There is a very old saying that is not possible to predict the invention of the wheel. Why not? Because by the time you have defined whatever it is you believe that you have failed to predict, you will have invented it. The description of the thing is the same thing as the invention of the thing. That is so, at least, where the “thing” is simple, but it is not true where the thing is complex and we might succeed in describing some of its properties despite having no idea how to bring it about.
We have the same problem with what is now usually called “superintelligence” because it is almost a matter of definition that any concept of intelligence that was greater than our own would, if we were to be able to understand it at all, necessarily need description. And we cannot describe something that we cannot imagine. Alongside the impossibility of reinventing the wheel would come the impossibility of imagining a new colour. We can imagine various combinations and shades and hues and tints and so forth of the colours we already know, but it is impossible to imagine a new one, and were anyone to claim to have done so they would find it impossible to verify unless they were able to produce an object or painting, or a piece of paper or something in which that new colour was instantiated.
So if we cannot imagine, still less invent superintelligence, but must just wait until it arises, might the same be true of being in a state of multiple consciousness? Our lives are spent – or so at least “the story goes” – in states of largely single consciousness with peripheral states of consciousness circling around of things that we are marginally aware of. To suggest that we be capable of a kind of multi tasking in consciousness in which we were able to be conscious of many things at once, each of them as fully as we are conscious of anything, would seem to invite a kind of schizoid life. And we would find it, we do find it – perhaps I should just say “I find it” – difficult, even impossible to imagine what that might be like. But there is no reason in principle why entities should be incapable of multiple consciousness, especially if consciousness is merely a by-product of the evolution of certain kinds of brain. If the brain is limited to one kind or one state of consciousness at a time, then it is at least conceivable that a more complex brain or machine might be capable of multiple consciousnesses. That we cannot imagine something is not a guarantee that it cannot be done or happen. Some positivists would claim otherwise, but they would quite simply be wrong.
The possibility of multiple consciousnesses raises the question of what it is that is important, even valuable, perhaps unique, certainly worthwhile about the one consciousness that we have already. We all know that we will die, that “none of us is getting out of here alive”, and that death will at least as far as we are concerned be not so much marked by as accompanied by the cessation of our conscious existence. Some people sadly and tragically lose their consciousness before the death of their physical bodies, but if I might be allowed the extravagance of calling it “under normal circumstances”, we carry our consciousness to our deathbeds, and at the moment of our physical passing our consciousness similarly ceases to be.
Those of you who have read other posts on this blog, and especially those who have followed the trajectory of my thought about the nature of the self and the solution to the mind/body problem, will be aware about I am a committed and even passionate dual-aspect theorist. I do not believe that there is a mind/body problem, or at least I believe that it is a problem that has been resolved for centuries, even millennia; since Aristotle, in fact. In essence: if an entity has a body with a particular kind of nervous system, it will be to whatever extent conscious, there will be a from-inside-looking-out element of that entity that is capable of interacting with its environment and with other entities. There is absolutely nothing mysterious about this: it is just a by-product of the evolution of certain kinds of nervous system and in particular the brains of human beings. Because there is nothing remarkable about it, there is no reason to suppose that it could not in some circumstances be improved upon or even completely superseded by either an evolved or an engineered future system. Whether such systems can be superintelligent or multi-conscious is of course not yet determined, but everyone that knows anything about this field fully expects what has now come to be called “the Singularity” to arise at some stage in the next 10 or 20 years. The Singularity is the emergence of a successor intelligence of sufficient power to take over the determination of its own and all future evolution. In other words it is an intelligence that will design all future intelligences and redesign or reconceive the very notion of what it is to be intelligent, and this leads some commentators to the rueful observation that the first AI we create that has genuinely super intelligent capacities will also probably be the last. And if you ask me to tell you what such a superintelligence might look like, I refer you to the impossibility of predicting the invention of the wheel. Although I think we are not compelled to remain completely silent on the subject.
