and first nature/second nature. Again, McDowell himself does not use “dualism” to label his own positions. He does use “relaxed naturalism” and “naturalism of second nature” to name his picture. In the next section, I will explain how a third dualism of McDowell is crucial to understand his naturalism. After that, I will attempt to connect McDowell scholarship to contemporary analytic metaphysics, in a way he himself would not approve.
The ignored dualism is between environment and world from Hans-Georg Gadamer (1960/2004). Gadamer writes that although “the concept of environment was first used for the purely human world… this concept can be used to comprehend all the conditions on which a living creature depends. But it is thus clear that [people], unlike all other living creatures, [have] a ‘world,’ for other creatures do not in the same sense have a relationship to the world, but are, as it were, embedded in their environment” (1960/2004, p. 441). He further, in the same context, relates this openness to the world to human’s possession of languages: “this world is verbal in nature… that language is originally human means at the same time the [people]’s being in the world is primordially linguistic” (ibid., p. 440). Relating this back to the two dualisms introduced in the previous section, environment is first nature, which is governed by natural laws, while world is second nature, which is governed by rational rules.
It has been problematic for the literature to ignore this Gadamarian dualism because in doing so one fails to register the fact that when McDowell speaks of the “world,” he means specifically the Gadamarian world, which is different from the physical world as usually understood. Moreover, unlike the above two dualisms, the one between environment and world raises a specific difficulty for McDowell. In Mind and World, he argues that humans like us have both environment and world. But how is it possible to have been initiated into having the world in view and at the same time still have access to the environment? Would the environment become something like Kantian noumena (Cheng, 2021)? These two points are distinctive, and if one does not consider this dualism, they cannot be seen clearly as distinctive issues. On other occasions I have discussed relevant issues and shall not repeat them here; in what follows I will focus on how contemporary analytic metaphysics can potentially shed light on issues concerning the above dualisms.
Dualisms have been forcefully criticised for understandable reasons, the most obvious one being that if the two realms are different in kind, it is unclear how they can interact at all. For example, how is it possible for the two realms to interact given Descartes’ system? The pineal gland does not seem to be a reasonable solution, since it is itself in one realm. How is it possible for concepts and intuitions to work together given Kant’s system (1781/1787/1998)? Schematism does not seem to be a reasonable solution, even if Kant has tried hard to explain how it has both characteristics. How is it possible for the senses to be grasped by creatures having physical nature given Frege’s system (1892/1930)? The list can go on. Now, McDowell does not call his own distinctions or dichotomies “dualisms,” but that does not mean he would not face such problems. If the realm of law/first nature/environment and the space of reasons/second nature/world are different in kind, how do they interact? How is it possible for creatures living in the realm of law/first nature/environment to begin with to be initiated into the space of reasons/second nature/world? For the ease of exposition, in what follow I shall not mention all three of them at once, but the below discussions will apply to all of them.
Contemporary analytic metaphysics and metametaphysics have provided much conceptual resource in this regard. In the more familiar context, the question is: we all agree that there are levels of analysis, but are there corresponding levels of reality? To answer this question, various “bridges” have been proposed and explicated. In this section I will explore five such candidates, and as we shall see, in some cases it is unclear whether McDowell would and should accept such suggestions.
Let’s begin with reduction. Contemporary discussions of reduction often begin with Ernest Nagel’s model (1949, 1961, 1970), which he describes as follows:
A reduction is effected when the experimental laws of the secondary science (and if it has an adequate theory, its theory as well) are shown to be the logical consequences of the theoretical assumptions (inclusive of the coordinating definitions) of the primary science. (1961, p. 352)
The basic idea is that a theory TR can be reduced to another theory TB if and only if TR is derivable from TB via bridge laws or coordinating definitions. In the past few decades, it has drawn many criticisms (e.g., Feyerabend, 1962, 1966; Churchland, 1986; Bickle, 1998), and many other models have been proposed (e.g., Oppenheim and Putnam, 1958). For our purposes, there is no need to get involved into these controversies. Suffice to see that to reduce the space of reasons to the realm of law, in whatever sense, is completely alien to McDowell’s thinking. More specifically, reduction belongs to what he calls “bald naturalism,” according to which the space of reasons is to be fully explained by the realm of law. Whatever sense of reduction is in place, it cannot capture the distinction between the space of reason and the realm of law as McDowell conceives of it.
