Grounding, Physicalism, and the Explanatory Gap

Working draft. Please do not quote or cite without permission. Comments very welcome. Title removed for blind review Zach Blaesi The University of Texas at Austin zachblaesi.com Abstract Recently, it has been suggested that the notion of (metaphysical) ground has an important role to play in developing physicalism about mentality. For there are reasons to think that Grounding Physicalism About Mentality (GPM) has advantages over traditional reductive and non-reductive versions of physicalism about mentality. In this paper, I argue that a new spin on an old objection to physicalism—that it leaves an ‘explanatory gap’—undermines the enthusiasm for GPM. I start by arguing that truths about the essences of things have an important role to play in explaining certain grounding phenomena. I then argue that this ultimately creates a dilemma for GPM: either GPM leaves a distinctive explanatory gap, or it collapses into a version of reductive physicalism. Keywords: explanatory gap; physicalism; mind–body problem; grounding; essence 1 Introduction Contemporary metaphysics is marked by a revived interest in the notion of (metaphysical) ground.1 Recently, a number of grounding theorists have suggested that ground is exactly the notion needed to successfully develop physicalism about mentality—the view that the mental is ‘nothing over and above’ the physical.2 It is also frequently announced that the notion of ground is needed to formulate (or state the minimal commitments of) physicalism. But even if the latter idea is a mistake, there are still reasons to think that the notion of ground can be used to develop a version of physicalism about mentality—call it ‘Grounding Physicalism About Mentality’ (‘GPM’ for short)—that is superior to other well-explored physicalist theories of mind.3 In short, the appeal of GPM is that it promises to occupy a middle position between certain reductive and non-reductive versions of physicalism about mentality. On the face of it, GPM is less demanding than reductive versions of physicalism, because it does not entail that mental phenomena are reducible to physical or functional phenomena. On the other hand, unlike some 1 See Fine 2012 and Rosen 2010 for introductions to the notion and discussions of its importance to metaphysics. 2 See, e.g., Dasgupta 2014, deRosset 2013, and Rosen 2010. 3 Understood in this way, GPM needn’t be a view that all physicalists must accept in order to count as physicalists. As Lewis [1983] tells us, physicalism ‘is not to be identified with any one . . . theory of mind. It is a thesis that motivates a variety of theories of mind: versions of Behaviourism, Functionalism, the mind–body identity theory, even the theory that mind is all a mistake’ (361). 2 Z. Blaesi versions of non-reductive physicalism, it purports to adequately capture the idea that the mental depends on (and is explained by) the physical. My aim here is largely negative. I argue that a new spin on an old objection to physicalism—that it leaves an ‘explanatory gap’ [Levine 1983]—undermines the initial appeal of GPM. While the explanatory gap problem has been heavily discussed in the philosophy of mind, many of these discussions construe physicalism about mentality as an identity thesis. By contrast, I take GPM as my target and show that—even though ground is an explanatory notion—GPM confronts at least as much of an explanatory gap problem as any other version of physicalism. I then argue that this creates a dilemma for the Grounding Physicalist About Mentality (‘GPMist’ for short): either the view is incompatible with a popular response to the explanatory gap problem, or it ultimately collapses into a version of reductive physicalism. While the first horn of this dilemma puts GPM at a serious disadvantage to reductive versions of physicalism, the second undermines one of the central motivations for the view. As a preview, consider the claim that some experience of pain is grounded in a physical event. If that claim is true, then it is plausible that it is necessary that if a physical event of that type occurs, then it grounds the experience of pain. But what explains that grounding pattern? I argue that this question is to be answered by appeal to ‘essentialist truths’ about the grounded. To the extent that the essence of pain leaves the question unanswered, GPM leaves an explanatory gap. To bridge this gap, we need a better understanding of the essence of pain. One way to do this is to provide a (partial or full) ‘real definition’ of pain—to state (at least in part) what it is for an individual to be in pain. Yet, if our definition is a success, we end up with a version of physicalism that has no clear advantage over the traditional reductive alternatives. The paper proceeds as follows. In §2, I motivate GPM by discussing the shortcomings of some traditional versions of physicalism about mentality and then providing reasons to think that GPM avoids these shortcomings. In §3, I argue that the grounding patterns involving consciousness are to be explained by essentialist truths about consciousness, even if those truths do not ground any of those grounding patterns. In §4, I discuss some reasons for thinking that GPM leaves an explanatory gap. Finally, in §5, I argue that a popular response to the explanatory gap problem is unavailable to the GPMist, and I establish my dilemma. 