(OO-L): This chapter considers Hume's proposal that we are made up entirely of particular mental states and events: the bundle view. An argument for the bundle view is based on the claim that the traditional idea of substance is dismissed. The bundle view is then shown to follow naturally from widely held claims about diachronic and synchronic personal identity. Reid's objection that bundles of thoughts cannot be thinkers is elaborated and endorsed. It is then argued that the bundle view cannot easily avoid the thinking-animal problem. There follows a critical discussion of two related views: that we are bundles of universals1 and that we are something like computer programs. Both are found to be hopeless.
- Bundle theories
- Traditional arguments for the bundle view
- Personal identity and the bundle view
- Can thoughts think?
- Thinking animals2 once more
- Bundles of universals3
- The program view
6.1 Bundle theories
6.2 Traditional arguments for the bundle view
- We have now considered a number of views according to which we are material things: organisms, things coinciding materially with organisms, and spatial and temporal parts of organisms. I cannot think of any other promising materialistic view of what we are4. So let us turn to views according to which we are not material.
- One such view is that we are composed of mental states or events: particular beliefs, desires, sensations, emotions, and so on. In particular, each of us is composed of all and only his own mental states or events. Our parts may include both occurrent states or events – things actively going on within the mind, such as your current philosophical cogitations – and non-occurrent states and dispositions lying dormant, such as your memories of last summer and your taste in furniture. Or our parts may be particular mental qualities or “tropes”. But none of our parts are material things. We are not made of matter. Though our bodies may be made of matter, the parts of our bodies are not parts of us. Call this the bundle view.
- We need to distinguish the bundle view from other claims that sound similar. One is that our parts include both mental states and a material body that is not itself composed of mental states. Because this suggestion has no obvious attraction over the bundle view, and most of what I will say about the bundle view applies equally to it, I won’t discuss it separately.
- More importantly, a number of claims besides the one that concerns us have been called “bundle theories”. First, there is supposed to be something called the bundle theory of the mind: the view that the mind is nothing but a bundle of mental states. What this means depends on the meaning of the word ‘mind’. If 'mind' means 'thinking being', then we are minds (since we are thinking beings), and the view that our minds are bundles is the view that we are bundles. But those who assert that minds are bundles are sometimes merely making a claim about the nature of our mental lives, namely that they are not unified in the way that certain philosophers have thought. This is not a view about what we are5. In fact it looks consistent with any view about we are.
- Another bundle theory says that all concrete objects, and not just ourselves, are composed of particular states or qualities. This “global” bundle theory is not the same as the bundle view. Someone could hold either view without holding the other: someone could say that we are bundles but that unthinking objects such as trees are not, but rather traditional substances; or someone might think that all concrete objects are bundles but deny that we exist. I will argue in §6.5, though, that friends of the bundle view are better off with a global bundle theory.
- A third sort of bundle theory says that concrete objects are composed not of particular states or qualities, but of universals6. The view is roughly that Kilimanjaro (say) is composed of a certain height, a certain shape, a certain geological structure, and so on – not a particular height, shape, and geological structure that necessarily belong to that mountain alone, but the very same height, shape, and so on that other mountains might share. Applied to ourselves, the view would be that we are composed of psychological universals7. We will consider this view in §6.6.
- Then there is the view that we are "logical constructions" out of mental states. Ayer once wrote, "We know that a self, if it is not to be treated as a metaphysical entity, must be held to be a logical construction out of sense-experiences" (1946: 125). (By “a metaphysical entity” Ayer meant an immaterial substance, and he thought had shown talk of immaterial substances to be meaningless.) What does it mean to say that we are logical constructions? The phrase 'logical construction' is a tricky one. To say that Fs are logical constructions out of Gs sometimes means that Fs are sets, in the mathematical sense, built up out of Gs. But Ayer did not think that you and I were sets. (The idea that we might literally be abstract objects that have their members essentially is hard to take seriously.) As Ayer used the term, a logical construction is not a kind of thing at all. We cannot say that, among the things that there are, some are logical constructions and some are not. That would be like saying that among the things that there are, some are real and some are unreal – no one thinks that there are leprechauns, and that they belong to a kind called “unreal objects”. When Ayer said that we are logical constructions out of sense-experiences, he meant that all statements that appear to be about ourselves, or about thinking beings generally, could be translated without loss of meaning into statements that make no reference to thinking beings, but only to sense-experiences (1946: 63). When we say, “Paul heard a noise,” what we are saying is equivalent to some longer and more complicated statement that refers to or quantifies over nothing but mental states.
- This is not the view that we are composed of experiences. (Ayer is explicit about this: 1946: 127f.; see also Pike 1967.) The logical-construction view does not say that we have thoughts as parts. In fact it says nothing at all about our metaphysical nature. It is not a metaphysical claim at all, but rather a claim about meaning – specifically the meaning of "person talk". In fact it is unclear what it could mean, on the logical-construction view, to ask what we are8. If all so-called talk of people is equivalent in meaning to talk of things other than people, what could it mean to ask what sort of things people are?