What about supervenience? As Brian McLaughlin and Karen Bennett (2018) point out, “Everyone agrees that reduction requires supervenience” (3.3). Given this only, rejecting reduction is not in itself a rejection of supervenience. But does supervenience require reduction? David Chalmers thinks not (1996), and the consensus seems to side with him. Often reduction requires property identity or entailment, and if so, it is too strong for supervenience to require it (also see Kim, 1984, 1990). Therefore, rejecting reduction does not force us to reject supervenience. If the space of reasons supervenes on the realm of law, then every change in the space of reasons is due to changes in the realm of law, according to the major intended meaning of supervenience. It is not too clear whether McDowell would or should accept this, unfortunately. A basic distinction might be helpful: on a weaker understanding, supervenience is simply a covariation relation; on a stronger understanding, supervenience involves determination or fixation. It seems that McDowell has to reject the stronger understanding, as he cannot allow that happenings in the space of reasons are determined or fixed by happenings in the realm of law. It is less clear whether he can allow for the weak understanding, but it seems more reasonable to expect he can, as this weak notion is so weak that if one denies it, that would presumably cut off any connection between the two realms. Therefore, I will tentatively hold that McDowell can allow supervenience that does not involve determination or fixation.
As for (material) constitution, it is often invoked to explain, for example, the relation between a statue and its corresponding lump and form, or the relation between persons and their bodies (Baker, 2000). Does the realm of law constitute the space of reasons? If the latter has its own existence, which is different in kind from the former, then constitution cannot be the right relation. Here we already assume that constitution is not identity. For those who hold constitution is identity, it is even less likely to saddle McDowell with constitution. It will be useful to contrast this with realisation, which often appears in the context of functionalism and functional roles (Putnam, 1967). It is indeed natural to invoke functional roles to understand rational relations, and it also seems natural to think that happenings in the space of reasons are or can be realised by happenings in the realm of law. Realisation is often taken to be incompatible with reduction (Fodor, 1974), which is congenial to McDowell’s picture.
In addition to the above potential relations, there is an increasingly popular one that has to be considered: grounding. If fact A is grounded by fact B, fact A obtains in virtue of fact B’s obtaining. Is it right to say that facts in the space of reasons obtain or hold in virtue of the-realm-of-law fact’s obtaining? It does not sound right within McDowell’s framework. To think otherwise is to commit the naturalistic fallacy: to attempt to derive ought from is (McDowell, 1996a). Another way to see the problem is to consider Kit Fine’s suggestion: according to Fine, if fact A is grounded by fact B, the essence of A would require it to be grounded by B (1994, 2012). But this again does not seem right for McDowell: the essence of reason does not require it to be grounded by natural laws, though as it happens in our world the space of reasons and the realm of law have very close connections. To get clear about these notions is extremely important. Just to give an example, in a different context it has been claimed that “the protective body map grounds the sense of bodily ownership,” and that “[t]he affective quality constitutes the phenomenology of bodily ownership” (de Vignemont, 2018, p. 264). Assuming the sense and the phenomenology mean the same thing here, one wonders about the relations between the protective body map, the affective quality, and the sense or the phenomenology of bodily ownership. In that paper, the author does not explicate these relations, but to get clear about them is of course crucial. Perhaps what is meant is that the phenomenology of bodily ownership is identical to the affective quality, which is grounded in the protective body map. But in any case, these relations need to be taken seriously in philosophical formulations.
This completes our preliminary discussions of potential metaphysical relations connecting the space of reasons and the realm of law. We have seen that while reduction, constitution, and grounding are not suitable, weak supervenience and realisation might be feasible. But this is not the end of the story. In the next section I will discuss yet another metaphysical connection – emergence – and argue that strong emergence is required to cash out the McDowellian picture.
The key to understand McDowell’s dualisms, I submit, is to hold that what crucially bridges the two realms is strong emergence, which might or might not compatible with realisation and supervenience. I will come back to that later, and focus on emergence for now. Put it bluntly, emergence is taken to capture both dependence and autonomy. This is congenial to McDowell’s picture, as he would not wish to reject dependence altogether, but he demands that the space of reasons is autonomous in certain sense. Like many other notions in this area, emergence is often divided into weak and strong. Let’s begin with the weak version. Weak emergence uncontroversially acknowledges the realities of entities and features posited in special sciences such as biology and psychology, while insisting on physicalism. Even just from this, it can already be seen that it is not something McDowell can accept: his relaxed naturalism is no physicalism. Before moving onto strong emergence, let’s review the most prominent problem for weak emergence. In doing so we will understand both weak emergence and strong emergence more thoroughly. According to the standard formulation, weak emergence accepts the following five theses (O’Connor, 2020):
Jaegwon Kim (1993, 1998) famously argues that the above five points jointly entail a thesis that is unacceptable:
Overdetermination: Emergent effects are generally causally overdetermined by distinct individually sufficient synchronic causes
Weak emergentists would insist that the Distinctiveness in question is so weak that it does not generate overdetermination, while Kim himself rejects Distinctiveness altogether (4) and opts for reductionism. Eliminativism rejects Reality (2), epiphenomenalism rejects Efficacy (3), substance dualism rejects Supervenient Dependence (1), and strong emergentism rejects Physical Causal Closure (5). This brings us to what fits McDowell’s picture better: strong emergentism.
