p40

A Humean Non-Humeanism David Builes Forthcoming in Philosophical Studies Abstract How should we account for the extraordinary regularity in the world? Humeans and Non-Humeans sharply disagree. According to Non-Humeans, the world behaves in an extraordinarily regular way because of certain necessary connections in nature. However, Humeans have thought that Non-Humean views are metaphysically objectionable. In particular, there are two general metaphysical principles that Humeans have found attractive that are incompatible with all existing versions of Non-Humeanism. My goal in this paper is to develop a novel version of Non-Humeanism that is consistent with (and even entails) both of these general metaphysical principles. By endorsing such a view, one can have the explanatory benefits of Non-Humeanism while at the same time avoiding two of the major metaphysical objections towards Non-Humeanism. 1. Introduction The world is an astonishingly regular place. Apples fall, planes fly, and bread nourishes. The astonishment only increases when one looks at fundamental physics. Fundamental physical entities always seem to behave according to certain elegant, ironclad, mathematically precise rules. Why should this be so? A natural first reaction to the astonishing regularity of the world is to think that things don’t just happen to behave as they do, but, in some sense, they have to behave as they do. This “have to” is cashed out in different ways by different “Non-Humean” approaches to natural necessity.1 Perhaps, for example, there are fundamental causal powers or dispositions in the world, or primitive laws of nature, or necessity relations between universals, which necessitate that things behave in the ways that they do.2 1 Although the terminology of “Humeanism” and “Non-Humeanism” is now entrenched, it is not at all clear that David Hume himself was a Humean (e.g. see Strawson 2014). 2 Mumford (2004), Bird (2007), and Shoemaker (1980) defend the view that there are fundamental powers/dispositions. Carroll (1994, 2008) and Maudlin (2007) defend the view that laws are fundamental. Armstrong (1983), Dretske (1977), and Tooley (1977) defend the view that there are necessitation relations between universals.2 However, “Humeans” regard the necessary connections that Non-Humean views posit as metaphysically suspect. For Humeans, laws of nature merely summarize simple and informative natural regularities, and facts about causation, counterfactuals, and dispositions are all ultimately explained in terms of such laws.3 The main goal of this paper is to formulate a version of Non-Humeanism that avoids two of the main metaphysical objections that Humeans have made towards Non-Humean views. I begin by describing these two metaphysical objections and noting how they apply to existing Non-Humean views (section 2). Then, I will turn to formulating a version of Non-Humeanism that avoids these two metaphysical objections (sections 3-5). The version of Non-Humeanism that I will describe involves combining three different metaphysical views, each of which can be (and has been) defended for independent reasons. However, my goal isn’t to further defend any of these three views in isolation. Rather, my goal will be to show how the combination of these views constitutes an attractive overall Non-Humean picture of natural necessity: one which keeps the Non-Humean intuition that there’s a robust sense in which things have to behave in the way they do, while at the same time being compatible with some of the main metaphysical scruples of Humeans. 2. Two Humean Principles The first guiding metaphysical principle that motivates Humeans often goes by the name of Humean Supervenience, which is described by David Lewis as follows: [A]ll there is to the world is a vast mosaic of local matters of particular fact, just one little thing and then another…For short: we have an arrangement of qualities. All else supervenes on that. (1986b: ix-x) Let us make this vision a bit more precise. The kind of view that Lewis is advocating is a strengthening of the view that truth supervenes on being. According to Lewis, “truth supervenes on being” just in case “truth is supervenient on what things there are and which perfectly natural properties and relations they instantiate” (Lewis 1992: 218). However, there is disagreement about the nature of (perfectly) natural properties. Some conceive of natural properties as “pure powers”, where the nature of such properties is exhausted by their modal relations to other properties (e.g. how such properties are disposed to interact with other kinds of properties). For example, one might think that the property of having a particular mass just is the property of 3 Although most self-described Humeans use this “best-system” approach to laws to account for other nomic notions, one can accept an underlying Humean metaphysics without adopting the best system approach. For example, van Fraasen (1989) has argued that Humeans should deny that there are any laws at all, and Wilson (2009) has defended an alternative Humean account of causal facts that does not appeal to general laws. See Friend (2021) for further discussion of different versions of Humeanism.3 being disposed to resist acceleration in certain ways and interact gravitationally with other objects in certain ways.4 However, others conceive of natural properties as “qualities” or “categorical” properties, where the nature of such properties is not determined by their modal relations to other properties. Rather, such properties have their own intrinsic, non-relational nature.5 For example, it is natural to think that our pre-theoretic conception of colors involves thinking of (say) being red as a quality.6 With this distinction in hand, we can now formulate the following two claims: Truth Supervenes on Being (TSB): Every fact supervenes on what things there are and which perfectly natural properties and relations they instantiate. Truth Supervenes on Qualitative Being (TSQ): Every fact supervenes on what things there are and which perfectly natural qualities and relations they instantiate.7 The first Humean principle that will be our main focus is TSQ. It is easy to see how standard Non-Humean views violate TSQ. For example, views according to which there are fundamental causal powers violate TSQ because facts about the distribution of fundamental powers do not seem to supervene on facts about the distribution of fundamental qualities. Other Non-Humean views go so far as to violate TSB, which is a weakening of TSQ. For example, according to certain primitivist views about laws, there are fundamental facts of the form [L(P)], where “L” is a fundamental operator interpreted as “it is a law that” and “P” is some proposition that is a law of nature. Since these facts are fundamental, there is no reason why such facts would have to supervene on facts about the distribution of perfectly natural properties and relations in the world. 