p22

Experience and Time: A Metaphysical Approach By David Builes and Michele Odisseas Impagnatiello [Forthcoming in Analytic Philosophy] Abstract What is the temporal structure of conscious experience? While it is popular to think that our most basic conscious experiences are temporally extended, we will be arguing against this view, on the grounds that it makes our conscious experiences depend on the future in an implausible way. We then defend an alternative view of the temporal structure of experience from a variety of different objections. Along the way, we hope to illustrate the wider philosophical ramifications of the relationship between experience and time. What one thinks about the temporal structure of experience is, we believe, deeply interconnected with issues concerning whether consciousness is vague or precise, whether conscious states can be reduced to physical states, whether phenomenal properties are intrinsic properties, and whether phenomenal consciousness can “overflow” access consciousness. As we will see, even seemingly unrelated metaphysical questions, such as the debate between Humean and Non-Humean accounts of natural necessity, bear on questions about the relationship between experience and time. 1. Introduction Consider the experience of listening to your favorite song. This is a temporally extended experience, which is composed of more specific, shorter-lived experiences. Reflecting on this kind of example naturally leads one to draw a distinction between basic and non-basic experiences (or phenomenal properties). You have non-basic experiences, like the experience of listening to your favorite song, in virtue of having other, more specific experiences, such as the auditory experience of a specific chord in that song. Basic experiences are ones that you do not have in virtue of having any other experiences.1 1 This basic/non-basic distinction is closely related to the fundamental/non-fundamental distinction, but Physicalists will not think that any experience is absolutely fundamental. For Dualists, the distinction between basic and non-basic experiences coincides with the distinction between fundamental and non-fundamental experiences. For more on “in virtue of” explanations, see Rosen (2010), Audi (2012), and Dasgupta (2017).2 How short-lived are your basic experiences? A popular view is that such experiences cannot be instantiated over durationless instants, but rather that they must have some minimal positive temporal extension. However, because we can still intelligibly consider the question of whether or not a person is conscious at some particular time (e.g. we can ask whether you were conscious at midnight last night), we will formulate such a view as follows: Minimal Positive Duration: There is some positive minimum length of time L such that, if you are conscious at time t in virtue of instantiating a basic phenomenal property P, then P is instantiated during some temporal interval T with length at least L, where t ∊ T. We will be arguing against Minimal Positive Duration. After briefly introducing some of the main considerations in favor of Minimal Positive Duration (section 2), we develop an argument against Minimal Positive Duration (sections 3-5), and then defend an alternative account of the temporal structure of experience that is responsive to some of the main motivations for Minimal Positive Duration (sections 6-10). Beyond the debate about Minimal Positive Duration, however, our broader goal is to illustrate the wider philosophical ramifications of the relationship between experience and time. What one thinks about the temporal structure of experience is, we believe, deeply interconnected with issues concerning whether “consciousness” is a vague or precise notion, whether conscious states can be reduced to physical states, whether phenomenal properties are intrinsic properties, and whether phenomenal consciousness can “overflow” access consciousness. As we will see, even seemingly unrelated metaphysical questions, such as the debate between Humean and Non-Humean accounts of natural necessity, bear on questions about the relationship between experience and time. 2. Minimal Positive Duration Why believe Minimal Positive Duration? We will briefly go over three such reasons here. First, we seem to be directly aware of change and duration in experience. Consider, for example, the experience of seeing a moving object: we’re not merely aware of the object being at different positions at different times; rather, we also seem to be aware of the motion itself. Prima facie, the phenomenology of such experiences can’t be reduced to instantaneous experiences (even supplemented with, say, memories of previous experiences: the experience of seeing a moving object is quite different from that of seeing an object at a certain location while remembering it was a different location). In slogan form: in these cases, we have an experience of succession,3 and not merely a succession of experiences. To properly account for experiences of change and duration, many have thought that our experiences themselves must be spread out in time.2 A second consideration is that many philosophers have found it difficult to even conceive of