2 2.1 The Shift to Grounding Physicalism About Mentality From Reductive Physicalism to Non-Reductive Physicalism In this paper, I understand ‘reductive physicalism’ broadly to include any view that is committed to reducing mental phenomena to phenomena that are definable in wholly non-mental terms. I use the phrase ‘reducible to’ as a placeholder for both identity and real definition [see Rosen 2015]. Historically, certain versions of functionalism were considered to be non-reductive, because they were thought to avoid a commitment to identifying mental states with lower-level physical states. In my terminology, however, the mind–brain identity theory, functionalism, and certain Title removed for blind review 3 versions of representationalism are reductive theories, because they are committed to providing biconditional analyses of mental states in wholly non-mental terms. I also assume for the sake of argument that extant versions of reductive physicalism fail. Some of these views face the Multiple Realizability Problem: they are unable to accommodate the apparently empirical datum that mental states are ‘realized’ in sufficiently complex systems of varying physical makeup and in the same systems as a result of different physical processes at different times. But they all appear to face the more general problem of ultimately ‘leaving out’ the most important part of consciousness: the ‘what-it’s-likeness’ of experience. Call this the Elimination Worry. The Elimination Worry is an important motivation for developing a thoroughgoing non-reductive version of physicalism, one that is not committed to reducing mental states to states that are definable in wholly nonmental terms. The task is then to render mental states physicalistically acceptable without reducing them to physical or functional states. One extensively discussed approach to this task appeals to supervenience. Supervenience-based non-reductive physicalism about mentality (‘SNRPM’ for short) consists of two central claims. The first is that the mental supervenes on the physical. This claim can be spelled out in a number of different ways, but for illustrative purposes, we can understand it as the global supervenience thesis that any metaphysically possible world that is a physical duplicate of the actual world is a mental duplicate of the actual world. The second claim is that mental phenomena are not reducible to phenomena that can be defined in wholly non-mental terms. There are two major problems with SNRPM. First, supervenience is far too weak of a notion to guarantee that the mental is ‘nothing over and above’ the physical. For example, as many philosophers have noted, both emergentists and physicalists about mentality can agree that the mental supervenes on the physical. They nonetheless disagree as to whether (or how) mental phenomena depend on (or are explained by) physical phenomena. Since supervenience does not adjudicate this dispute, it is not sufficient for nothing-over-and-aboveness. Call this the Explanatory/Dependence Problem.4 Second, the supervenience of the mental on the physical cries out for explanation. If a version of physicalism lacks the resources to provide such an explanation, then that counts significantly against the view. As Kim [1989] puts it, ‘If the global supervenience of the mental on the physical were to be proposed as an unexplainable fact that we must accept on faith, I doubt that we need to take the proposal seriously’ (42). This is not a problem for the reductive physicalist, who is in a position to explain the supervenience of the mental on the physical in terms of the reducibility of the mental to the physical/functional. By contrast, SNRPM is committed to brute modal correlations of simple mental properties with complex physical properties. These correlations are ‘modal danglers’ that count against the view for the same reason that brute psychophysical laws count against dualism. Call this the Dangler Worry. 4 This is a problem not only for SNRPM (construed as a theory of mind) but also for attempts to state the minimal commitments of physicalism in purely modal terms (so-called ‘supervenience formulations’ of physicalism). 4 Z. Blaesi While I have understood SNRPM in terms of global supervenience, these points generalize to other supervenience theses as well. For these reasons, a number of philosophers reject SNRPM for some version of reductionism. The problem, as Horgan [1993] puts it, is that reductive versions of physicalism ‘usually end up susceptible to counterexamples of one sort or another’ (579). What we need, Horgan suggests, is a way of ‘construing higher-order properties which (i) do not provide reductive sufficient conditions, but nevertheless (ii) render the physical supervenience of these properties materialistically explainable anyway’, thus making room ‘for the higher-order properties as part of the physical world’ (580). I suspect that the recent interest in metaphysical ground stems in part from the hope that it can help us to do exactly that. I make no claim that SNRPM is the only viable version of non-reductive physicalism. There are alternatives, such as constitution versions, determinable–determinate versions, and powers-based subset versions.5 That being said, that there is no consensus on these other versions of physicalism is a reason to explore GPM as a new alternative. So, for my dialectical purposes, I will simply set these other views aside. 2.2 From Superdupervenience to Grounding To make GPM explicit, I assume that grounding is a multigrade relation between facts, and I follow Rosen [2010] in understanding facts to be ‘structured entities built up from worldly items—objects, relations, connectives, quantifiers, etc.—in roughly the sense in which sentences are built up from words’ (114).6 I use ‘[P]’ as a function term for the fact that P, which obtains just in case P. When [P1 ], [P2 ], . . . ∈ ∆, I write ‘∆ < [Q]’ for ‘[P1 ], [P2 ], . . . ground [Q]’. Ground is first and foremost an explanatory notion: if ∆ < [Q], then the facts in ∆ metaphysically explain [Q]. Grounding is also a relation of relative fundamentality: if ∆ < [Q], then the facts in ∆ are more fundamental than [Q]. It is necessitating: if ∆ < [Q], then it is metaphysically necessary that if the facts in ∆ obtain, then [Q] obtains. And it is factive: if ∆ < [Q], then the facts in ∆ and [Q] exist and obtain. While the primary notion of ground is full, we can define a notion of partial ground (≺) in terms of it: [P] ≺ [Q] just in case [P] fully grounds [Q] on its own or in combination with other facts. Partial grounding is a strict partial order, and full grounding can be weakened to partial grounding by subsumption: from ∆, [P] < [Q], it may be inferred that [P] ≺ [Q]. With these qualifications in place, I understand GPM to be the following thesis: Grounding Physicalism About Mentality (GPM): The obtaining mental facts are fully grounded in the fundamental physical facts. By ‘fully grounded’, I mean that there is a chain of ground leading up from the fundamental physical facts to the mental facts such that there is no stricter or fuller 5 See, e.g., Pereboom 2011, Yablo 1992, and Wilson 1999, respectively. the sake of readability, I occasionally write as if properties, states, and events are relata of the grounding relation; on such occasions, the reader should understand my claims in terms of facts involving those entities. 