Strong emergentism “[maintains] that at least some higher-level phenomena exhibit a weaker dependence/stronger autonomy than weak emergence permits. This often takes the form of rejecting physical realisation, affirming fundamental higher-level causal powers, or both” (ibid.). More specifically, it affirms Supervenient Dependence, Reality, Efficacy, and Distinctiveness, which fits well with McDowell’s picture: according to that picture, happenings in the space of reasons weakly supervene on happenings in the realm of law, though whether that involves necessitation is a matter of dispute. Happenings in the space of reasons are of course real, causally efficacious, and distinct. Given that it is non-physicalist naturalism, it is sensible for it to hold that the picture violates Physical Causal Closure.
Things are complicated here, however. For McDowell, second nature does have the kind of autonomy strong emergence affirms, but it is unclear whether he rejects physical realisation. One possibility is that he does not reject realisation per se, but he rejects that realisation has to be physical. Moreover, since he affirms causation in the space of reasons, he should affirm fundamental higher-level causal powers (Gaskin, 2006). These are all thorny issues. On top of that, strong emergence in general faces three general problems (O’Conner, 2020):
Incoherence or inexplicability (Bennett, 2017)
This objection concerns whether the reality has a fundamental level, and if so, how it is possible to have a level that does not depend on any other level. This is, to be sure, a very difficult problem, and there is no obvious resource in McDowell’s system. However, this seems to be a general problem for not only strong emergentists but also for those who believe in fundamentality.
Anti-naturalism or evidential paucity (McLaughlin, 1992)
This is the common worry that strong emergence is incompatible with standard naturalism cum physicalism. This is not a big problem, though, as we already know that McDowell’s picture is not standard naturalism cum physicalism.
Collapsing (Kim, 1998)
This is about whether the so-called “novel powers” are already possessed by lower levels. Here McDowell would insist that notions of higher-level phenomena are necessary for psychological explanations. This does not by itself solve the problem, but at least McDowell is again not alone here. Daniel Dennett’s intentional stance (1987), for example, also insists on this point (also see List, 2019, in the context of free will).
Now, it should be admitted that strong emergentism is difficult to defend, as we just saw. However, it still has some prominent contemporary defenders (O’Connor, 2000; Wilson, 2021). Given the increasing difficulties faced by physicalism, other theories such as idealism, panpsychism, and emergentism have come back. These controversies will not be solved anytime soon, and strong emergentism should be at least regarded as a potential option.
I will end this paper with the following conservative moral: it seems clear that McDowell’s dualisms – the realm of law/the space of reasons, first nature/second nature, and environment/world – requires strong emergence, given the above considerations. For some, this can be a reductio ad absurdum for McDowell’s picture. This paper does not set out to defend strong emergence. However, it is significant enough to clarify that McDowell’s picture does require strong emergence, for McDowell’s Wittgensteinian quietism refuses to engage any constructive philosophy, and therefore would not engage any explicit discussion concerning reduction, supervenience, constitution, realisation, grounding, and emergence. Making this commitment explicit is a crucial step towards connecting McDowell’s picture back to the purview of analytic metaphysics. This is in conflict with his own metaphilosophy, but that is quite another matter.
This branch of analytic metaphysics is called “interlevel metaphysics,” which studies questions such as whether reality has a leveled or layered structure, and what relations underpin the structure. Note that even if reality has no such leveled structure, this branch of metaphysics is still useful, as identity and reduction are relations we wish to understand as well. In addition to the candidate relations discussed above, other relations include mereological composition, the determinable-determinate relation, causal mechanism, and so on. This is a massive and messy area, and it is difficult for anyone to master many materials in it. To understand the natures of these relations and see how they can help us cash out McDowell’s relaxed naturalism would be a daunting task, but it is a worthwhile project to pursue, and I hope more will devote themselves to this task in the future.
 Peter Godfrey-Smith (2010) has a nice discussion of why McDowell thinks it is a bad idea to have certain dualisms and then try to bridge the two sides. This is related to his quietism, which I criticise in Cheng (2021).
 Some of these discussions benefit from a seminar led by Duen-Min Deng. The mistakes, if there is any, are all mine.
 Things are really complicated here. For the relations between weak emergence, strong emergence, and grounding, see Leuenberger (2020).