durationless experiences. It is easy to conceive of a very short pain, or a very short musical note, but we arguably cannot conceive of a durationless pain, or a durationless note. Try to imagine, for example, a possible person who exists for just one instant, an instant in which the person feels intense pain. We suspect that many philosophers will find it difficult to conceive of such a case. In order for something to be painful, it seems, it must be painful for some positive amount of time. Butterfield (1996: 210) appeals to a similar example involving the experience of grief, which he uses to motivate the claim that it is “undeniable and maybe even necessary” that conscious states require duration. Although it is hard to give explicit arguments for these intuitions, many philosophers have taken them to be phenomenologically obvious, as when Strawson (2009) simply writes, “One thing I take for granted is that experience takes time: it can’t exist or occur at an instant, where an instant is defined as something with no temporal duration at all” (256). A final consideration concerns the physical correlates of conscious experience. As Lee (2014) emphasizes, the physical correlates of conscious experiences are plausibly all extended in time. Neural firings, for example, cannot take place in an instant. However, if the physical correlates of phenomenal properties are extended in time, then the phenomenal properties that they realize are also plausibly extended in time. This follows from the following principle, which Lee endorses: The Temporal Identity Principle: Every phenomenal property is instantiated over the same temporal interval as its corresponding physical correlate.3 One way to motivate The Temporal Identity Principle goes by way of certain views about the metaphysical nature of consciousness. For example, Physicalist views according to which experiences are identical to their physical correlates must accept The Temporal Identity Principle. Similarly, Russellian Monist views according to which experiences are identical to the intrinsic categorical basis of their physical correlates must also accept The Temporal Identity Principle.4 Other formulations of Physicalism and Russellian Monism, according to which 2 This kind of argument has a long history of controversy. For an overview of this kind of argument and its potential ramifications for the temporal structure of experience, see Dainton (2008a, 2018). 3 Following Shoemaker (1981), Lee (2014) distinguishes between an experience’s “total” physical correlate and its “core” physical correlate. An experience’s total physical correlate is the minimal set of physical events that metaphysically necessitate the corresponding experience (perhaps conditional on the psycho-physical laws if Dualism is true). An experience’s core physical correlate is “the part of the total [physical correlate] that differentiates the type of experience in question (e. g. the activity in V4 that determines that an experience is of blue rather than of red)” (3). Since it is plausible that contemporary accounts of the total and core physical correlates of any experience are temporally extended, our discussion in the main text can stay neutral between these two understandings of an experience’s physical correlate. 4 For defenses of Russellian Monism, see Strawson (2003), Chalmers (2013), and Goff (2017).4 experiences are grounded in (the intrinsic nature of) their physical correlates, are not logically committed to The Temporal Identity Principle, but the principle still remains very natural. After all, standard examples of grounding all involve a synchronic relationship between the grounding facts and the grounded facts, in which the grounding facts obtain at the very same time(s) as the grounded facts. For example, if the existence of a party is grounded in certain facts about the activity of various people, the temporal duration of such a party will exactly match the temporal duration of those corresponding activities. Similarly, if the mass of a mereologically complex object is grounded in the individual masses of its parts, then this will also be a synchronic relation. In these and many other canonical examples, there is no temporal discrepancy between the obtaining of the grounding facts and the grounded facts.5,6 One could also motivate a weaker version of The Temporal Identity Principle that would serve our purposes equally well: The Temporal Inclusion Principle: If a phenomenal property is instantiated over a temporal interval T1, then its corresponding physical correlate must be instantiated over some interval T2, where T2 is a subset of T1 (T2 ⊆ T1).7 Since The Temporal Inclusion Principle implies that the temporal extent of a phenomenal property is at least as long as the the temporal extent of its corresponding physical correlate, it can also be used to argue that phenomenal properties are temporally extended. The Temporal Inclusion Principle can be motivated by the following natural principle (which is conditional on the psycho-physical laws to be amenable to Dualists): Phenomenal Internalism: Given the psycho-physical laws (if there are any), if x and y are physical duplicates throughout some temporal interval T, then x instantiates 5 In fact, in his introductory treatment of grounding, Rosen (2010) explicitly says that “the grounding relation is a synchronic relation” (218). However, Rosen (2017: 280) later expresses skepticism that the synchronicity condition holds in complete generality (e.g. he considers the example of the semantic content of a word at some time being grounded in temporally extended facts about patterns of usage in the wider linguistic community). For our purposes, we only want to claim that there is prima facie reason to endorse The Temporal Identity Principle given grounding formulations of Physicalism and Russellian Monism. 