6 For Title removed for blind review 5 account of why the mental facts obtain. In other words, the fundamental physical facts might fully mediately ground the mental facts by grounding various intermediate facts (such as chemical, biological, or neurological facts).7 Given that full grounding is necessitating, GPM entails that for any obtaining mental fact m, there is some collection of physical facts ∆ such that it is metaphysically necessary that if every fact in ∆ obtains, then m obtains. As a result, GPM entails the global supervenience of the mental on the physical. For if the physical facts metaphysically necessitate the mental facts, then any metaphysically possible world that is a physical duplicate of the actual world is a mental duplicate of the actual world. So, if GPM is true, the supervenience of the mental on the physical is explained by the the nature of ground. Hence, on the face of it, GPM avoids the Dangler Worry. However, GPM does not entail necessitation in the other direction [cf. deRosset 2013, 6]. It is consistent with GPM that the same mental properties are grounded in different physical properties. For suppose [Fa] grounds [Ga]. It follows that □(Fa → Ga) but not that □(Ga → Fa). For all we have said, [Ga] might obtain without [Fa] obtaining. It is for this reason that GPM avoids the Multiple Realizability Problem that confronts certain versions of reductive physicalism, such as the mind–brain identity theory. GPM also appears to rule out emergentism. For one, since ground is an explanatory notion, GPM entails that the mental facts are ultimately explained by the fundamental physical facts. Moreover, given that grounding is a relation of relative fundamentality, GPM entails that the mental is non-fundamental [cf. Dasgupta 2014, 563]. If emergentism is the view that the mental is a fundamental, physically unexplainable phenomenon in the sense that it is not metaphysically grounded in the physical, then GPM avoids the Explanatory/Dependence Problem. Last but not least, GPM minimizes the Elimination Worry, the charge that physicalism about mentality leaves out the most important part of consciousness. For given the factivity of ground, GPM entails that certain facts about phenomenal consciousness both exist and obtain; and since partial grounding is irreflexive, it also entails that the obtaining facts about phenomenal consciousness are distinct from their fundamental physical grounds. However, GPM does not entail that phenomenal properties are reducible to physical/functional properties. It is consistent with GPM, for example, that the essence of pain is exhausted by its ‘what-it’s-likeness’. This completes my sketch of GPM. I now turn to the main aim of the paper: to argue that the grounding patterns involving consciousness are to be explained by truths about the essence of consciousness and then to show why this creates a dilemma for GPM. 3 Grounding and Essence As I have characterized it, GPM incurs a commitment to grounding facts: facts about what grounds what. As a result, the GPMist confronts the difficult question 7 The immediate grounds for a fact are those that ‘need not be seen to be mediated’ and that are to be located ‘at the next lower level’ [Fine 2012, 51–52]. Full ground is mediate in that some relationships of immediate ground can be chained together. 6 Z. Blaesi of whether the grounding facts involving mentality are themselves grounded in physical facts. Bennett [2011] and deRosset [2013] endorse a principle from which it follows (together with GPM) that they are. Schematically: Collapse: If ∆ < [Q], then ∆ < [∆ < [Q]]. So, for example, if the fact that Jones’ c-fibers are firing ([P]) grounds the fact that Jones is in pain ([Q]), then [P] also grounds [[P] < [Q]]. Dasgupta [2014] provides some important reasons to reject Collapse and suggests that grounding facts are partially grounded in ‘essentialist truths’. The essentialist truths with respect to a given entity specify (in part) what it is to be that entity. Following Fine [1995], we can take the essentialist truths with respect to x to be formed from an operator □x , which can be pronounced as ‘it lies in the essence (or nature) of x that’. Consider {Socrates}, for example. It lies in the essence of {Socrates} to contain Socrates as its sole member; that’s just part of what {Socrates} is. While I am sympathetic to the view that grounding facts are partially grounded in essentialist truths, I will simply assume for the sake of argument that Collapse is true. Instead, I will focus on a related grounding phenomenon and argue that, even if Collapse is true, essentialist truths still have an important explanatory role to play—and it is this explanatory role that ultimately spells trouble for GPM. As a result, while some of my arguments are inspired by Dasgupta [2014], they do not depend on the controversial claim that grounding facts are partially grounded in essentialist truths. Suppose the fact that Jones is in pain is grounded in the fact that his c-fibers are firing. It is then plausible to think that if another person’s c-fibers were firing, then that person, too, would experience pain in virtue of c-fiber firing. And so on for any other (possible) individual. This suggests that there is a grounding pattern involving c-fiber firing and pain.8 Drawing from Fine [2012], we can express this idea precisely. Suppose that [P] grounds [Q]. [P] and [Q] will have as constituents certain items a1 , a2 , . . . and b1 , b2 , . . . , respectively. Using [P(a1 , a2 , . . . )] and [Q(b1 , b2 , . . . )] to make this explicit, we can generalize away from the particular grounding connection to: Grounding Pattern: Necessarily, for any x1 , x2 , . . . and y1 , y2 , . . . , if φ(x1 , x2 , . . . ), ψ(y1 , y2 , . . . ), and P(x1 , x2 , . . . ), then P(x1 , x2 , . . . ) < Q(y1 , y2 , . . . ), where φ(x1 , x2 , . . . ) and ψ(y1 , y2 , . . . ) are conditions that in fact hold of a1 , a2 , . . . and b1 , b2 , . . . , respectively. To put the idea loosely, any time some physical event grounds an experience of pain, say, that event has some property such that, necessarily, every event that has that property grounds the experience of pain. This principle may not be completely obvious, but it is easy to multiply examples that support it. Thus, if the letter is scarlet, then its being scarlet grounds its being red. But that’s true of any letter, and 8 I use the example of c-fiber firing throughout for illustrative purposes. The important point is that if the fact that Jones is in pain is really grounded in some physical facts (whatever they may be), then there is a grounding pattern involving facts of those kind and facts involving the instantiation of pain. Title removed for blind review 7 if the letter had been crimson instead, its being crimson would have also grounded its being red. Crimson and scarlet share something in common with all determinate shades of red: it is necessary that they ground the instantiation of red whenever they are instantiated. The central question to be explored here is what explains grounding patterns in general and the putative grounding patterns involving consciousness in particular. One option is that these patterns are grounded in and thus explained by their instances. The advocate of Collapse might then argue that the grounding patterns do not pose some further problem. For if the grounding patterns are to be explained by the particular grounding facts, then the real mystery is why each particular grounding connection holds. If only we could explain every grounding fact, we would have a complete explanation of the grounding patterns. No further mystery would remain. There are two problems with this proposal. In the first place, each of the universal generalizations in question falls within the scope of a modal operator, and while there is no consensus on what grounds necessitated truths, no one says that necessitated universal generalizations are even partially grounded in their instances. So, it’s simply false that we can explain the grounding patterns involving consciousness by explaining (in piecemeal fashion) why each particular grounding connection holds. But for the sake of argument, let’s set the modal operators aside. It’s true that if each universal generalization is grounded in its instances, then the grounding patterns admit of a grounding explanation. But that isn’t the sort of explanation we were looking for. As Dasgupta [2014] puts it, ‘We want to know why all those instances turned out alike—just repeating the instances is no answer’ (570). Could it be that the grounding patterns are to be explained by the essence of some property that the grounds all share? In general, this isn’t the case. It’s true that for any entity, if it exists, then its existence grounds the existence of its singleton. From numbers to people, this pattern holds. But there is nothing common to numbers and people that could explain the grounding pattern involving sets and their members. In addition, as Rosen [2010] points out, if the grounding pattern involving pain and c-fiber firing is to be explained by the essence of c-fiber firing, then ‘the analgesic neuroscientist who knew everything about the detailed physiology of c-fibers and their role in the functional economy of the organism but who knew nothing about pain would have an incomplete understanding of what it is for a c-fiber to fire’ (133). But that seems implausible: ‘[I]t is hard to see why [the neuroscientist’s] understanding of the essence or definition of this particular neurological kind should be defective’ (133). Perhaps the grounding patterns are to be explained by the essence of some property the grounds all share together with the essence of ground [cf. Fine 2012, 77]. Such a view seems to imply that there exists an extremely complicated truth of the form □ground (P ∧ Q ∧ . . .), which somehow mentions each and every grounding pattern involving consciousness. There are two major problems with this view. First, it just doesn’t seem to be in the essence of ground that the grounding patterns hold: if one is ignorant of a single grounding pattern, one is not thereby ignorant of 8 Z. Blaesi what ground is. That is why the colour-blind metaphysician might fully grasp the essence of ground despite knowing nothing of the grounding patterns involving the colours. Second, on this view, if at least one grounding pattern is knowable only a posteriori, then it is impossible to grasp the full essence of ground a priori. This suggests that the epistemology of ground is perversely unlike the epistemology of other established metaphysical tools. It might be suggested that the essence of ground does not mention the grounding patterns by name but instead entails them. This might well be true. Perhaps it is essential to ground that there be grounding patterns if anything grounds anything at all. That might explain why there are grounding patterns involving pain and c-fiber firing on the assumption that the instantiation of c-fiber firing in fact grounds the instantiation of pain. But it doesn’t explain why the instantiation of c-fiber firing grounds the instantiation of pain in the first place, and that’s precisely what we were after. My