- If the logical-construction view suggests any metaphysical view, it is that there are really no people, but only things other than people – the mental states or what have you that figure in Ayer’s translations of person talk and which presumably account for the appearance of there being people. If there were people – if our thoughts had subjects – then surely our talk of people would refer to or quantify over them, and thus would not be synonymous with statements that refer to or quantify only over mental states. But this is to put a metaphysical gloss on an anti-metaphysical view. On the logical-construction view, the statement “there are no people” can mean only that there are no mental states (or whatever it is out of which people are "logically constructed") of the appropriate sort – which of course logical constructivists deny. To ask about the metaphysical nature of people, they say, is to misunderstand the meaning of person-talk. It is like asking about the metaphysical nature of sakes: to wonder whether there is such a thing as Kolya’s sake, and if so what sort of entity it might be, is to misunderstand the meaning of the word ‘sake’. ‘Kolya’s sake’ is not an expression that purports to refer to anything. Ayer is saying that neither is ‘Kolya’. To use another analogy, asking what the logical-construction view says about the metaphysical nature of people is like asking what logical behaviorism says about the metaphysical nature of mental states, or what phenomenalism says about the metaphysical nature of physical objects.
- Most of this book rests on the assumption that the logical-construction view is false. I could say a good deal about why I think it is false, but this is not the place for it. In any case it is not an account of what we are9. The bundle view I want to consider says that we are bundles of mental states and events.
- Because we think, it follows that bundles of mental states think: the subject of my thoughts is something composed of those very thoughts. Now it is hard to understand how a bundle of mental states could think (see §6.5 below), and someone might suggest instead that we are bundles that don’t think. I can see no attraction in this “unthinking-bundle view”. No one who takes us to be unthinking bundles will suppose that other things think our thoughts. That is, no one will suppose that we don’t think our thoughts, but other beings – things that are not even parts of us- -do think them. The unthinking-bundle view could be true only if nothing thinks our thoughts: that is, if thoughts occur, but have no subjects or thinkers. But if nothing thinks our thoughts, does it not follow that we don’t exist? First-person singular statements, such as that I am now awake, refer to their subject, the being that makes them and whose thoughts they express. If there is no such subject, they don’t refer at all. And if the word ‘I’ never refers to anything, then there is no such thing as I, just as if the word ‘Atlantis’ never refers to anything there is no such thing as Atlantis. (Assuming, anyway, as the unthinking-bundle view does, that ‘I’ is a referring expression in the sense of §1.4.)
- Might someone suppose instead that we exist but don’t think? That our personal pronouns refer to unthinking beings? That would be strange. But it would be even stranger to suppose that they refer to bundles of thoughts. If personal pronouns and proper names referred to something, but never to thinking beings, what unthinking beings would they refer to? Maybe some things would be better candidates for their reference than others. Given that we say such things as “Thatcher is a human being” and “Thatcher is sitting down”, it would be absurd to say that ‘Thatcher’ referred to a bicycle or a tree. If anything, we might expect it to refer to a human organism – one that according to the unthinking-bundle view is for some reason unable to think. That’s the sort of thing it appears to refer to. But why suppose that it refers to a bundle of thoughts?
- In any case, the unthinking-bundle view is not really an answer to our question. Although it answers the question, What do our personal pronouns and proper names refer to?, it does not answer the question, What sort of beings think our thoughts and perform our actions? Indeed, it assumes that these two questions have different answers. And I said in §1.4 that the second question was to have priority over the first should the two diverge. The interesting idea behind the unthinking-bundle view, as I see it, is that nothing thinks our thoughts. We will come to that suggestion in due course.
- Let us return now to the bundle view that concerns us, the claim that we are composed of mental states. This is not a complete account of what we are10. Though it implies that we are concrete particulars, that we are not substances, and that our parts are particular mental states, it does not say which or what sorts of mental states are parts of us and why, or which properties are essential to us, or whether we persist through time. We will come to these questions shortly. Still, the idea that we are made up entirely of thoughts tells us a good deal about what we are11, enough to rule out many rival accounts. It stands in stark contrast with the other views we have considered. That is is enough to work with.
- The bundle view is counterintuitive for some of the same reasons as the brain view is. It implies that we never strictly see ourselves or each other, and that we are wholly invisible and intangible: you can no more touch a bundle of thoughts than you can touch a dream. So it is perhaps unsurprising that it has few defenders. Hume proposed that each of us is "nothing but a bundle or collection of different perceptions" (1978: 252) – though even he found it hard to believe. Quinton says that each of us is "a series of mental states connected by continuity of character and memory" (1962: 398; see also 1973: 97-105), and I have already mentioned Rovane's claim that a person is "a set of intentional episodes" (1998: 172; see also Campbell forthcoming). I suspect, however, that the bundle view has a large underground following. I will argue that a number of popular views about personal identity support it. First, though, I will briefly review some traditional arguments for the bundle view.
6.3 Personal identity and the bundle view
- The bundle view was once considered the obvious alternative to our being simple immaterial substances (by Hume and Quinton, for instance). Arguments against substance dualism were taken to support the bundle view. Nowadays we know better.