6 Perhaps the most plausible way to resist The Temporal Identity Principle goes by way of Dualism, according to which experiences are merely nomologically connected to their physical correlates. While Dualism is certainly compatible with The Temporal Identity Principle, it is also compatible with the claim that experiences occur at the “end” of (or even strictly after) their corresponding physical correlates. One could even independently motivate this kind of Dualism by considering a version of Interactionist Dualism where physical correlates are the cause of phenomenal properties. Given the premise that causes must precede their effects, the physical correlate of an experience would then have to precede the experience it causes. We ourselves are somewhat skeptical of this motivation even granting Interactionist Dualism, due to the possibility of simultaneous causation. See Huemer and Kovitz (2003) and Mumford and Anjum (2011) for arguments that causation is always simultaneous. 7 Since we intend The Temporal Inclusion Principle to be weaker than The Temporal Identity Principle, T1 can be be a “subset” of T2 if T1 = T2 (i.e. T2 need not be a proper subset of T1).5 phenomenal property P during T if and only if y instantiates phenomenal property P during T.8 Phenomenal Internalism states that perfect physical duplicates throughout some interval must also be phenomenal duplicates throughout that interval. It is therefore closely related to the plausible claim that phenomenal properties are intrinsic properties. In fact, one popular way of defining intrinsicality goes by way of properties that are always shared among perfect duplicates.9 To see why Phenomenal Internalism supports The Temporal Inclusion Principle, suppose for reductio that there could be a physical correlate P of an experience E that violated The Temporal Inclusion Principle. So, consider some x that instantiates E during T1 and P during T2, where T2 is not included in T1. Then, there could be some y that is a perfect physical duplicate of x during the interval T1, yet y does not instantiate E during T1 because y does not have the required physical correlate for E (because y can be physically very different from x during the times inside T2 and outside T1). So, we think there are at least three good reasons for thinking that our experiences are temporally extended. However, one might worry whether these three considerations only motivate the following claim, which is weaker than Minimal Positive Duration: Positive Duration: If you are conscious at time t in virtue of instantiating a phenomenal property P, then P must be instantiated during some temporally extended interval T, where t ∊ T. It is compatible with Positive Duration, but not Minimal Positive Duration, that (basic) phenomenal properties can be arbitrarily short-lived. However, the main motivations for Positive Duration also motivate Minimal Positive Duration. For example, whatever the physical correlates of conscious experiences are, it is natural to think that they will have some positive minimal temporal extension. The temporal grain of biologically relevant processes in the brain, for example, is certainly not below half of the Planck time. Moreover, insofar as one believes that our experiences of change and duration must be accommodated by temporally extended experiences, one ought to think that these experiences have some minimal temporal length. After all, hearing a note of C flow into a note of D is a change that occurs on the order of milliseconds rather than picoseconds. For these reasons, there doesn’t seem to us to be a theoretically natural and principled position that endorses the motivations behind Positive Duration, while at the same time rejecting Minimal Positive Duration.10 8 x instantiates a phenomenal property P “during” T just in case P is instantiated over some temporal interval that is a subset of T. 9 See Marshall and Weatherson (2018) for an overview of different ways of analyzing the notion of an intrinsic property. 