Similarly, it may seem natural to us that the omega-point intelligences, for reasons of historical or archaeological research, or compassion, or moral duty, or mere whimsy, will eventually create virtual-reality renderings of us, and that when their experiment is over they will grant us the piffling computational resources we would require to live for ever in ‘heaven’. (I myself would prefer to be allowed gradually to join their culture.) But we cannot know what they will want. Indeed, no attempt to prophesy future large-scale developments in human (or superhuman) affairs can produce reliable results. As Popper has pointed out, the future course of human affairs depends on the future growth of knowledge. And we cannot predict what specific knowledge will be created in the future — because if we could, we should by definition already possess that knowledge in the present.{7}
It is not only scientific knowledge that informs people’s preferences and determines how they choose to behave. There are also, for instance, moral criteria, which assign attributes such as ‘right’ and ‘wrong’ to possible actions. Such values have been notoriously difficult to accommodate in the scientific world-view. They seem to form a closed explanatory structure of their own, disconnected from that of the physical world. As David Hume pointed out, it is impossible logically to derive an ‘ought’ from an ‘is’. Yet we use such values both to explain and to determine our physical actions.
The poor relation of morality is usefulness. Since it seems much easier to understand what is objectively useful or useless than what is objectively right or wrong, there have been many attempts to define morality in terms of various forms of usefulness. There is, for example, evolutionary morality, which notes that many forms of behaviour which we explain in moral terms, such as not committing murder, or not cheating when we cooperate with other people, have analogues in the behaviour of animals. And there is a branch of evolutionary theory, sociobiology, that has had some success in explaining animal behaviour. Many people have been tempted to conclude that moral explanations for human choices are just window-dressing; that morality has no objective basis at all, and that ‘right’ and ‘wrong’ are simply tags we apply to our inborn urges to behave in one way rather than another. Another version of the same explanation replaces genes by memes, and claims that moral terminology is just window-dressing for social conditioning. However, none of these explanations fits the facts. On the one hand, we do not tend to explain inborn behaviour — say, epileptic fits — in terms of moral choices; we have a notion of voluntary and involuntary actions, and only the voluntary ones have moral explanations. On the other hand, it is hard to think of a single inborn human behaviour — avoiding pain, engaging in sex, eating or whatever — that human beings have not under various circumstances chosen to override for moral reasons. The same is true, even more commonly, of socially conditioned behaviour. Indeed, overriding both inborn and socially conditioned behaviours is itself a characteristic human behaviour. So is explaining such rebellions in moral terms. None of these behaviours has any analogue among animals; in none of these cases can moral explanations be reinterpreted in genetic or memetic terms. This is a fatal flaw of this entire class of theories. Could there be a gene for overriding genes when one feels like it? Social conditioning that promotes rebellion? Perhaps, but that still leaves the problem of how we choose what to do instead, and of what we mean when we explain our rebellion by claiming that we were simply right, and that the behaviour prescribed by our genes or by our society in this situation was simply evil.
These genetic theories can be seen as a special case of a wider stratagem, that of denying that moral judgements are meaningful on the grounds that we do not really choose our actions — that free will is an illusion incompatible with physics. But in fact, as we saw in Chapter 13, free will is compatible with physics, and fits quite naturally into the fabric of reality that I have described.
Utilitarianism was an earlier attempt to integrate moral explanations with the scientific world-view through ‘usefulness’. Here ‘usefulness’ was identified with human happiness. Making moral choices was identified with calculating which action would produce the most happiness, either for one person or (and the theory became more vague here) for ‘the greatest number’ of people. Different versions of the theory substituted ‘pleasure’ or ‘preference’ for ‘happiness’. Considered as a repudiation of earlier, authoritarian systems of morality, utilitarianism is unexceptionable. And in the sense that it simply advocates rejecting dogma and acting on the ‘preferred’ theory, the one that has survived rational criticism, every rational person is a utilitarian. But as an attempt to solve the problem we are discussing here, of explaining the meaning of moral judgements, it too has a fatal flaw: we choose our preferences. In particular, we change our preferences, and we give moral explanations for doing so. Such an explanation cannot be translated into utilitarian terms. Is there an underlying, master-preference that controls preference changes? If so, it could not itself be changed, and utilitarianism would degenerate into the genetic theory of morality discussed above.
What, then, is the relationship of moral values to the particular scientific world-view I am advocating in this book? I can at least argue that there is no fundamental obstacle to formulating one. The problem with all previous ‘scientific world-views’ was that they had hierarchical explanatory structures. Just as it is impossible, within such a structure, to ‘justify’ scientific theories as being true, so one cannot justify a course of action as being right (because then, how would one justify the structure as a whole as being right?). As I have said, each of the four strands has a hierarchical explanatory structure. But the fabric of reality as a whole does not. So explaining moral values as objective attributes of physical processes need not amount to deriving them from anything, even in principle. Just as with abstract mathematical entities, it will be a matter of what they contribute to the explanation — whether physical reality can or cannot be understood without also attributing reality to such values.
In this connection, let me point out that ‘emergence’ in the standard sense is only one way in which explanations in different strands may be related. So far I have really only considered what might be called predictive emergence. For example, we believe that the predictions of the theory of evolution follow logically from the laws of physics, even though proving the connection might be computationally intractable. But the explanations in the theory of evolution are not believed to follow from physics at all. However, a non-hierarchical explanatory structure allows for the possibility of explanatory emergence. Suppose, for the sake of argument, that a given moral judgement can be explained as being right in some narrow utilitarian sense. For instance: ‘I want it; it harms no one; so it is right.’ Now, that judgement might one day be called into question. I might wonder, ‘Should I want it?’ Or, ‘Am I really right that it harms no one?’ — for the issue of whom I judge the action to ‘harm’ itself depends on moral assumptions. My sitting quietly in a chair in my own home ‘harms’ everyone on Earth who might benefit from my going out and helping them at that moment; and it ‘harms’ any number of thieves who would like to steal the chair if only I went elsewhere for a while; and so on. To resolve such issues, I adduce further moral theories involving new explanations of my moral situation. When such an explanation seems satisfactory, I shall use it tentatively to make judgements of right and wrong. But the explanation, though temporarily satisfactory to me, still does not rise above the utilitarian level.