- Others argued for the bundle view (and for the global bundle theory more generally) on the grounds that the very idea of a substance is incoherent. This reasoning usually began with a certain picture of what substances are supposed to be, something like this:
To perceive an object is to perceive its qualities: shape, size, motion, and so on. Some philosophers suppose that there must be more to a thing than just its qualities: there must also be something that stands under and supports them. They posit something called a substance to play this role. Their view is that an ordinary thing like a cat is made up not only of the furriness, the feline shape, and the sinuous movements that we perceive, but also of a substance in which those qualities inhere. Yet even those who say this admit that we never perceive the substance itself. How could we? The substance is by definition something apart from its qualities, and therefore incapable of characterization. It is not furry or feline or moving. It has no qualities at all; it only supports the qualities we observe. And the mere fact that the substance supports certain qualities tells us no more about how it is in itself than the fact that something supports certain books tells us how it is in itself. It is a mere “something, we know not what”. A substance is therefore a mysterious theoretical entity: a metaphysical abstraction of the most dubious sort. We might tendentiously call this the Lockean picture of substance (Campbell 1990: 4-11 is a recent example). If this is what substances are, sensible philosophers will have nothing to do with them. What could we be, then, if not substances? Bundles of qualities, presumably. There is little else that we could be, on the Lockean picture. We needn't be bundles of mental qualities only: we might be made up partly of brute physical qualities as well. But some sort of bundle view will be almost inevitable.
- The core of the Lockean picture is the idea that what is furry or shaped or moving in the strictest sense is not a substance, but a particular quality or trope. That leaves the substance with nothing to do but fix the qualities in place: it stands to them much as a lump of soft clay stands to colored feathers stuck into it. So the qualities the substance supports do not characterize it, but merely clothe it. If there is anything the cat’s qualities characterize, it is the bundle of those qualities.
- I reject this picture root and branch. The idea of a thing uncharacterized by any qualities makes no sense to me. But I don’t agree that the thing the qualities characterize – the thing that is furry or moving – is itself a quality, or a bundle of qualities. What I understand by 'substance' bears little resemblance to this absurd picture of a bare particular clothed in qualities. As I see it, a substance is not a metaphysical abstraction, but an ordinary thing: the cat that is furry and moving. The cat is not a compound made up of qualities and the substratum that supports them. Nor is it made up of qualities alone. It is made up of other substances: cells, molecules, and atoms. What makes it a substance is not that it is “something apart from its qualities”, but that it is not itself a state or a quality of something else. It is not qualities that we perceive, but substances.
- A substance is not "incapable of characterization": to characterize something is precisely to say what qualities it has. That is at any rate the anti-Lockean picture I was taught. It provides nothing like a complete theory of substances, and leaves plenty of hard questions unanswered. But I see nothing wrong with it. In particular, I see no reason to accept the Lockean picture, or the traditional arguments for the bundle view that presuppose it.
6.4 Can thoughts think?
- To my mind, the most serious argument for the bundle view (the bundle view of ourselves, not the global bundle theory) has to do with personal identity. Since the time of Locke, philosophical orthodoxy has assumed that personal identity is grounded in psychological facts. No account of what we are12 fits better with this assumption than the bundle view.
- Consider the view that you could move from one human animal13 to another via Shoemaker’s "brain-state transfer" procedure. No substance, material or otherwise, thereby moves from one animal to another (the ontology of temporal parts aside, anyway). What does move? If Shoemaker's description of the process is correct, it transfers the particular mental states realized in the first animal's brain to the second animal's brain. And if it is possible to move you by moving nothing but mental states, the natural conclusion is that you are composed of mental states (Campbell forthcoming).
- More generally, many philosophers think that our identity through time consists entirely in facts about mental states or events. What is necessary and sufficient for a person x existing at one time to be identical with something y existing at another time, they say, is for the mental states x is in at the first time to stand in certain relations – causal ones, perhaps – to the mental states y is in at the second time. In fact the very question of personal identity over time is sometimes stated as what is necessary and sufficient for mental states or events occurring at different times to belong to the history of a single person (Grice 1941; see also Perry 1975: 7- 12). Now what sort of thing could have its persistence determined entirely by facts about relations among mental states? Well, something composed of mental states could. It is doubtful whether any concrete object made up entirely of things other than mental states could survive or perish just by virtue of relations among mental states. The obvious conclusion is the bundle view. (Or perhaps the view that we are made up partly of mental states and partly of something else. Again, I take this to be an uninteresting variant of the bundle view.)
- Or one could argue for the bundle view from considerations about what determines how many of us there are at any one time. We might wonder whether the result of cutting the neural connections between the cerebral hemispheres would be two people in one body, or whether there might be several people “inhabiting” a single human organism at once in an extreme case of multiple personality. A deeper question is what sort of facts would settle the matter. What would make it the case that there were two people sharing a single human animal14? Or is it possible at all? For that matter, what makes it the case that there is just one person associated with an ordinary human animal15, and not two or more? Many philosophers say that the answer lies in psychological facts: just as (they say) facts about psychological continuity16 over time determine whether we have one person or two in “diachronic” cases, facts about psychological unity at any one time determine how many of us there are in “synchronic” cases. They say that simultaneous mental states belong to the same subject if and only if they are in some sense unified. The reason why the mental states of an ordinary human animal17 are all the thoughts of a single person is that they are unified in the right way. But if they were sufficiently disunified, they might be the thoughts of two different people.