What has this to do with the discussion that preceded it? Well, if our inside-looking-out-ness is our consciousness, that consciousness is not the entirety of our neurological activity, and there is a good case for regarding our consciousness as in some sense a metaphor for our full selves. In other words it reveals something of ourselves that says something about ourselves and that is indicative of greater truths about ourselves, the kinds of truths that one might associate with a statement along the lines of “he is the kind of person who…” where we try to predict someone’s behaviour from what we know about them. So the cessation of my consciousness along with the death of my physical body entails as far as we know only the the cessation of a subjective experience for me, and the cessation of an interactive experience for you insofar as you interact with me. In the case of multi-consciousness we would anticipate a similar cessation if the physical substrate died, ceased to function, malfunctioned, rusted, decayed or ran out of power. But during its existence a multi-conscious entity could interact differently with different people, and would one suppose be able to interact constructively with itself, rather like an advanced AI interacting with itself in order to play a better game of chess or Go. Under normal circumstances the multiple consciousness would exist in segregated worlds, but there could be moments when it shared its experiences, learned from itself, and generally as you might put it engaged in conference calls with all its other conscious selves in order to bring them into some kind of synchronicity.
Does this make any kind of sense? Is it an interesting or useful thought? The reason for raising it is that we are unduly hung up on the experience of consciousness, and it really requires huge efforts of concentration and thought to rid ourselves of or to diminish in someway the impact of that prevalent worldview on the way we see everything. Even if, as those of you who have read other things in this blog will know, we can rid ourselves of the pernicious notion of the soul, which constantly interrupts the flow of the conversation about what AI is capable of achieving, it is much more difficult to imagine a world in which what mattered was something other than consciousness. And this brings us to Derek Parfitt.
Those who know Parfit’s work, and most of us came to it through his monumental Reasons in Persons, have been struck by the brilliance of the thought experiment that he introduces in Part Three of that book, while at the same time being somewhat bemused by its apparent disconnection from most of the rest of the book and indeed the three monumental tomes On What Matters that succeeded it. Parfit died very early in 2017, I think on 2 January, and so we will never perhaps know what he really thought mattered,but what he appears to have been doing in his work, as I have only just come to appreciate, is systematically to rid ourselves of the delusion as he thinks of it or at least the misperception as we might more kindly call it, that our consciousness is what matters. Indeed the whole of Reasons and Persons is an attempt to debunk that motion not just for his reader, but perhaps more particularly for himself. Parfit repeatedly draws our attention to the fact that he is trying to make himself feel better about his own death, And he claims towards the end of the book to have achieved that by effectively recontextualising the importance of consciousness or perhaps one could put it more strongly decontextualising the importance of consciousness.
If one understands the task Parfit undertook as being in essence to rid ourselves of the notion that it is our conscious existence that matters about us, contrary to all our instincts and presuppositions, then the third part of Reasons and Persons not only makes sense but is the single most important thing that he ever wrote. What he shows there is that common assumptions about the prevalent centrality and importance of consciousness are untenable under certain conditions that he sets out in his thought experiments. Some critics would argue that the impossibility of realising his experiments in any practical sense renders them useless or inadmissible or impossible or unimportant or in irrelevant, but Parfit’s point was not to suggest the physical possibility of a telly transporter, it was to point out that the mere possibility in thought of such a device introduces contradictions into our notions of the self and particularly our notions of consciousness that render it inconceivable that it could be what we take it to be. In other words what Parfit is doing in Part Three of Reasons and Persons is demonstrating not only that the notion that consciousness is what matters needs to be abandoned and debunked, but that everything associated with that notion including that we take it to be the central feature of our lives that authenticates the importance of our existence, and therefore the things that we take to matter in pursuit of the fulfilment of existence as we conceive it, needs debunking and abandoning as much as the notion of consciousness itself.
