William S . Robinson Developing Dualism and Approaching the Hard Problem

Arguments for property dualism offer a strong challenge to materialist views, but even if they are regarded as successful, a large task remains, namely, to develop a positive account of the place of non-physical properties in the world — one that holds some promise of eventual satisfaction regarding the hard problem. After noting some difficulties in current approaches to this task, this paper outlines one possible line of development for a dualistic view. Like all other suggestions for routes to progress in this area, this one is speculative. However, the empirical findings that would support this line of development lie within current epistemic possibility. Moreover, the conceptual changes that would be required are intelligible from our present vantage point, and have parallels in views that are accepted in science and by non-dualist philosophers. The dualism to be discussed in this paper claims that there are phenomenal qualities, and that these are different from, and not composable from, the properties and relations found in our natural sciences. Phenomenal qualities include the colours, the tastes, the odours, and other sensory qualities; painfulness, itchiness, and other bodily sensation qualities; and qualities that occur in emotions such as anger, jealousy, and elation. Dualists hold that many words have a Journal of Consciousness Studies, 21, No. 1–2, 2014, pp. 156–82 Correspondence: William S. Robinson, Iowa State University. Email: [email protected] [1] Some may wish to add to this list, likely candidates being feelings in a non-sensory ‘fringe’ such as familiarity or confidence (Mangan, 2001), and specifically cognitive phenomenology (see Bayne and Montague, 2011, for a variety of positions). Issues connected with these additions must be left for another day: the focus of this paper is on sensory qualities. C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n double use: they can be used for phenomenal qualities and they can also be used for physical properties, typically ones whose instances are causally related to instances of phenomenal qualities. The characteristic claim of the dualism of this paper is not that there are no physical colours — for example, not that there is no set of reflectance profiles that is appropriately classified under a term such as ‘red’. The distinctive claim is that our experiences (a) are non-physical, qualitative events; that is, they are, or essentially involve, instances of phenomenal qualities, i.e. instances of properties that are not instances of physical properties; and (b) these property instances have no further, or hidden, physical nature. There are many arguments for property dualism (see, for example, Chalmers, 1996; White, 2007; Goff, 2011; Nida-Rümelin, 2007; Robinson, 2004). This paper does not add to these arguments and, although its author is evidently sympathetic to them, the paper does not, strictly speaking, assume that any of these arguments are successful against materialist views. The stance of this paper is, instead, that arguments for property dualism are sufficiently strong to keep dualism in play, as a view that deserves to be further developed. Churchland (1984, p. 19) has taunted dualists with this thought: ‘Compared to the rich resources and explanatory successes of current materialism, dualism is less a theory of mind than it is an empty space waiting for a genuine theory of mind to be put in it.’ We might call this the charge of Poverty of dualism. One aim of this paper is to describe a series of developments that are at present epistemically possible, and that, taken together, would answer the charge of Poverty. There is currently much interest in panpsychism (see, for example, Strawson, 2009; Seager and Allen-Hermanson, 2012) and in Russellian monism (See, for example, Chalmers, 1996; Stoljar, 2006). Difficulties for both views are well-known. Consciousness for DEVELOPING DUALISM 157 [2] Most qualitative events are instantiations of several phenomenal qualities. This fact raises some significant issues, but since they do not affect the arguments of this paper, they will be set aside here. They have been addressed in Robinson (2004). It is common for contemporary philosophers to insist that colour terms (and terms for other sensory qualities) have their primary use, or their only correct use, as denoters of properties of physical objects (or, perhaps, physical surfaces). But it is also common to recognize the intelligibility of a dissenting view, although it is often insisted that a dualist’s experiential properties must properly be denoted by some special term such as ‘reddish’ or ‘phenomenal red’. The double use of words affirmed in the text is thus not substantively different from familiar understandings of dualism. It differs only in its denial of the appropriateness of special terms for quality words in their phenomenal use. [3] Cp. Owen Flanagan (1992, p. 35): ‘The view that subjectivity is easy to understand if we think of it as part of an immaterial world has proved to be an illusion. The view actually explains nothing...’ C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n inanimate objects, whether fundamental particles or tables, is counterintuitive, and the combination problem for panpsychism remains formidable. Russellian monism is equally counter-intuitive if it attributes phenomenal qualities to fundamental particles, but if it avoids this problem by appeal to protophenomenal qualities, it faces its own combination problem — how can protophenomenal qualities combine to produce a red sensation, or a pain? (See, for example, Nagasawa, 2008.) This paper neither elaborates on nor adds to these difficulties. They are mentioned here only as motivations for an alternative development. At present, panpsychism and Russellian monism are proposals whose development is speculative and incompletely worked out. Perhaps they will eventually be given a more satisfactory articulation. In the meantime, it is reasonable to ask whether there can be an articulation of dualism that is alternative to these views. This paper aims to develop one such alternative. There is one further motivation that will be important for our discussion. A mainspring of writings on consciousness in the last 18 years has been Chalmers’ (1995) hard problem. Developing dualism in the present climate requires offering some sort of response to this problem. Some authors, however, seek to dismiss this problem in a way that has been given a succinct expression by O’Hara and Scutt (1996): ‘For a problem to be a genuine problem, some sort of idea of a solution must be available...’ But the hardness of the hard problem is that it is not easy, where ‘easy’means that ‘we have a clear idea of how we might go about explaining them’ (Chalmers, 1995, p. 203). Putting these two thoughts together suggests the result that the very thing that is supposed to make the hard problem hard actually makes it not a genuine problem. Let us call this the Hard Problem Problem (HPP) — the problem of maintaining that the hard problem (HP) is a genuine problem. Neither materialists nor non-materialists need accept the premise of O’Hara and Scutt that leads to the HPP. However, it would count as a virtue of a dualistic theory if it could lay the HPP to rest. We shall see that the way of developing dualism that is outlined in this paper is 158 W.S. ROBINSON [4] Michael Tye (2000, p. 34) raises a similar problem about the explanatory gap: ‘Since an explanatory gap exists only if there is something unexplained that needs explaining, and something needs explaining only if it can be explained (whether or not it lies within the power of human beings to explain it), there is again, no gap.’ In Consciousness Online 4, a paper by Glenn Carruthers and Elizabeth Schier (2012) has the title ‘Dissolving the Hard Problem of Consciousness’. Its first sentence is: ‘In this paper we attempt to dissolve worries around the hard problem of consciousness by showing that there is no good argument for the existence of such a problem.’ C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n entitled to claim this virtue. This is not to say that it offers a solution to the HP itself. That solution depends on empirical matters that are not yet known, and on possible conceptual changes that are envisioned in this paper, but not (yet) actually in force. It will, however, be useful to begin by considering a constraint on the solutions of both problems. 1. A Constraint on Solutions of the HP and the HPP It will be helpful to formulate the HP as consisting of the following two questions. Why, given that there are our neural events and their causal relations to each other and to the world at large, should there be any phenomenally qualitied events at all? Why, given that there is a particular kind of neural event (standing in its particular web of causal relations) should there be this particular kind of qualitative event that it yields? (‘Yields’ is a studiously neutral term covering identity, correlation, and causation. Identity theorists will bridle at ‘correlation’, representationalists will reject ‘identity’, and dualists will reject both. But all agree that where there are neural events of certain kinds there are phenomenal qualities, and ‘yields’ is introduced to have a brief way of capturing this commonality.) These questions express an intellectual dissatisfaction. For whatever reason, we have an ‘intuition of distinctness’ (Papineau, 2002) that seems to tell us that phenomenal qualities just cannot be the same as any physical properties, or any combination of them or relation among items with physical properties. This leaves us with two sets of properties and no obvious reason why one should yield the other. A solution to the HP requires providing a reason why neural events yield qualitative events. Correlatively, a solution to the HPP requires providing a sketch of such a reason, or at least a proof that such a reason could be worked out. It requires providing a set of ideas for which it is plausible that further developments could remove our intellectual dissatisfaction. What is needed is a point of view, about which we can imagine our descendants feeling that of course, given a defensible view of how things might very well be, our neural events would yield qualitative DEVELOPING DUALISM 159 [5] The phenomenal concept strategy seeks to explain away this intuition. It is clear that dualists are not convinced by this strategy but, as in other cases, it is beyond the scope of this paper to explain why. [6] Thanks to an anonymous referee for suggesting the last clause of this sentence. It is, however, difficult to imagine how there could be an existence proof of a reason that would provide intellectual satisfaction even though no sketch of the reason itself could be derived from it. C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n events, and of course particular neural events would yield just the kind of qualitative events that they do. It may be that such a feeling of understanding of the neural event– qualitative event relation is not possible. Price (1996), following work by Rosch (1994), has argued that causal relations are by nature explanatorily gappy, and that we have several means of hiding this gappiness from ourselves. On this view, what is special about the mind/body case is not that there is an explanatory gap, but that in this case we are uniquely unable to apply our customary means of hiding the gap from ourselves. Perhaps we will eventually have to concur in some such account. But we should not accept it without first making a serious attempt to achieve intellectual satisfaction about the relation between neural events and qualitative events. The first step of such a project is to set out a possible route by which such satisfaction might be obtainable. This is the project to be pursued in this paper. 