- There are different accounts of what this unity amounts to. Kant thought that what made something a mental state of a particular being, and hence a mental state at all, was that being's ability to combine or synthesize it with its other mental states – to unite those states, as he put it, "in one self-consciousness" (1929: B134). More recent accounts exploit the fact that many mental states are disposed to interact in special ways with other mental states. It is characteristic of desires, for instance, to interact with beliefs to produce action: roughly, your desire for something tends to cause you to act in ways that you believe will satisfy it, unless you have stronger competing desires. That seems to be part of what it is for something to be a desire – and also part of the nature of belief. The claim that the entire nature of all mental states consists in such facts about their causal dispositions is the core of the functionalist theory of mind. That theory is controversial. But few would dispute that these causal roles are at least part of the nature of many mental states.
- It appears to follow from this that many mental states necessarily come in packages: in order for something to be a desire, for instance – as opposed to a memory or a visual sensation or something non-mental – it has to occur as part of a network of beliefs and other states with which it is disposed to interact, directly or indirectly, in characteristic ways. Call such packages mental systems. (Let us not confuse matters by calling them minds. You and I are minds in the sense of thinking beings; but we don't want to conflate this platitude with the contentious idea that we are mental systems. For what it's worth, I find the word 'mind' as a count noun unhelpful in thinking about personal identity.) Mental states belong to the same mental system if and only if they relate causally to one another and to actions (or are disposed so to relate) in the right way. A mental system is something composed of mental states related to one another and to certain actions, but not to anything else, in this way.
- Now mental systems match up pretty well with people, or thinking beings generally. If I want an orange, and believe that there is an orange in the bag and that it is within my power to get it out of the bag, this will ordinarily result in my attempting to do so, unless I take that to be incompatible with some other goal of mine. My desire for an orange will not combine in this way with anyone else's beliefs to cause action. Ordinarily all and only the mental states of a given thinker will be parts of a single mental system. Where there is more than one independent mental system, there is ordinarily more than one thinker. So it is tempting to suppose that being parts of the same mental system is what it is for mental states to belong to the same thinking being. This is Shoemaker's view:
It is only when the belief that it is raining and the desire to keep dry are co-personal that they tend (in conjunction with other mental states) to lead to such effects as the taking of an umbrella; if the belief is mine and the desire is yours, they will not directly produce any joint effects. And it seems that if a belief and desire do produce (in conjunction with other mental states) just those effects which the functional characterizations of them say they ought to produce if co-personal, then in virtue of this they are co-personal....Whether mental states...should count as belonging to the same person, or mind, would seem to turn precisely on whether they are so related that they will jointly have the functionally appropriate sorts of effects. (1984: 94; see also 1997: 294) The claim is that, necessarily, for every person or thinking being (“mind”), there is exactly one mental system, all and only the elements of which are the mental states of that being; and for every mental system there is exactly one thinking being whose mental states are the elements of that system. Mental systems and thinkers must match up one to one. This is the psychological individuation principle of §2.9.
- This principle may sound attractive, and Shoemaker is not alone in advocating it. It is also closely connected with his explanation of why organisms are unable to think (see §2.5 above), giving opponents of animalism18 another reason to look favorably on it. And those who accept the orthodox view that psychological facts determine our identity over time may find it natural to suppose that psychological facts also determine how many of us there are at any one time.
- The psychological individuation principle suggests that thinking beings are themselves mental systems. How could the number of mental systems necessarily fix the number of anything but mental systems? Remember: the number of mental systems is determined entirely by causal relations among mental states and actions. And it is hard to see how causal relations among mental states and actions could entail both the existence and the precise number of things that are not even partly made up of mental states or actions. But if thinking beings have mental states among their parts, then you and I have mental states among our parts, which is a version of the bundle view.
- It is especially hard to see how the psychological individuation principle could be compatible with our being material things. (I assume that no material thing has mental states as parts.) Any material thing that could have mental properties at all, it seems, could be mentally disunified. Nothing could guarantee that the mental states of any material object must be psychologically unified in the way that Kant and Shoemaker demand of a mental subject. Think of an extreme case of multiple personality, in which many of the usual interactions among mental states break down. This would be a being whose beliefs, desires, perceptual states, and so on don’t interact in the usual way to produce actions, any more than your mental states interact with mine to produce action. They would not form a unified mental system. Even if there is enough interaction among the mental states of a human organism in any actual case of multiple personality for them to form a mental system, this doesn’t seem to be a necessary truth.
- Suppose there really were such psychological disunity within a human being. If every mental state must belong to a mental system, these disunified states would have to belong to different mental systems. In that case there would be two or more unified mental systems associated with one human organism at once. According to the psychological individuation principle there would therefore be two or more thinking beings – two or more people – “sharing” that animal. Could those people be material things? Well, what sort of material things could they be? They would have to be physically different, else there would be nothing to explain their mental differences. (No materialist would suppose that two people could be physically identical, with the same surroundings and history, yet differ radically in their mental properties.) It seems that each would have to be a different part of the animal’s brain. But that presupposes thinking-subject minimalism; and as we saw in Chapter 4, minimalism faces no end of trouble19.