4 For defenses of this kind of position, see Chakravartty (2003, 2007), Mumford (2004), and Bird (2007). 5 While the distinction between categorical/dispositional properties is commonplace in discussions of natural properties, it is controversial exactly how to understand this distinction (see Taylor 2018 for a discussion of some of these controversies). It is beyond the scope of this paper to enter into the details of these controversies here, but it should be noted that for our purposes the distinction between categorical and dispositional properties is not simply a modal one. For example, Bird (2007: 66-67) defines a categorical property as any property that does not necessitate a corresponding power or disposition. However, such a definition doesn’t allow conceptual room for the “grounding view of powers” that will be introduced in the next section. See Coates (2020) for a hyperintensional characterization of the distinction between qualities and powers in terms of essence and grounding that will be suitable for our purposes. 6 Of course, upon reflection one might come to think that colors are dispositional properties (e.g. ones that are disposed to cause certain sensations in humans in normal conditions). Chalmers’ (2006) conception of “Edenic” colors are paradigm examples of qualities. 7 In other contexts, “qualitative” properties are meant to contrast with “haecceitistic” properties, such as the property of being Alice (e.g. see Dasgupta 2017). However, in this sense of “qualitative”, both qualities (e.g. being red) and dispositions (e.g. being fragile) are “qualitative”. For the entirety of this paper, I will be using the phrase “qualitative property” to refer to qualities rather than non-haecceitistic properties.The second Humean principle that we will be focusing on is the following: 4 Hume’s Dictum (HD): There are no metaphysically necessary connections between wholly distinct entities.8 To say that two entities are “wholly distinct” is to say that they are (i) distinct and (ii) do not have any mereological or spatiotemporal overlap.9 Clearly, there are necessary connections between an entity and itself (e.g. necessarily, if [x is F], then [x is F]), and there are also necessary connections between entities that overlap (e.g. necessarily, if half of an apple is red, then the whole apple must be red-on-one-half). The Humean can even acknowledge that there might be not-so-trivial necessary connections between an entity and itself: for example, an object’s having 5 units of mass necessitates that it does not have 7 units of mass, and perhaps an object’s having a certain shape necessitates that it has a certain color. However, if two entities are wholly distinct, then it is natural to be skeptical of any metaphysically necessary connection between them. Suppose there were two lights such that it was metaphysically necessary that for any time t, the first light is on at t iff the second light is on at t. This connection cries out for explanation. How can the second light even “know” about the first light? Even if it could “know” about the first light, why is it impossible for the second light to be out of sync with the first light? Perhaps more importantly, HD could also be supported by a conceivability argument. If two entities are wholly distinct, then no matter what the first object is like, it seems conceivable that the second object could be in any possible intrinsic state.10 Many Non-Humean views also violate HD. For example, consider a Non-Humean view according to which the laws of nature are explained in terms of the fundamental causal powers of physical entities. On such a view, if the universe were deterministic, then the intrinsic state of the past (which will include a specification of the causal powers had by things in the past) will necessitate the intrinsic state of the future, violating HD. To my knowledge, no existing formulation of Non-Humeanism is consistent with both TSQ and HD. Insofar as Humeans are committed to TSQ and HD, it is therefore no wonder why they have rejected all forms of Non-Humeanism. In what follows, my goal will be to formulate a version of Non-Humeanism that is consistent with, and even entails, both TSQ and HD. 8 It is a subtle question how Humeans should formulate HD more precisely. For example, a Humean might wish to reformulate HD in a way that allows there to be a necessary connection between Socrates and his singleton set. For our purposes, these kinds of counterexamples won’t be relevant, but for a wide ranging critical discussion of various different formulations of Hume’s Dictum, see Wilson (2010). 9 The choice of defining “wholly distinct” in terms of lack of merelogical and spatiotemporal overlap comes from Wilson (2010: 604-607). 