But if consciousness needs debunking, would not multiple consciousness need debunking all the more? Only if we continued to carry forward into the multiple conscious existence I am envisaging the same kind of precious attachment to the subjective experience of consciousness that has bedevilled human existence since we first emerged on the planet. This notion has given rise to more of the evil in the world than any other. That I believe my experience of my consciousness to be the centrally most important and valuable thing in the universe leads me to favour myself over others in a way that cannot contribute to the net collective well-being of the species, or the planet on which it lives with so many other creatures. That being so, privileging of consciousness is the single greatest environmental disaster in human history. It is responsible for war, for cruelty, for insensitivity, for environmental rape, and for all the other evils that contribute to the misery that is human existence. We have almost literally no awareness of our corporate collective existence. We are as little like ants as it is possible to be. Nobody understands, still less can conceive, what a genuinely collective existence that privileged the well-being of the whole community might entail.
Some might argue that Marx tried to argue for a similar shift in our way of life, but the legacy of his attempt has been almost as destructive as the legacy of consciousness itself, so we must regards his attempt as a failure. Marx’s biggest mistake was to imagine that you could lose capitalism and instantiate communism or collectivism as it is better called, without tackling the way we privilege individual human consciousness, which is to say, he imagined that you could on ideological grounds introduce and sustain collectivism without addressing the very thing that makes collectivism not only impossible but inconceivable.
Marx’s mistake in other words was to think that people are capable of being benevolent and community-centred and, as I have called it before, other-centred, without a fundamental shift in the way they think about themselves, and the way they as a result seek to position themselves in the collective life. Perhaps the best-known Marxist principle, “from each according to his ability and to each according to his needs” embodies precisely the problem that he needed to address if his revolution was to succeed in a sustainable way: the problem of centring everything on the abilities and needs of individuals, which are the symptoms of their obsession with their own lives and especially their consciousness of their own lives.
So how are we to proceed? There is, after all, no future in this. If almost all humankind are obsessed with the prevalence of their own consciousness, and not at all interested therefore in collective well-being except insofar as it affects that self-consciousness, then there is no incentive for anyone to change anything. Parfit is aware of this problem. In fact he is so aware of it that he persistently deferred as a full discussion of it while he deals first with the vagaries of language in which he sees the linguistic habits that underpin our preference for consciousness so deeply embedded that he can conceive of no way to get rid of them. No way that is unless one first reformed language itself. This is what brings him to On What Matters, but I regret to say that I think he makes the same mistake as Marx, notwithstanding that from a distance one can see within his intellectual compass precisely the tools necessary not only to avoid the mistake but to put right, to rectify the mistake Marx made, and almost all others who want to be social reformers make, that is the failure to address the prevalence of consciousness and our consequent preoccupation with preserving it – more particularly our own – as the central and single most important feature in human existence. (It would be interesting to research the extent to which this failure on Marx’s part is attributable to his Hegelian background.) Having convinced himself that there is nothing much to be worried about in his own death, Parfit seems to move elsewhere, when what he should really set about doing is persuading everyone else of exactly the same thing, for without that there can be no progress. Our own awareness of ourselves and our own consciousness leads us to a passion for the preservation of consciousness. That is the single most pernicious habit that human beings have and is responsible for everything that goes wrong in the world.
There might be two kinds of multiple consciousness: serial and parallel. Serial multiple consciousness is easy to imagine: we would just switch from one consciousness to another rather as a CPU in a computer slices up its time to manage different threads. Parallel multiple consciousness is much harder, if not impossible to imagine because we are so obsessed – I nearly said single-mindedly preoccupied, which of course is just the point – with the experience of consciousness rather than with its significance.
Can there be a consciousness we are not aware of? Is that not a contradiction in terms? Well, perhaps, but let us stay with the idea: could we – could our brains – really be more like GPUs than CPUs, with our experienced consciousness assigning different tasks to different GPUs and leaving them to process in the background? In that case the “blocks” of the GPU, or different GPUs in a more complex architecture, would act rather like consciousnesses we were not aware of, processing things as they would be processed by consciousness, but without us being aware of them.