- So the psychological individuation principle looks incompatible with our being material things. It appears to rule out any account of what we are20 apart from the bundle view. If this is right, it has important implications. It is not only an argument for the bundle view, but an argument against materialism generally, based on a widely held principle about personal identity. It means that you can’t be a materialist and at the same time accept everything we are inclined to say about personal identity. This is especially inconvenient for those materialists – usually advocates of the constitution view – who argue against animalism21 on the grounds that it is incompatible with our convictions about personal identity. It now turns out that their own view is incompatible with those convictions. Being a materialist is harder than it looks.
6.5 Thinking animals24 once more
- I hope I have shown that there is something to be said for the bundle view. Let us now look at it more critically.
- To my mind, the most forceful objection to the bundle view is expressed in this quotation from Reid:
I am therefore [according to Hume’s bundle view] that succession of related ideas and impressions of which I have the intimate memory and consciousness. But who is the I that has this memory and consciousness of a succession of ideas and impressions? Why, it is nothing but that succession itself. Hence, I learn that this succession of ideas and impressions intimately remembers and is conscious of itself. I would wish to be further instructed whether the impressions remember and are conscious of the ideas, or the ideas remember and are conscious of the impressions, or if both remember and are conscious of both....This, however, is clear, that the succession of ideas and impressions not only remembers and is conscious, but that it judges, reasons, affirms, denies – nay, that it eats and drinks and is sometimes merry and sometimes sad. If these things can be ascribed to a succession of ideas and impressions, in a consistency with common sense, I should be very glad to know what is nonsense. (1940: 378) As I understand him, Reid is objecting to the idea that a bundle or “succession” of thoughts should think or act. There may be such things as bundles of thoughts, but it is a metaphysical blunder to suppose that such things are the subjects of the thoughts that compose them. Reid doesn’t say why it is a blunder: he takes the claim to be patently absurd once we set it out clearly. That seems to me to be right. (Even philosophers sympathetic to the bundle view have agreed: see for instance Pike 1967: 163.) But can we say more to help those not yet convinced?
- Well, consider the idea that a particular thought might think that very thought. Might your belief that it’s cloudy believe that it’s cloudy? Could your love of hot curry love hot curry? Does your dream of white horses dream of white horses? Surely not. Even hardened bundle theorists will accept that. Nor can your dream of white horses believe that it’s cloudy, or have any other thought. If we know anything, we know that thoughts don’t think – just as games don’t play and dances don’t dance.
- Now, could one hold that something composed of many thoughts might think, even if no individual thought can22? The mere fact that a thing is composed entirely of parts that don’t think needn’t prevent it from thinking. A thinking thing could be composed of unthinking atoms, for instance. (So it seems, anyway.) How might a thing made up of nothing but unthinking parts think? Well, it could think if its parts cooperated to produce thought: if they each did something other than thinking, and these individual sub-psychological activities came together to add up to an act of thinking on the part of the being they compose (see §8.4). It is hard to imagine any other way in which something composed of unthinking things could think. But could individual thoughts cooperate to produce thinking? Might they each contribute something sub-psychological, so that these contributions added up to a mental act? It would seem not. Individual thoughts are acts of thinking. They are not sub-psychological ingredients of acts of thought. And combining lots of acts doesn’t give you an actor. Combining lots of games doesn’t give you a player. Combining lots of thoughts doesn’t give you a thinker. The idea that bundles of many thoughts might think seems no more credible than the idea that an individual thought might think.
- Sensible as this may reasoning may sound, not all philosophers are convinced. Some deny that there is any real distinction between states and events on the one hand and the things that are in those states or that participate in those events on the other – between thoughts and thinkers, or dances and dancers, or games and players. “Physical objects,” wrote Quine, “conceived thus four-dimensionally in space-time, are not to be distinguished from events or, in the concrete sense of the term, processes23”. If there is no distinction between physical objects and events, or between actors and acts, then there is no obvious absurdity in the claim that dreams might dream, beliefs might believe, or bundles of thoughts might think.
- The idea that both bundles and individual thoughts think seems to imply that every thought has at least two thinkers, namely the thought itself and the bundle it is a part of. That would leave it unclear what could make it the case that you were a bundle rather than an individual thought, or how you could ever know. I don’t suppose anyone tough-minded enough to deny the distinction between actors and acts is going to lose any sleep over this. But it certainly is messy.
6.6 Bundles of universals30
- Suppose for the sake of argument that a bundle of thoughts really could think. What about the thinking-animal problem – the problem of how we can know that we are not those human animals25 that appear to think our thoughts?
- Well, what sort of things would animals be if the bundle view were true? Would they too be bundles of states or events or qualities? Or would they be something else – substances, perhaps? I think friends of the bundle view will want to say that animals are bundles. If the human animal26 located where you are were a substance, we should expect your mental states to be states of it (or perhaps states of your brain; in any case they would be states of some substance or other). If there are substances as traditionally conceived, then some states, anyway, will be states of those substances. That’s what substances are for. But if your thoughts are states of an animal, that animal ought to be a subject of those thoughts. It ought to have them. It ought to think. What could be the difference between being in a state of belief and believing, or between being in a state of hunger and being hungry? And if animals think our thoughts, yet are non-bundles different from ourselves, the thinking-animal problem arises once more.