2. A Short but Unsuccessful Route Some philosophers already hold the view that causal relations are essential to properties (e.g. Shoemaker, 1980; 1998; 2007; Swoyer, 1982; Bird, 2005). So, it is not difficult to imagine that our descendants become persuaded to accept this way of thinking. It should be noted that causal essentialism can be adopted by dualists. ‘In all metaphysically possible worlds in which A is instantiated, A causes B (in circumstances C)’ does not imply that A and B are the same property. Indeed, in general, effect properties are distinct from the properties of their causes, so causal essentialism can quite naturally be taken as the assertion of necessary connections between distinct properties. There is a second belief that we can imagine our descendants to hold, namely, the view that neural events cause qualitative events. It is true that references to the ‘NCC problem’ — the problem of the neural causes of consciousness — are sometimes avoided in favour of references to the ‘NCCC problem’ — i.e. the problem of the neural causes or correlates of consciousness. But there are many philosophers even now who would be willing to embrace the causal arm of this 160 W.S. ROBINSON [7] Note that this formulation does not imply that qualitative events are reducible to the physical. Thus, although such a reduction would likely provide the desired intellectual satisfaction, proposing reduction as a requirement would be to impose a constraint on the form of solutions that dualists (and others) are free to reject. [8] Of course, this view offends Humean sensibilities. For its defence, see, for example, Wilson (2010). C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n disjunction, and one can surely imagine the possibility of this view becoming more popular than it is at present. Someone who holds both of these views might well conclude that the causal relations between certain of our neural events and our qualitative events are metaphysically necessary. We would, of course, find out which neural event properties yield our qualitative events only by empirical investigation. That fact might account for our tendency to think that the NE–QE relations are contingent, and to regard the relation as mysterious. But, on the combination of views being considered, the sense of mystery is founded on an epistemic illusion of contingency. If our descendants become thoroughly persuaded of causal essentialism, they might very well feel that events instantiating our phenomenal qualities have to occur given the neural events of the kinds that have arisen in the brains of many animals in the course of evolution (or, have to be caused by those neural events). We shall see that something like this suggestion will play an important role in our sketch of a solution to the HP. As it stands, however, this simple account will not provide what is required of it. The reason is that the most persuasive argument for causal essentialism turns on the idea that transworld identification of properties depends on using causal profiles; but this reason does not evidently apply in the case of phenomenal qualities. To explain, consider the proposal that in W1 — a possible world different from the actual world — certain items are electrons, but the laws of W1 are different from ours, and these items do not repel each other. A very plausible reaction to this proposal is that there could be no principled reason that would justify the proponent of such a view in calling the indicated items ‘electrons’. They might just as well be counted as neutrinos with a mass different from actual-world neutrinos, or as particles of a kind that do not exist in our world. The point is generalizable to all particles, and to ordinary objects such as bricks and windows. There is no principled way of defending a claim such as ‘In W1 throwing bricks at glass windows does not break them’. It is just as plausible to deny that W1 really contains bricks, or windows that are made of glass, and not some other substance that presents the same superficial appearance. This kind of point, however, does not seem to generalize to phenomenal qualities. It at least appears to us that ‘blue’ or ‘sweet’, in their uses to describe the qualities in our experience, are not identified DEVELOPING DUALISM 161 [9] Shoemaker (2007, p. 142): ‘Basically, my case for CTP [causal theory of properties] in my earlier work comes down to the claim that there is no plausible truthmaker for the identification of properties in different worlds having different causal profiles.’ C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n by us through any causal relations. If we want to communicate about our experiences, we need to use common predicates, and, very plausibly, that will involve reliance on causal relations, as occurs, for example, in ‘the colour I normally see when I look at a cloudless sky’. But we do not appear to have any difficulty in imagining possible worlds in which looking at cloudless skies causes a different kind of experience. It at least appears to us that our words for phenomenal qualities are rigid designators whose reference is fixed by us through acquaintance with those qualities, without any involvement with assumptions about causal relations. It may be that this way of looking at our words for phenomenal qualities harbours some kind of error. If we come to see such an error clearly, and learn how to convincingly dispel it from our thinking, the short route might become available. But as long as it seems to us that phenomenal qualities do not have essential causal relations, and as long as we have no reason to think otherwise, we will not have intellectual satisfaction about the proposition that neural events of particular kinds cause particular kinds of qualitative events. We will not be able to feel that particular kinds of neural events have to yield the particular kinds of experiences that they do — or, indeed, any kinds of experiences at all. Progress may, however, be made by attending to some facts about complexity. The following section introduces some key considerations that will be developed in the remainder of the paper. 3. Some Reflections on Complexity Our qualitative events and their properties are complex in various ways. Colours have degrees of saturation as well as hue and brightness, and hues themselves may be unique or mixed. Sounds have timbre and loudness as well as pitch. Tastes and smells produced by good cooks or good vintners are often described as ‘complex’. Where we find such complexity, we also find a foothold for plausible suggestions for plurality in its causes. Cooks use several herbs and spices in the same dish, there are many chemicals in wines, musicians distinguish between fundamentals and overtones, artists mix their paints, and so on. The complexity of our phenomenal qualities is, however, orders of magnitude less than the complexity of neural properties that we must suppose to be instantiated in neural events that yield qualitative events, and orders of magnitude less than the field properties proposed by electromagnetic field theories (Pockett, 2002; McFadden, 162 W.S. ROBINSON C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n 2013). There are thousands of distinguishable phenomenal qualities, and therefore thousands of relevantly different kinds of neural event types, or field types, that yield them. So far as we have any reason to believe, the properties of neurons that are relevant to their yielding of qualitative events are their degrees of activation. To obtain sufficient diversity in the yielding neural events, we must therefore suppose that they consist of the activation levels of many neurons. Field theories will require fields that extend over regions that include many neurons. A consequence of the disparity between the degree of complexity found in phenomenal qualities and in brain events that yield our qualitative events is that we are unlikely to find anything plausible about a connection between a particular kind of qualitative event and the particular kind of brain event that yields it. For dualists, this means that no amount of focusing on the nature of a phenomenal quality is likely to provide any intuitive sense that the cause of its instances should be a brain event of type N, where ‘N’ is given as a set of activation levels of a group of neurons, or fields that depend on neural activity. It may very well feel natural that mixing pure yellow and pure green should result in a colour that strikes us as more complex than either element in the mixture, but the relative simplicity of phenomenal qualities, as compared with the likely brain event causes of their instances, will fail to suggest a natural connection between them, so long as we focus simply on one phenomenal quality and one brain event type. This line of thinking may be resisted in various ways: (a) We each have ~100 billion neurons, so perhaps there are single neurons, each of which causes its characteristic phenomenal quality when activated. However, aside from its ill fit with contemporary vision science, this suggestion does not seem capable of providing intellectual satisfaction. ‘Activating this neuron makes you have a salty taste, activating that one makes you see red’ seems about as unexplanatory as a claim could be. (b) Taxonomizing neuron types is an ongoing project, but we already know there are hundreds of distinguishable types. Perhaps a fully developed taxonomy would reveal enough different types so that DEVELOPING DUALISM 163 [10] We can think of such sets of activation levels as vectors, and different yielders as different points in a high-dimensional vector space. For elaboration of this way of thinking see P.M. Churchland, e.g. his (2012). Some discussions of sensory systems emphasize distinct regions, e.g. primary visual cortex vs. primary auditory cortex. But nothing in our sciences gives mere difference of location a causal role. The different brain regions must contribute different kinds of events, if they are to yield different qualitative events. And these different kinds, so far as we have any reason to believe, must be differences in patterns of activations of sets of neurons, or, perhaps, differences in the values of field strengths over some region. C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n we could say ‘Activation of type 1 neurons produces blue experiences, activation of type 2 neurons produces an experience as of oregano...’ This is an improvement on (a), in that it relates types to types, rather than individuals to types. But it is intellectually unsatisfying: what does the shape, or number of dendrites, or branching pattern have to do with colours or odours? (c) A more philosophically motivated resistance may come from representationalists, for whom phenomenal qualities are in external objects, while neural events distinctively track them and mediate discriminative responses. This proposal, however, does not evidently lay questions about complexity to rest. Colours in objects will have to be reflectance profiles (or, perhaps, dispositions to produce characteristic reflectance profiles in normal lighting conditions, or molecular structures that ground such dispositions), tastes will be molecular structures, sounds will be complex waves. These properties have far greater complexity than what is apparent in our phenomenal qualities. If we say that phenomenal properties just are these complex physical properties, we will have to explain how a property can appear in some way, while failing to appear to have the complexity it actually has. And, of course, we will have to do so without appealing to the idea that external properties are merely causes of experiences that either have or represent different, less complex properties. This fact makes it problematic to say that we represent external properties, but do not represent them as they actually are. But if we cannot say that, the only way to preserve representationalism will be to hold that, in addition to the complex external properties that are usually taken to cause events in our sensory cortices, there are properties of external objects that are simpler — and therefore different — from reflectance profiles, molecular structures, and so on. This amounts to a form of dualism, since these simpler properties do not figure in any of our natural sciences. Such a view would not likely lead to intellectual satisfaction, because it is unparsimonious, and because it would naturally invite questions of the form ‘Why does complex external property P1 give rise to representations of the particular simpler external property Q1 that it does?’ 4. Complexity and Simplicity Our reflections on complexity can, however, be turned to significant advantage, once we notice that the complexity of neural causes and 164 W.S. ROBINSON [11] Tye (2009) gives an especially clear version of this kind of view. C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n the relative simplicity of qualitative event types locate neural events and qualitative events with respect to a common property. We have no name for this property, but there is no bar to introducing one. Following Spinoza’s use of ‘motion and rest’ (a pairing of apparent contraries) as a term of art designed to bring out the fact that being at rest is a property on the same scale as being in motion, let us use ‘complexity and simplicity’ — abbreviated to CAS — to designate the common property that is manifested in a certain kind of complexity in neural events and a certain kind of simplicity in qualitative events. This section and the next explain both the terms in this definition, and the justification for regarding complexity and simplicity as two manifestations of a common property. ‘A certain kind of complexity’ has to be taken seriously. CAS is not manifested in just any kind of complexity, but only in complexity of the kind that occurs in the brain events that we may discover to be the causes of qualitative events. It is, of course, an empirical assumption that there is a property that fits this description. Perhaps, at some point in the future, we will come to see the search for such a property as hopeless. But at present, it seems to many to be a worthwhile research programme: it’s the one commonly referred to as the NCC (or NCCC) problem. There are complex neural events that yield qualitative events (either directly or through production of fields), but there are also neural events that must be quite complex in their way, but do not yield qualitative events — or do not do so, at least as far as we have any reason to believe. Examples are (a) neural events in the visual system that give rise to colour constancy. The apparent colour of a patch in the visual field depends on the character of illumination, but that can be detected only from a visual field that is larger than the patch. What we see thus depends on a considerable amount of neural processing; but the earlier stages of this processing do not, as far as we have any reason to believe, yield qualitative events of their own directly — they do so only by causally contributing to later events that are directly correlated with visual experiences; (b) our brains control secretions to our digestive systems, contractions of digestive muscles, release of substances from glands, and so on. These processes are DEVELOPING DUALISM 165 [12] If ‘disjunctive properties’ were allowed, there would be no empirical assumption here — a ‘property’ would be guaranteed so long as there were some brain event property or other for each event that yielded a qualitative event. This understanding is to be excluded; here and throughout, ‘properties’ are to be understood as natural, non-disjunctive, projectible commonalities. [13] But see Schwitzgebel (2013) for dissent on this assumption. C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n complex, in their way, but have no correlates in consciousness, as far as we have any reason to believe. It is, of course, logically possible that there is no feature of brain events that is common to all of those that yield qualitative events and absent from all of those that do not. But it is not an unreasonable expectation that such a property will be found. And it is not beyond imagining that it will turn out