This is far more plausible than it sounds. We are not, contrary to much that we might imagine, aware of thinking: we are aware of thoughts as the output from thinking, rather like the output from a neural net the working of whose hidden layers we know little or nothing about. Thoughts are the things that emerge in words and pictures and sounds and sensations and so forth; thinking is not something we are usually conscious of. Thinking takes place “in the background” and can sometimes absorb enormous resources, so much so that we find our conscious existence stressful and barren while we wait for the results of the background processing. My subjective experience of creativity is that it is a mixture of really unpleasant barrenness when nothing seems to be going on, but which experience tells me is just the waiting before the revelation, punctuated by moments of blinding insight when suddenly everything gets kick-started again and I feel alive. Were there to be no such instances there would be very little to live for. If this distinction between thinking and thought comes as a surprise, imagine the instruction “think harder!”: how do we do that? We can’t, not directly; all we could hope to do would be to devote more input resources to whatever background processes might work on them and hope for the best. We have almost no control over our thinking.
So it is ironic when we hear pundits worrying about what AI may do, especially that we are even now beyond the point where anyone really understands what advanced AI is doing on a cumulative global basis, but we are in the same position with our own brains. “The one thing that I cannot know is the next thing that I shall think.” Back to predicting the invention of the wheel: our brains will do what they will do; we can only control the input data upon which they work, not how they work on it. Brains are not like neural nets: we cannot redesign them to match new and different tasks, but we can educate them to be more proficient at more things, providing the resources needed for them to work along parallel lines of consciousness that may (and probably will) one day surprise us.
On Being Many
Conventional psychology regards people whose live their lives with many different affective states as suffering from a mental illness such as MPD – multiple personality disorder – but this is a prejudice unless that multiplicity of selves creates problems dealing with everyday existence. Unfortunately, this same conventional psychology leads us to expect people to be, so-to-speak, seamlessly and recognisably themselves. “She wasn’t herself today” is used of someone who is in some sense out of sorts, but the phrase also betrays an assumption: that we should be recognisably ourselves, and our one self.
There is no reason why this should be the case; in fact, there is every reason why it should not. If different strands in our thinking that are embedded in different “parts of” or processes in our brains, result in the emergence of different ideas and behaviours, then there is every reason to expect that we will not always appear the same and every reason to suppose that we might from time to time appear different. The world may expect us always to present it with the same “self”, but that is no reason why we should, and it is perfectly possible that we are incapable of meeting such an expectation. Only grant this and a great deal that currently troubles us about the way people behave suddenly becomes less surprising and, as a result, less alarming; only grant this and we will learn an important lesson about life: that people are seldom entirely what they seem to be since most people are from time to time living these different lives.
An interesting question is then what triggers our switching from one personality to another, and how and whether we can control those changes. We should not be too quick or ready to diagnose someone who is inconsistently themselves as mentally ill: they may instead be several perfectly healthy versions of themselves, each sharing the same life in the same body. Our preoccupation with the continuity of our subjective experience of consciousness makes us reluctant even to consider such a possibility, but Parfit’s thought experiment enables us to see that there is not, in practice, any continuity of self, and that the notion of continuity of self is an illusion, perhaps even a delusion.
Consider for a moment the version of the teletransporter thought experiment where the machine leaves my body intact but creates several copies. As I walk towards the machine, knowing that this will be the outcome of walking through its archway, I may perfectly well be troubled – that I am troubled and should not be is precisely Parfit’s point – by the question which of the entities that emerges from the machine will be continuous with the one entering it; in other words, which of these three future bodies will have the experience I call being “me”? All, none, one or some?
And it is the unanswerability of this question that gives Parfit’s experiment its cutting-edge and shows that the experience of consciousness is not what we imagine it to be, and certainly not the locus of our personality. Neither is it permanent, so the notion that my self consists in that continuity must be incoherent. Parfit’s point – although it is a difficult point to assimilate, still less accommodate – is that the notion of my own death is not more to be feared that the duplication entailed in the experiment depicted above: that I am not able to decide which of the entities will be “me” suggests that none of them will be “me” any more than the others. And if none of them will be “me” any more than any of the others, then the notion that any of them could genuinely be me must be an illusion. a fortiori, since none of them being me makes no difference to me now, the cessation of my being me that I fear in death makes no sense either. What I want to “hand on to”, my “me-ness”, by virtue of establishing the continuity between one of these future selves and my current self, cannot be “hung on to”. All I can say is that each of the three entities beyond the machine will momentarily believe itself to be me and have the instantaneous memory of having passed through a machine with a history beforehand that is my history. Thereafter their histories will diverge, first by a little and later by a lot as they live their necessarily divergent lives.