claim is that the grounding patterns involving consciousness are to be explained by the essence of consciousness. This follows from a general picture of ground: that grounding patterns are always to be explained by the essences of grounded items. As Fine [2012] puts it, ‘It is the fact to be grounded that “points” to its grounds and not the grounds that point to what they may ground’ (76). Why does the existence of a plurality of objects ground the existence of the set of those objects? Because it lies in the essence of a set that if its members exist, then the existence of those members grounds the existence of the set of those members. Why does the instantiation of a determinate shade of red ground the instantiation of red? Because it lies in the essence of red that if some object is a determinate shade of red, then its being that shade of red grounds its being red. It is easy to multiply examples. At this point, it might be objected that the grounding patterns involving consciousness are to be explained by ‘metaphysical laws’, which, contrary to Rosen [2017], are not essentialist truths. There may be some plausibility to the idea that the grounding pattern involving red and the determinate shades of red is to be explained by a determinable–determinate law; the grounding pattern involving sets and their members by a set-formation law; and so on. So why not posit some psychophysical grounding laws as well? While there are a number of reasons to be suspicious of this proposal, the decisive reason to reject it in the present context is that it is dialectically inimical to GPM. It forces the GPMist to substitute brute metaphysical laws for brute supervenience relations. Like the psychophysical laws of dualism, these ‘metaphysical danglers’ cry out for explanation and yet have none. The essentialist picture is not obviously subject to this same criticism. Essentialist truths specify what it is to be something. Like identities, they are ‘autonomous’ in that they are not apt to be explained in the first place [see Dasgupta 2014, §7]. If the essentialist picture is right—and the shortcomings of the alternatives suggest that it is—then the grounding patterns involving pain are to be explained by the essence of pain. The arguments for this picture also show why the grounding facts cause trouble for GPM. For the essence of c-fiber firing no more explains the particular instances of the grounding pattern involving c-fiber firing and pain than Title removed for blind review 9 it explains the grounding pattern itself. It is of crucial importance to note that these explanations need not be grounding explanations. That is, essentialist truths might explain grounding facts and patterns without themselves grounding them. Indeed, according to Lowe [2012], ‘[W]e should regard essence-based explanation just as one more distinctive species of explanation’ (938). As a result, even if the GPMist endorses Collapse, there is still significant pressure on the GPMist to accept the claim that the grounding patterns involving consciousness are to be explained by the essence of consciousness. This is the first step to establishing my dilemma for GPM. 4 The Explanatory Gap In the previous section, I argued that the grounding patterns involving consciousness are to be explained by truths about the essence of consciousness. I will now argue that some of the usual reasons for thinking that physicalism leaves an explanatory gap are also reasons for thinking that the essence of consciousness does not explain the grounding patterns involving consciousness. To the extent that the essence of consciousness does not explain those grounding patterns, GPM lacks an explanation of those patterns. As a result, GPM confronts at least as much of an explanatory gap problem as any other version of physicalism. This may come as a surprise, because one might have expected GPM to bridge explanatory gaps, not open them up. Of course, GPM may bridge certain explanatory gaps. For if a given conscious experience really is fully grounded in a physical event, then that conscious experience is fully metaphysically explained by that physical event. However, this is consistent with there being no explanation for why physical events give rise to certain conscious experiences rather than others. Consider: (2) □∀x(x’s c-fibers are firing → ([x’s c-fibers are firing] < [x is in pain])). If the essence of pain does not rule out the possibility of there existing an x such that x’s c-fibers are firing and yet x’s c-fiber firing fails to ground x’s being in pain, then even if experiences of pain are metaphysically explained by physical events, GPM leaves an explanatory gap in that it does not explain why having one’s cfibers fire should feel the way that it does rather than some other way or no way at all [cf. Levine 1983, 358]. So why think that the essence of consciousness fails to explain the putative grounding patterns involving it? First, suppose that c-fiber firing is one possible ground for pain. In order to provide an essentialist explanation of (2), it seems that the GPMist must maintain that: 10 Z. Blaesi (3) □pain ∀x(x’s c-fibers are firing → ([x’s c-fibers are firing] < [x is in pain])).9 One problem with (3) is that it is conceivable that there be a system that is identical to an ordinary conscious human being in all physical respects but that differs with respect to its phenomenal experiences. For example, when this system’s c-fibers fire, it might feel pleasure rather than pain. Alternatively, the system might be a complete zombie: functionally and behaviourally indiscernible from an ordinary human being but completely unconscious. The claim that conceivability entails metaphysical possibility is controversial, so zombies might be metaphysically impossible. But it is plausible that the conceivability of zombies at least provides defeasible evidence for thinking that statements such as (3) are false. Second, a number of philosophers are sympathetic to a ‘Revelation’ thesis with respect