- Here is another argument against a “mixed” bundle view. Anyone who says that we are bundles but organisms are substances will want to deny that organisms and other substances think. Why wouldn’t substances think? Presumably because it is impossible for them to think: substances could have non-mental properties, but never mental properties. Something about mental properties would prevent substances from having them. That would open a deep metaphysical gulf between the mental and the non-mental. It would not be substance dualism, the view that some substances have mental properties and others have physical (or non-mental) properties and nothing could have both. It would be an even more profound sort of mind-body dualism: not only could no substance with physical properties have mental properties, but no substance of any sort could have mental properties. The boundary between the physical and the mental would be the boundary between substances and non-substances.
- Bundle theorists are unlikely to find this attractive. Nor does the bundle view provide any support for it: the claim that thinking things are bundles rather than substances in no way suggests that material things should be substances rather than bundles. If the subjects of mental states are bundles, we should expect the subjects of non-mental states to be bundles as well. Let us suppose, then, that not only we ourselves, but human animals27 too are bundles of states or events or qualities. More generally, all ordinary concrete objects are bundles: there are no substances as traditionally conceived.
- But this would not yet solve the thinking-animal problem. If animals are bundles of states, which states make up your animal body? We should expect them to include those physical states and activities we attribute to you: your height, your mass, the activities of your digestive system, and so on. But wouldn’t they also include your mental states? If the non-mental activities going on within the animal are parts of the animal, why shouldn’t the mental activities going on within it be parts of it too? And if the animal has thoughts as parts, how can it fail to think? You think, according to the bundle view, by having thoughts as parts. Why shouldn’t the animal also think by having thoughts as parts? If the animal digested in virtue of having digestive activities as parts but didn’t think in virtue of having mental activities as parts, that would again be a metaphysical dualism of the mental and the non-mental; and we should want to know what makes the mental so different from the non-mental.
- One could get round this problem by saying that we are those animal bundles: we are not bundles of mental states only, but bundles composed of both mental and non-mental states. That would mean that we are animals. This would solve the thinking-animal problem, all right. However, it would raise the problem of how we know we are not thinking bundles of mental states. And wasn’t the whole point of the bundle view to offer an alternative to our being animals? Bundle-theoretic animalism28 would have no obvious advantage over the usual “substance animalism”29.
6.7 The program view
- I will say no more about the view that we are bundles of mental particulars. What about the view that we are bundles of mental universals31? The idea is that I am composed of such properties as believing that it’s cloudy and feeling hungry – properties that I might share with others. In particular, I am composed of those psychological properties that I have or instantiate. Now I have a hard time understanding this. I have enough trouble thinking about universals32 by themselves. The idea that I myself might be made up of nothing but universals33 – abstract objects not strictly located in space and time – sounds to me like the sort of thing that comes to one in a dream after eating too many oysters. But I will venture a few brief remarks.
- It is not obvious what advantages the “universal bundle view” has over the particular bundle view. If it is hard to see how a bundle of mental particulars could think, it is even harder to see how a bundle of mental universals34 could think. And if a bundle of mental particulars would have no grounds for believing that it was not an animal, a bundle of mental universals35 would seem to face the same difficulty.
- The universal bundle view also has troubles of its own. Start with the “problem of distinct discernibles”. There could be someone else psychologically just like me. It is of course enormously unlikely that anyone else has a mental life exactly like mine, right down to the smallest detail – so unlikely that we can be confident that it isn’t the case. But it doesn’t seem absolutely impossible. It would be like winning the lottery a million times in a row without cheating, rather than like winning the lottery without buying a ticket. It would not be possible, however, if I were a bundle of psychological universals36. For then my doppelgänger and I should be composed of the very same universals37. Our parts would be not merely exactly similar, but numerically the same. And surely the same universals38 cannot compose two different objects. At least it looks impossible to me – though my lights are dim in these regions39.
- Another problem is that I have different mental properties at different times. I once believed that I was 20 years old. I used to like bubble gum. There was a time when I had never heard of George W. Bush. Not any longer. But a collection of universals40 cannot be composed of different universals41 at different times. The collection composed of universals42 A, B, and C cannot come to be the collection of B, C, and D, let alone the collection of D, E, and F. These can only be three different collections. So it seems, anyway. Which collection of universals43 might I be, then? If I am the collection of A, B, and C, I must always be that collection. I cannot come to be a numerically different collection, for the simple reason that one thing cannot come to be another, numerically different thing. But if I must always be the collection of A, B, and C, then presumably I must always instantiate those universals44: I must always remain just the same and can never change. Yet if I know anything, I know that I do change.
- Universal bundlers may reply that I am composed of temporal parts (O’Leary Hawthorne and Cover 1998: 208). If I first instantiate the psychological universals45 A, B, and C (and only those), then later instantiate B, C, and D, and still later C, D, and E, then these three bundles – A-B-C, B-C-D, and C-D-E – are each temporal parts of me. Every bundle, all and only the elements of which I instantiate at some time, is one of my temporal parts. I am a bundle of bundles of universals46.