10 For a defense of the claim that a suitably idealized kind of conceivability entails possibility, see Chalmers (2002).3. Part One: The Grounding View of Powers 5 We’ll begin by considering a view that implies TSQ, even though (by itself) it does not imply HD. In order to abide by TSQ, everything must supervene on the global distribution of qualities. But if we only have qualities to work with, and no fundamental causal powers, laws, or necessitation relations between universals, how can we possibly explain why things behave in the way that they do? How can we explain why apples fall, planes fly, and bread nourishes?11 It would somehow have to be that the qualities themselves fully explain why things do what they do. In other words, it must be that the dispositions or causal powers of any given entity are wholly grounded in the qualities that that entity possesses. On this view, while there would be no fundamental causal powers, physical entities can still have derivative causal powers in virtue of having the particular qualities that they have. Such a view has been independently motivated in different ways by Jacobs (2011), Kimpton-Nye (2021), Tugby (2021, 2022), and Builes (2022), among others.12 Let us call the view that fundamental qualities ground (and hence necessitate) corresponding dispositional properties the Grounding View of Powers (GP).13 Perhaps the easiest way to get a grip on GP is by way of a concrete example. In the philosophy of mind, a number of different philosophers have been attracted to Russellian Panpsychism (e.g. Russell 1927, Strawson 2006, Chalmers 2015, and Goff 2017), which is the view that (at least some of) the categorical properties of fundamental physical entities are phenomenal properties (i.e. properties that characterize what it’s like to be a certain entity). The Phenomenal Powers View, defended in different ways by Langsam (2011), Mørch (2014, 2018a, 2018b, 2019), Goff (2020), and Builes (2020) is the view that categorical phenomenal properties may ground certain physical dispositions purely in virtue of their intrinsic phenomenal character (i.e. what they feel like).14 Combining Russellian Panpsychism with the Phenomenal Powers view results in a 11 Here I am implicitly making the Non-Humean assumption that Humean explanations of these phenomena are inadequate. For more on the question of whether Humean laws can genuinely explain, see Lange (2013, 2018), Schumener (2019), Dorst (2019), Bhogal (2020), Kovacs (2020), Hicks (2021), and Emery (forthcoming). 12 For example, Kimpton-Nye (2021) motivates such a view on the grounds that it avoids some of the central objections towards standard versions of “dispositional essentialism”. 13 The label “the grounding theory of powers” comes from Tugby (2021). A closely related view to GP is the “Powerful Qualities” view, according to which fundamental physical qualities are identical to (rather than ground) dispositional properties (e.g. see Strawson 2008 and Heil 2010) . Such a view is similar in spirit to the grounding view, but I will be focusing on the grounding version of the Powerful Qualities view here since it is easier to make sense of. It is interesting to note that the way that Coates (2020) makes sense of the Powerful Qualities view makes it equivalent to GP. 14 Although Goff (2020) is skeptical that phenomenal properties are capable of grounding dispositions, he defends the possibility of closely related “consciousness+” properties grounding dispositions.6 version of GP, where fundamental physical entities possess fundamental phenomenal qualities, which ground the dispositional properties of those entities. Of course, both Russellian Panpsychism and the Phenomenal Powers View are very controversial, and it is far beyond the scope of this paper to defend these views here. However, it is worth at least giving some initial motivation for the Phenomenal Powers view in order to flesh out the example further. The Phenomenal Powers view is often motivated by reflecting on our own phenomenological experiences. It is natural to think, for example, that the feeling of pain makes subjects who experience it try to avoid it simply in virtue of how bad it feels. Pleasure might make subjects who experience it try to pursue it simply in virtue of how good it feels. Someone who feels tired or exhausted might be disposed to stay in bed simply in virtue of how tired they feel. These connections between these experiences and their effects are certainly defeasible. Of course, in our own case, human subjects endure pain for all sorts of reasons. Someone might endure short-term pain in order to avoid more pain in the future (e.g. going to the dentist), or in order to experience a greater pleasure (e.g. masochism), etc. However, in the absence of any interfering causes like these, it seems like pain has a “default” power to make subjects at least try to avoid it. An important lesson behind this kind of example of GP is that it shows that the grounding connection between a quality and its corresponding power needn’t be arbitrary. Rather, there seems to be an intelligible connection between the qualitative character of pain and its corresponding effects: it makes sense why someone having an experience with the qualitative character of pain would try to avoid it. In fact, Mørch (2014, 2018a, 2018b, 2019) has argued that it is inconceivable that the phenomenology of pain has any other causal power than the one that it in fact has. While a defense of this particular claim would take us too far afield, it does show that the defender of GP doesn’t need to be committed to the view that the necessary connection between a quality and its corresponding disposition is brute, arbitrary, or otherwise unintelligible.15 Having gone through this example, however, it should be stressed that GP on its own is certainly not committed to either Panpsychism or the Phenomenal Powers view. In fact, one could think that both views are extremely implausible while still endorsing GP. The example is only meant to be a concrete case that illustrates the conceptual possibility of a view like GP. Another kind of example that has been used to motivate GP are geometrical properties (e.g. see Heil 2003: 86 and Martin 2008: 44-45). Intuitively, the (intrinsic and non-dispositional) property of being a sphere (partly) grounds a corresponding disposition to roll in suitable circumstances, whereas the property of being a cube does not ground a similar disposition, because the intrinsic non-dispositional nature of a sphere differs from the intrinsic non-dispositional nature of a cube. 15 As Kimpton-Nye (2021) points out, however, one might also believe that the relevant grounding relations are “opaque”, in the sense that it is conceivable that the grounding fact obtains without the grounded fact. See Schaffer (2017) for an extended defense of the general view that opaque grounding relations are pervasive.7 Although GP has been independently motivated in many different ways, for our purposes the main attraction of GP is that it complies with TSQ.