to phenomenal experience. As Goff [2017] presents Revelation, in experiencing pain, say, the experience is directly presented to one in such a way that one is an epistemic position to know all of the essentialist truths about pain and to know with something close to certainty that oneself is really in pain. Goff argues that Revelation provides the best explanation of the putative fact that certain introspective judgments about our phenomenal experiences are super-justified in that the degree of rational confidence we are entitled to hold in them is comparable to the degree of rational confidence we are entitled to hold in our judgments about the basic truths of mathematics and logic (and significantly higher than the degree of rational confidence we are entitled to hold in our perceptual judgments about the external world). Suppose, for example, that someone presents me with two colour chips, one red and the other orange. Through introspective reflection alone, I can form the judgment that what it’s like for me to experience this (the red chip) is to have an experience like that (demonstrating an aspect of my experience). I can also form the judgment that what it’s like for me to experience this (the red chip) is similar to what it’s like for me to experience that (the orange chip). Moreover, I am entitled to a very high degree of rational confidence in these judgments. I can rationally doubt that the chips are similar. After all, I can rationally doubt that the chips even exist—perhaps I am dreaming or in the Matrix. But without some special defeater, I cannot rationally doubt that my experiences of the chips are similar. In this respect, I am in a unique epistemic situation vis-à-vis my phenomenal experiences. Revelation predicts that I am in such an epistemic situation, and to the extent that it reflects how things seem to be upon reflection, it provides a good explanation of my epistemic situation. Unless the GPMist can provide a better explanation of our unique epistemic situation vis-à-vis our phenomenal experiences, we have a strong reason to accept Revelation (or so the Revelation argument goes). The problem for GPM is that, if Revelation is correct, one should be in an epistemic position to know (3) primarily on the basis of painful experiences. But 9 It may be that (3) specifies only the mediate essence of pain. For intuitively, the immediate essence of a thing includes only what has a direct bearing on that thing. The mediate essence of a thing, by contrast, is subject to chaining and includes the essences of those things that figure into its immediate essence. For discussion of this distinction, see Fine 1995. Title removed for blind review 11 one is in no such position. If you are skeptical, then the next time you stub your bare toe on a piece of furniture, just pay extra attention to the unpleasant experience that you undergo and ask yourself how in the world you are supposed to get from that to anything like (3). In light of these problems, the GPMist might fall back on the seemingly modest claim that it lies in the essence of pain that it should have some physical ground. Fine [2012] is sympathetic to this claim (at least when ‘ground’ is understood as ‘natural ground’ as opposed to ‘metaphysical ground’). However, even if the modest claim is true, it doesn’t explain why it is necessary that if the putative physical grounds for pain obtain, they ground pain. Consider an analogous metaethical claim about rightness: it is essential to rightness that it should have some descriptive ground. This may explain why it is necessary that for any act, if that act is right, then it has some descriptive right-making feature(s). But it clearly doesn’t explain why, if act utilitarianism is correct, it is necessary that utility maximization is a right-making feature. At this point, it is worth emphasizing that I am not endorsing the conceivability or Revelation arguments here. My point is simply that if these arguments succeed, then they support the conclusion that GPM leaves an explanatory gap. It falls outside the scope of this paper to motivate these arguments in any detail. My aim for this section has been simply to show that GPM confronts at least as much of an explanatory gap problem as any other version of physicalism. This is the second step to establishing my dilemma for GPM. For in the next section, I argue that, to the extent that GPM has an explanatory gap problem at all, it has more of a problem than reductive versions of physicalism. 5 Closing the Gap If GPM leaves a prima facie explanatory gap, can the GPMist find some way to close it? I think so—but not without abandoning one of the central motivations for the view. However, in order to appreciate this point, it is first important to appreciate why GPM’s prima facie explanatory gap is distinctive. This requires looking at a popular response to the explanatory gap problem in the philosophy of mind. The popular response to the claim that physicalism about mentality leaves an explanatory gap is to argue that if physicalism is true, then nothing has been left unexplained and therefore there is no explanatory gap. For example, Tye [1999] tells us: Take the referent of the term “Q” and the referent of the term “this feeling”—conceive of those referents as you will—why is the former the same as the latter? If this is how the question is understood, then there is no significant question here for the physicalist. Only one state exists, conceived of in two ways, and that state must be self-identical. On this interpretation, then, there is no need for an answer and no explanatory gap. (717) Similarly, according to Papineau [2002]: 12 Z. Blaesi A mind–brain identity simply says of something that it is itself. . . . I say that once you really accept that pain, say, really is some material M, then you will see that this requires no more explanation than does Mark Twain = Samuel Clemens. Identities need no explaining. (150; emphasis added) Rather, the so-called ‘explanatory gap’ is