10 One potential exception to this can be found in Dainton (2014), where Dainton argues that a ‘gunky’ view of conscious experience (according to which phenomenal properties are infinitely divisible) can respond to some arguments raised by Pelczar (2010b).6 Lastly, it is worth emphasizing that Minimal Positive Duration is weaker than other natural nearby principles. Minimal Positive Duration only makes a claim about the experiences that you, a particular human organism, will in fact have. It makes no claims about what experiences are possible for you to have, and it makes no claims about what experiences are possibly had by other possible creatures.11 So, in sum, we think that Minimal Positive Duration enjoys a great deal of plausibility. Nonetheless, we will be arguing against Minimal Positive Duration. Our strategy will be to argue that, when conjoined with another plausible principle about consciousness, Minimal Positive Duration implies the absurd conclusion that you have been conscious for an infinite amount of time in the past. 3. Your Experiences Do Not Wholly Depend on the Future It is natural to think that whether you are conscious here and now should not crucially depend on what may or may not happen later. We think this intuition can be made precise with the following thesis: Not Wholly Future Dependent: If you are conscious at time t, then there must be some phenomenal property P that you instantiate over some temporal interval T, where no time strictly later than t is a member of T, such that you are conscious at time t in virtue of instantiating P.12 Before we turn to defending Not Wholly Future Dependent (and explaining why it captures the intuitions that it is meant to capture), a couple of clarificatory points. First, Not Wholly Future Dependent is logically independent of both Positive Duration and Minimal Positive Duration. For example, the relevant phenomenal property in Not Wholly Future Dependent may be instantiated in an instant (e.g. over the “interval” [t,t]) or over an arbitrarily short interval, contrary to (Minimal) Positive Duration. Conversely, it might be that, whenever you are conscious at time t, you instantiate a temporally extended phenomenal property over the interval [t − L/2, t + L/2], which satisfies (Minimal) Positive Duration but not Not Wholly Future Dependent. Perhaps more importantly, Not Wholly Future Dependent is perfectly compatible with your present consciousness being future dependent. It is only meant to be incompatible with your present consciousness being wholly future dependent. Here is a toy example to illustrate this point. Consider some time interval [0,2], and suppose you instantiate multiple overlapping 11 In fact, if you will only ever have finitely many basic experiences throughout your life, each of which is instantiated over a positive interval of time, then L in Minimal Positive Duration can simply be the minimal length of those positive intervals of time. So, there would be no room for the possibility that Positive Duration might be true without Minimal Positive Duration. 12 See Phillips (2011, 2014) and Soteriou (2013) for defenses of certain ‘holistic’ views about the temporal structure of conscious experience that are in tension with Not Wholly Future Dependent.7 temporally extended phenomenal properties at time t = 1.13 For example, perhaps you instantiate one temporally extended phenomenal property P1 over the interval [0,1] and you instantiate another phenomenal property P2 over the interval [0.5, 1.5]. Both of these phenomenal properties might be sufficient for your consciousness at time t = 1. So, it might be that you are conscious at time t = 1 in virtue of instantiating P1 and you are conscious at time t = 1 in virtue of instantiating P2. This is a case of harmless overdetermination, such as a case where a disjunctive truth “P or Q” holds both in virtue of “P” being true and in virtue of “Q” being true. In our toy example, there is a sense in which the fact that you are conscious at time t = 1 is future dependent. After all, the fact that you are conscious at that time holds in virtue of (but not only in virtue of) the fact that you are instantiating P2, which is a fact (partly) concerning your phenomenology at future times. However, the fact that you are conscious at time t = 1 is not wholly future dependent, since it also holds in virtue of the fact that you are instantiating P1, which only concerns the past and present. In order to bring out the intuition that Not Wholly Future Dependent is supposed to capture, consider the following case: Two Rooms: Suppose you are in a closed room, and you are staring at a bright red wall. In another exactly similar room, there is a perfect duplicate of you staring at an exactly similar bright red wall. Everything goes normally in your room. Everything goes normally in your duplicate’s room, until noon. However, directly after noon, your duplicate will cease to exist (perhaps, for example, a bomb will explode and destroy their room).14 Question: is your perfect duplicate conscious at noon? It seems to us that the answer is clearly “yes”. Of course, you will be conscious at noon, because nothing strange is going on in your room. However, nothing strange is happening in your duplicate’s room at any time up to and including noon. If you were conscious at noon but your duplicate wasn’t, this would mean that your consciousness is wholly future dependent. Whether or not you are conscious at noon will crucially turn on what may or may not happen after noon. The implausibility of this case seems to us akin to the implausibility of backwards causation. 13 Many philosophers who believe that experience is temporally extended also believe that experiences overlap in this way. See, for example, Foster (1991) and Dainton (2008b).