- But this looks wrong, for reasons that have nothing to do with objections to the general idea that we are composed of temporal parts. Imagine that A-B-C, B-C-D, and C-D-E are my only temporal parts. And suppose that C-D-E is a later part of me than B-C-D, and that B-C-D is later than A-B-C. (It belongs to the idea of temporal parts that one can be in some sense later than another – though what could make one bundle of universals47 earlier or later than another is not obvious.) Now it seems possible for someone to instantiate the same psychological universals48 as I do but in a different order. For instance, someone could start by instantiating C, D, and E, then instantiate B, C, and D, and finally instantiate A, B, and C before ceasing to exist. His career would be just like mine only in reverse. Someone’s career could also be just like mine but rearranged in a more complex way. The universal bundle view appears unable to account for this possibility. It implies that such a person would have the very same parts as I have, making us both qualitatively and numerically identical. But surely we should be both qualitatively and numerically different.
- Technically minded philosophers can no doubt think of solutions to these problems; but they are unlikely to hold much attraction, apart from their ingenuity. Those drawn to the idea that particulars are composed of universals49 are probably better off with the logical-construction view of §6.1. They could say that statements about people and other particulars are true if and only if certain universals50 relate in such and such a way, without saying that particular things are actually composed of universals51 – much as phenomenalists say that statements about physical objects are true if and only if certain facts about sense-experiences hold, without saying that physical objects are composed of sense-experiences. Whatever its merits, though, this is not a view about what we are52.
- One final thought that has some affinity with the bundle view is that we are something like computer programs. We stand to our bodies or our brains as computer programs stand to the physical machines they run on. We are not made of matter, or of particular states or qualities. We’re not made of anything particular at all. Call this the program view.
- The word 'program', like many expressions, has a type-token ambiguity. For instance, the English language has some 300,000 words – that is, word types. If you have to submit an essay of not more than 5000 words, however, this means word tokens: if the word ‘and’ occurs ten times on page two it counts as ten words, not one. Just as we distinguish word tokens from word types, we need to distinguish the particular copy of the word-processing program Mariner Write 3.6.2 now running on my Macintosh from the type or universal of which it is an instance. The copy I am using to write these words, if there is such a thing, is presumably some sort of concrete electronic event or state. It has a fairly definite location, changes over time, and will cease to exist when I shut down the computer or erase it from the hard drive. But when we speak of the word-processor Write 3.6.2, as when we speak of the word ‘and’, we don't seem to be referring to any particular, changeable thing located at a particular time and place, but rather to the universal of which the particular copies are instances. In any case, it is computer programs as universals53 that interest us here. The idea that we are computer-program tokens would be a version of the bundle view. I want to consider the view that we are universals54.
- Let us not confuse the view that we are computer programs with the so-called computational theory of mind. This is roughly the claim that human cognition is a computational process: a matter of manipulating symbols according to mechanical rules. The program view may entail the computational theory of mind, but the computational theory does not entail that we are programs; in fact it has no obvious implications at all about what we are55.
- It is hard to find an explicit endorsement of the program view in the philosophical literature. This quotation from Dennett comes close:
If [as Dennett urges] you think of yourself as a center of narrative gravity..., your existence depends on the persistence of that narrative..., which could theoretically survive indefinitely many switches of medium, be teleported as readily (in principle) as the evening news, and stored indefinitely as sheer information. If what you are is that organization of information that has structured your body's control system (or, to put it in its more usual provocative form, if what you are is the program that runs on your brain's computer), then you could in principle survive the death of your body as intact as a program can survive the destruction of the computer on which it was created and first run. (1991: 430) The program view is also a common theme in science fiction, and I have found philosophers attracted to it when pressed in conversation.
- Here are some considerations that look like arguments for the program view. In his story "Where am I?" (1978), Dennett invites us to suppose that we could survive the complete destruction of our brains if the information encoded there were “downloaded” into an electronic computer. What sort of thing could literally be transferred by wire from a human being to a computer? It is tempting to say that no concrete object literally moves from brain to machine; all that really happens is that the machine comes to instantiate or realize the informational state that the brain first instantiated. All that is first “in” the brain and then “in” the computer is some sort of universal. And the universals56 that computers are designed to instantiate or realize are programs. So the view that you could be downloaded into a computer might suggest that you are a program. Those who accept the possibility of resurrection or reincarnation but hesitate to accept Cartesian dualism may find themselves drawn to the program view for a similar reason.
- Or consider fission cases. We don’t want to say that transplanting57 each of your cerebral hemispheres into a different head would necessarily destroy you. We want to say that both resulting people would be you. Yet there are two of them and only one of you, and two things cannot be one thing. And we don’t want to say, as the temporal-parts view does, that there were really two of you all along. Friends of the program view can say that both offshoots are you even though there was only one of you to begin with. They cannot, of course, say that each offshoot is numerically identical with you, but they can say that each is you in the sense of being a concrete instance or token of you: both could be you in the way that the tattered paperback on my shelf and the calf-bound volume on yours are both Moby Dick. Where there was previously only one instance of you, there are now two. What if one of the offshoots has a beard and the other doesn’t? Do you have a beard, or don’t you? Well, if you are a universal, you can’t strictly have any physical feature – beard, nose, sunburn, or what have you- -but speaking more loosely we can say that you have a beard insofar as one instance of you has a beard strictly speaking, and that you have no beard insofar as another instance of you has none. Asking whether you have a beard would be like asking whether Moby Dick has a torn cover. No account of what happens in the fission story is very attractive, and someone might prefer this story to the alternatives.