It is not enough to dismiss this thought-experiment as irrelevant because it cannot be achieved. That it can be conceived is sufficient to show that the notion of “my” future continuity with myself is untenable: the notion that there is a “me” that will be “in” one of these bodies but not the others makes absolutely no sense unless I retain belief in a soul (but see “Mind: the Gap” on this blog and other essays if you are still in thrall to that illusion). Parfit’s point is that there is no such “me”; there is only contemporary awareness and a sense of the past. Each of the three entities that leaves the machine will have that sense of contemporary awareness and sense of the past; each of them will therefore be “me” without any of them having been privileged by being the recipient of some unique quality, still less “thing”, that contains and constitutes “me”.
The absolutely infuriating question – at least, the question that absolutely infuriates me about the thought-experiment, if nobody else – consists in the incommensurability of this state of affairs. I habitually live my life on the assumption that “I” will enjoy the fruits of my labours and nobody else except insofar as I decide to share them: I study, work, think, struggle, take care of myself, save and plan; that three entities should be in a position to inherit the rewards for my efforts seems faintly improper (the question of the division of my assets is interesting but should not delay us here); that “I” might not be any one of them any more than any of the others seems more bizarre still; that “I” might be all of them seems impossible, multiple consciousnesses notwithstanding.
Parfit’s point is that the everyday, commonplace assumption that the self who wakes up in this body tomorrow morning will necessarily be “me” and the same me who is writing these words now is a conceptual mistake. It is the same mistake as imagining that only one of the three entities leaving the teletransporter will be me; it is far more coherent to imagine that all of them and none of them will be me. A perfectly good objection to this would consist of saying “Yes, but there will only be one body waking up tomorrow morning capable of thinking itself to be ‘you’, so where’s the problem?” The problem is that there is no more reason to grant that body “me-ness” than any single privileged body emergent from the teletransporter: the subjective experience of being me is neither a guarantee of continuity nor a requirement for it. What the body that wakes tomorrow morning will share with the three bodies emergent from the machine is a sense of being me and a sense of being a me who wrote something about this last night; nothing requires them to have continuity with that “me”; the person who walks into the teletransporter could as easily be evaporated with no impact on the thought experiment. The notion that it is being in possession of precisely this body that makes and enables me to be “me” is an unwarranted prejudice: being me consists in no more than being an entity with access to a history defined by a certain kind of relation with a narrative.
Somehow – this, too, is an infuriating thought – I feel that “I” am entitled to have access to the subjective experiences of me-ness enjoyed by all three of the entities since I am the one who has worked so hard to create the person whose subjective states they will all enjoy. And if I cannot experience all of them, why should I have any confidence that I will experience any of them? If the notion that it is the “I” writing this now who will have any of these experiences makes no sense, then what sense does it make to suggest that “I” will have any future experiences, teletransporter or not? And Parfit’s conclusion is that it makes no sense; indeed, that is precisely his point: we should not be concerned about our death because it cannot deprive us of anything other than the illusion that we will be denied future experiences of being “I” by it. Duplication or triplication will similarly deny us such experiences. Indeed, one feels a sorites paradox coming on: were there 1,000 emergent bodies exactly like mine, I could not enjoy the subjective experience of being all of them; ergo, I could not enjoy the subjective experience of being all of 999 were there only 999; ergo, … of two; ergo of one. There is no justification for believing that “I” will enjoy even the experience of being the one self whose subjective experiences this body supports; ergo I should not be concerned about my own death. But even Parfit struggles to persuade himself of that.
It follows that, if my subjective experience of my supposedly single consciousness is not the same thing as being me, nothing is lost if I am able to experience multiple consciousnesses no one of which is the entire me and each one of which is in some sense a complete me. Indeed, one might think nothing could possibly be more satisfactory: I get to experience all kinds of different possible states of being without the need ever to leave home.