a ‘cognitive illusion’ (Tye’s phrase), the remnant of an ‘intuition of distinctness’ (Papineau’s phrase). Notice, however, that this response to the explanatory gap problem presupposes an identity theory. The central claim is that it makes no sense to explain why a particular thing is identical to itself. By contrast, the GPMist denies that phenomenal properties are identical to physical/functional properties. The problem that the GPMist confronts is that it does make sense to explain the grounding patterns involving consciousness. Grounding patterns, unlike identities, are apt to be explained. Therefore, if the essence of consciousness fails to explain the putative grounding patterns involving it, then GPM leaves a genuinely explanatory gap, not just a mere intuition of distinctness. The popular response does not straightforwardly apply. Here’s another way of putting the point. Reductive versions of physicalism confront conceivability and Revelation arguments. Contrary to Levine [1983], however, they don’t face the further challenge of explaining why certain physical states should feel the way that they do. As a result, it is misleading to claim that reductive versions of physicalism have an ‘explanatory gap problem’, because this claim suggests that these views face a third problem in addition to the first two. By contrast, if GPM faces at least one of the first two problems, then it also faces a third. The GPMist may owe us an explanation not only of an intuition of distinctness with respect to the physical and the mental but also of the putative grounding patterns involving the physical and the mental. The GPMist’s only hope of closing this prima facie explanatory gap is to say something about the essence of consciousness that accounts for the grounding patterns involving it. This does not automatically commit the GPMist to giving a reductive account of consciousness. There are a number of non-reductive positions that arguably incur essentialist commitments. The colour primitivist might deny that there is a full real definition of red in terms of the determinate shades of red but maintain that it is essential to red that if some object is a determinate shade of red, then its being that shade of red grounds its being red. Thus, it is open to the GPMist to claim that the essence of pain lists out each of its possible grounds, even though there is no biconditional analysis of pain in terms of those possible grounds. There is still a big difference between these positions. The colour primitivist is not committed to giving even a partial analysis of red in non-colour terms. By contrast, the GPMist must say that the essence of pain mentions properties of a radically different kind. We might call such a position partially reductive.10 However, for epistemological reasons, there is a good case to be made that the GPMist is ultimately committed to providing a full reductive account of con10 That being said, as Rosen [2015] observes, ‘[I]t will be very hard to distinguish the [partially reductive view] from a [fully] reductive view that identifies [pain] with the disjunctive property λx (Φ1 (x) ∨ Φ2 (x) ∨ . . .), where the Φi ’s are the various [possible grounds for pain]’ (206, fn. 22). Title removed for blind review 13 sciousness. For even if the GPMist is unable to close the prima facie explanatory gap now, we should at least expect there to be a model for how the project is to be carried out in principle. So, the important question is whether the best model available to the GPMist requires providing a full reductive account of consciousness. Does it? In the first place, consider the paradigm cases in which we have justification to posit essentialist truths with respect to a given phenomenon without having a full reductive account of that phenomenon. Thus, even if knowledge is irreducible, we still have justification to believe that it is essential to knowledge that someone knows a proposition only if that proposition is true [cf. Rosen 2015, 194]. Other examples involve determinables and determinates, sets and their members, disjunctions and their disjuncts, and so on. However, in all of these cases, we have an a priori basis for positing these essentialist truths. Given the poor track record of a priori physicalism, GPMists should not appeal to these cases for a model. For even if it is essential to pain that it be grounded in c-fiber firing whenever one’s c-fibers fire, that does not seem to be the sort of thing that we are in a position to know a priori. The GPMist should instead appeal to cases of scientific discovery for a model. Consider the case of water. According to many philosophers, chemistry has revealed something important about the essence of water: it is part of its essence to be composed of hydrogen and oxygen. But that’s because chemistry gives us justification to believe that water just is H2 O. So, while it is plausible to regard this case as one in which we have acquired a posteriori knowledge of an essentialist truth, it is also one in which our knowledge is parasitic on our having a reductive account of the phenomenon in question. The other paradigm cases of scientific discovery fit this pattern. This should not be surprising. When we have no a priori basis for positing essentialist truths with respect to a given natural phenomenon, our best alternative proceeds by way of inference to the best explanation. Thus, we have justification to believe that it is essential to water to be composed of hydrogen and oxygen because (i) we have justification to accept the hypothesis that water just is H2 O and (ii) the essentialist truth follows from that hypothesis together with independent chemical information concerning H2 O. But why do we have justification to accept that hypothesis in the first place? The answer, it seems, is that it is the simplest explanation of the empirical observation that water and H2 O are correlated [see Block and Stalnaker 1999]. Unlike the hypothesis that water and H2 O are connected by a law of