- Finally, consider artificial intelligence. Many people think that it is possible in theory to build and program an electronic computer in a way that would produce thought as genuine as our own. It is no accident that this is commonly described by saying that computer programs may one day be intelligent. There is something odd about saying that computers – material objects made of metal and plastic – may one day be intelligent. If we were to produce an intelligent being by programming a computer, it would seem wrong to say that we had made the computer itself intelligent – that what was previously an ordinary desktop workstation had now acquired the ability to think. When the machine on which the crucial program was first run finally lands in the dustbin, no one would point to it and say that for a few exciting hours that piece of hardware was once intelligent. Nor would the intelligent thing seem to be a concrete electronic state or event going on within the computer at some particular time, or for that matter a material object the computer temporarily constitutes. It seems wrong to say that shutting down the program or erasing it from the computer's data-storage devices would destroy an intelligent being. If anything there is intelligent, it might be more natural to say that it continues to exist as long as the relevant information is still stored somewhere or other. And it would be tempting to say that the intelligent thing we had created could be stored on CDs and run on different machines, just like the word-processor Write 3.6.2. It is tempting to say, in other words, that the subject of artificial intelligence – the artificially intelligent being – would be a program.
- But if artificially intelligent beings would be computer programs, we should expect naturally intelligent beings such as ourselves to be programs too, or at least things of the same ontological kind as programs. If an electronic computer of the right sort (the right hardware, the right programming, the right surroundings, and so on) could “realize” an intelligent universal, then a biological organism of the right sort ought to be able to realize an intelligent universal too. How could the fact that the underlying physical processes are electronic and artificial in the one case and neurochemical and natural in the other make a difference to the metaphysical category of the resulting intelligent being? And if each normal human organism is the home of a thinking universal, it is hard to avoid the conclusion that we are those universals58.
- Despite these arguments, however, there are grave problems with the program view. For one, it is very hard to say which program, or which universal more generally, you or I would be. This human organism – my body – now instantiates all sorts of universals59. If it instantiates any program, it probably instantiates a vast number of them. And it is hard to see what could make just one of those programs intelligent and sentient. But if there are many intelligent, sentient programs “running on my biological hardware”, what could make it the case that just one of them was me?
- Even if it should turn out that there is a computer program, the running of which is uniquely responsible for my current mental life, it would be doubtful whether it has always run on my brain. Was the very program responsible for my current mental life responsible for it in my infancy? It is more likely that my brain ran a different program then. If brains run programs at all, they are constantly being reprogrammed. But I cannot be numerically identical with different programs at different times. So if I were the intelligent program now running on my brain, I should not be the program that ran on my brain when I was a child, or for that matter the program that will be running on my brain in a year’s time. My existence would be brief. Or perhaps I should be eternal and timeless, and my career as an embodied human person would be brief. Neither view is very attractive.
- Most obviously, universals60 don’t do anything. They don’t act. They don’t change. It isn’t the program type Write 3.6.2 – the program that you too can use on your own Macintosh – that converts my keystrokes into text, but rather a particular, local instance of it. Nor does Write change when I install it on my machine, or when I start it up or shut it down. Only the concrete instance of it running on my machine changes. Or rather, the program changes only in the way in which the number twelve changed by ceasing to be the number of apostles when Judas hanged himself: it undergoes only “Cambridge change”. It doesn’t undergo any real, intrinsic change. Computer programs as universals61 are inert and immutable.
- But I am not inert or immutable. If I know anything, I know that I am writing these words- -words that would not be written were it not for my actions. (Even if Descartes’ evil genius is deceiving me and I am not really writing, I am still doing something.) I know that I sometimes feel tired, sometimes hear the sound of the wind, sometimes wish I were somewhere else – and sometimes don’t. It couldn’t be the case that what seems to be real, intrinsic change in me is really only Cambridge change – that when I seem to grow more tired all that really happens is that something else – a particular human organism that instantiates me – grows more tired, while I myself remain ever the same, like the number twelve. If I am a universal, I don’t think or act- -not really. The thinker of these thoughts and the author of these words is not a universal. But if there is a concrete thing that thinks my thoughts and performs my actions, and an immutable universal that can be said to think and act only in the loose sense of having a concrete instance that think and acts, isn’t it clear that I am the concrete thing?
Footnote 19: Some of those troubles might be mitigated somewhat by combining minimalism with the psychological individuation principle. Such a combination might be worth exploring, though I doubt whether the end result would hold much attraction. I argue at greater length for the claim that the psychological unity principle rules out our being material things in Olson 2003.
Footnote 22: From here to the end of the Section, the text differs radically from the book.
Footnote 23: 1960: 171. It is not obvious that the ontology of temporal parts leads inevitably to this conclusion.
Footnote 39: Zimmerman 1998 is a useful discussion of some of these matters.
Text Colour Conventions (see disclaimer)
- Blue: Text by me; © Theo Todman, 2018
- Mauve: Text by correspondent(s) or other author(s); © the author(s)