to be some particular kind of complexity — one that will have as its species complex neural or field properties whose instances yield qualitative events of different kinds. It is that kind of complexity that will be the manifestation of CAS in neural events. It is also possible that the properties that distinguish occurrences that yield qualitative events will strike us as having a higher degree of complexity than events such as those in (a) and (b) in the previous paragraph. Work on the olfactory bulb in rabbits by Skarda and Freeman (1987; see Freeman, 2008) led them to the view that a sniffed odorant would push the olfactory system into the basin of a chaotic attractor distinctive of that odorant. If that view is right, a property that is common to all and only events that yield qualitative events would be a property that picked out a certain class of attractors. A perspicuous description of such a property would have to entail that its instances are very complex events, and it would have to differentiate those events from others that, in their own way, are also very complex. This paper is, of course, not committed to the view that the approach inspired by Freeman and colleagues is correct. Even so, their work can help us understand the possibility of a structure, involving both temporal duration and a large set of neurons, that has a distinctive form of exceptionally high intricacy. ‘A certain kind of simplicity’ likewise has to be taken seriously. The most fundamental manifestation of this kind of simplicity is the persistence of qualitative sameness through a temporal interval. This simplicity is on display by itself in olfactory experience. Smells can, of course, change fairly rapidly. One can get a whiff of something for a very brief time. But even a whiff has a noticeable duration; it is not a strictly instantaneous occurrence. And if one puts one’s nose up to a flower, one can have a sweet smell that lasts through one’s sniff. That is a kind of constancy, or stasis that persists through time. One’s neurons are engaged in complex firing patterns during the sniff, but the qualitative event is a continuum in time of the fragrance. The fragrance may have distinguishable components, and so not be absolutely simple. But (a) it is quite simple relative to the complexity of the 166 W.S. ROBINSON C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n neural or field events that yield it, and (b) each of the components shares the property of persisting as it is through a duration. All other qualitative events persist through durations, but some offer other species of the kind of simplicity through which CAS manifests itself in qualitative events. Visual events, for example, exhibit a spatial, as well as a temporal, continuity. Colour experiences are not collections of ‘colour points’ — they are collections of colour expanses. The auditory experience produced by tightening a string while bowing it is a glissando that exhibits a continuity of change of pitch. (Of course, that also involves temporal persistence, but the change itself is a simple continuity.) It may be suggested here that there are properties that have the required kind of relative simplicity, but are not instantiated in qualitative events. Now, if considerations of complexity and simplicity are to prove useful to a solution to the HP, this possibility will eventually have to be ruled out. It is, however, prima facie plausible that the kind of positive, continuous persistence that is found in qualitative events can be found uniquely in them. Our physical sciences typically progress by finding analyses that result in more, not less, complexity in the properties we attribute to the world. It is thus certainly imaginable that our descendants will continue to think that there is a distinctive kind of simplicity that pertains to the properties in qualitative events, and it will be assumed in what follows that they do so. CAS, then, is the property that manifests itself in brain events as the kind of complexity indicated, and manifests itself in qualitative events in the ways indicated. Because our science of the brain is not fully developed, we have only an imperfect understanding of the complexity in CAS, and we cannot even be sure that the empirical presuppositions of its coherence will prove to be borne out. This is one of the main reasons why we cannot now have a solution to the HP. But we do understand how our descendants may be able to form a better conception of the relevant kind of complexity of brain events. Let us imagine DEVELOPING DUALISM 167 [14] Consider a moving dot that changes colour part way through its journey (or, an illusion of this — the colour phi phenomenon). This case involves several species of continuity: simple temporal persistence, spatial spread of the colours, and continuity of motion. This severalness is one kind of complexity, but it must be distinguished from the special kind of complexity that yields qualitative events. Presumably, each aspect of continuity will be underlain by its appropriate species of the kind of neural complexity that yields qualitative events. Their combination in a single event accounts for the co-instantiation of several kinds of continuity, but it is being underlain by the appropriate species of neural complexity that yields each qualitative aspect. [15] There is a little more on this matter in the next note. Thanks to Bill Lycan for pointing out the need to consider this possibility. C op yr ig ht (c ) I m pr in t A ca de m ic 2 01 3 Fo r p er so na l u se o nl y -no t f or re pr od uc tio n that they have done so, and ask how they might proceed to build on