nature, or the hypothesis that it is essential to the property of being water that it be grounded in the (distinct!) property of being H2 O whenever that property is instantiated, the reductive hypothesis minimizes our ontological commitments. We are in a position to know that the reductive hypothesis is true because it leads to an increase in parsimony. This suggests that the GPMist’s only hope of closing the prima facia explanatory gap through empirical investigation is to reduce phenomenal properties to physical/functional properties. However, if the GPMist can pull this off, GPM will no longer be a non-reductive view. Yet, one of the primary motivations for GPM is that it promises to occupy a middle position between certain reductive and non- 14 Z. Blaesi reductive versions of physicalism. Therefore, if my overall argument succeeds, GPM faces the following dilemma: Dilemma: Either Grounding Physicalism About Mentality leaves a genuinely explanatory gap, or the Grounding Physicalist About Mentality is committed to there being a reductive account of consciousness, thereby giving up one of the central motivations for the view. So, what about just accepting the second horn of the dilemma? The problem is that a number of philosophers have concluded that no reductive account of consciousness will ever succeed. That is why GPM is such an enticing option. But on this horn of the dilemma, while the notion of ground apparently opens explanatory gaps, the notion of reduction does much of the heavy lifting in closing them. As a result, it is unclear why we need GPM as a new alternative to standard versions of reductive physicalism. The alternative is to accept that GPM leaves an explanatory gap. This comes at the cost of mystery, disunity, and complexity: mystery—because the grounding patterns cry out for explanation; disunity—because in all the paradigm cases, grounding patterns are explained by the essences of things; and complexity— because the grounding patterns involving consciousness cannot be subsumed under uncontroversial essentialist truths (e.g., about neurological kinds). Some philosophers may be willing to pay this price to avoid the problems of rival accounts. But many will take the existence of the explanatory gap to be strong evidence that there is a metaphysical gap between the physical and the mental. To echo Kim [1989], if the grounding patterns involving consciousness are to be proposed as unexplainable facts that we must accept on faith, why should we take the metaphysical proposal seriously? Following Fine [2012], we might give up the claim that consciousness is metaphysically grounded in the physical and instead defend the weaker claim that it is naturally grounded in the physical. This would be tantamount to embracing dualism. If that is a cost, it remains to be seen which cost is greater. 6 Conclusion I have argued that Grounding Physicalism About Mentality (GPM) initially promises to occupy a middle position between traditional reductive and non-reductive versions of physicalism. Its main appeal is that it has the resources to locate the mind in the physical world without reducing mental phenomena to physical or functional phenomena. I then went on to argue that the grounding patterns involving consciousness are to be explained by truths about the essence of consciousness. If the essence of consciousness does not explain those grounding patterns, I argued, then GPM leaves an explanatory gap. Moreover, this gap cannot be addressed in the usual way, because grounding patterns (unlike identities) are apt to be explained. Ultimately, the GPMist is forced to leave a distinctive explanatory gap or else embrace the reductionist project whose apparent failure motivated GPM in the first place. Perhaps, then, the notion of ground is much less useful to friends of Title removed for blind review 15 physicalism than has been recently suggested. References Bennett, Karen. 2011. “By Our Bootstraps.” Philosophical Perspectives 25:27–41. Block, Ned and Stalnaker, Robert. 1999. “Conceptual Analysis, Dualism, and the Explanatory Gap.” Philosophical Review 108:1–46. Dasgupta, Shamik. 2014. “The Possibility of Physicalism.” Journal of Philosophy 111:557–92. deRosset, Louis. 2013. “Grounding Explanations.” Philosophers’ Imprint 13:1– 26. Fine, Kit. 1995. “Ontological Dependence.” Proceedings of the Aristotelian Society 95:269–90. —. 2012. “Guide to Ground.” In Fabrice Correia and Benjamin Schnieder (eds.), Metaphysical Grounding, 37–80. Cambridge: Cambridge University Press. Goff, Philip. 2017. Consciousness and Fundamental Reality. Oxford: Oxford University Press. Horgan, Terence E. 1993. “From Supervenience to Superdupervenience: Meeting the Demands of a Material World.” Mind 102:555–86. Kim, Jaegwon. 1989. “The Myth of Non-Reductive Materialism.” Proceedings and Addresses of the American Philosophical Association 63:31–47. Levine, Joseph. 1983. “Materialism and Qualia: The Explanatory Gap.” Pacific Philosophical Quarterly 64:354–61. Lewis, David. 1983. “New Work for a Theory of Universals.” Australasian Journal of Philosophy 61:343–377. Lowe, E. J. 2012. “What is the Source of Our Knowledge of Modal Truths?” Mind 121:919–950. Papineau, David. 2002. Thinking About Consciousness. Oxford: Oxford University Press. Pereboom, Derk. 2011. Consciousness and the Prospects of Physicalism. Oxford: Oxford University Press. Rosen, Gideon. 2010. “Metaphysical Dependence: Grounding and Reduction.” In Bob Hale and Aviv Hoffmann (eds.), Modality: Metaphysics, Logic, and Epistemology, 109–36. Oxford: Oxford University Press. —. 2015. “Real Definition.” Analytic Philosophy 56:189–209. —. 2017. “Ground by Law.” Philosophical Issues 27:279–301. 16 Z. Blaesi Tye, Michael. 1999. “Phenomenal Consciousness: The Explanatory Gap as a Cognitive Illusion.” Mind 108:705–25. Wilson, Jessica M. 1999. “How Superduper Does a Physicalist Supervenience Need to Be?” Philosophical Quarterly 50:33–52. Yablo, Stephen. 1992. “Mental Causation.” Philosophical Review 101:245–80.