« Homo-gays & evolution | Gene Expression Front Page | Daniel Boorstin, R.I.P. »
February 29, 2004

The World Riddle

Warning: long.

In my post No Pain, no Gain I gave reasons for accepting the proposition:

“Some subjective sensations, such as pleasures and pains, are adaptive traits of organisms.”

The main reason is that pleasures are generally associated with biologically beneficial, and pains with harmful, circumstances, in such a way as to suggest that the association has evolved by natural selection. There are various apparent exceptions, but on closer examination most of these actually support the evolutionary explanation.

Prima facie this implies that these subjective sensations influence the behaviour of organisms, since otherwise they could not be selected. This also fits with our common-sense intuition that the intrinsic pleasantness and unpleasantness of sensations causes us to seek out or avoid them. If putting your fingers in boiling water produced the sensations of orgasm, there would be a lot of people with boiled fingers.

The idea that pleasure and pain have an adaptive function is not an eccentric invention of my own. It is probably what most biologists would take for granted (so long as they are not trying to be philosophers): for example, “Just as an itch can motivate defensive scratching, pain is an adaptation that can lead to escape and avoidance” (R. Nesse and G. C. Williams, Evolution and Healing, p. 35), or “Ordinarily, pain is a very useful adaptive mechanism - a gift, not a curse” (V. Ramachandran, The Emerging Mind, p. 18).

The snag is that the laws of physics leave no room for the subjective quality of sensations to make any difference to behaviour. The brain seems to be a purely physical system. Electrical impulses pass along nerve fibres, and at the junctions (synapses) between nerve endings in the brain, neurotransmitter chemicals are secreted, which, in sufficient quantities, stimulate the next nerve cell to ‘fire’. Everything seems to proceed in accordance with the laws of classical (non-relativistic, non-quantum) science, such as conservation of energy, momentum, and charge; Maxwell’s field equations, etc. These are all essentially deterministic laws. The state of the system at one time is in principle completely determined by its state at previous times. This leaves no gap in the chain of causation where the subjective quality of sensations can intervene and make a difference.

At this point, some defenders of ‘free will’ are inclined to appeal to quantum indeterminacy to open up such a gap. I don’t like this approach because:

a) it is just plain boring - almost as boring as the frequent misuse of Gödel’s Theorem in this context (but I may have a separate rant about that)

b) the basic units of nerve action - the discharges along the axon or across the synapse - are too large, involving millions of atoms, for quantum indeterminacy to come into play (see also Schrödinger: What is Life?/Mind and Matter, CUP, p. 92); and

c) in any event, what we want for the present purpose is not indeterminacy but a causal influence of a kind not recognised by purely physical laws. If you feel a severe pain, there is nothing ‘indeterminate’ about your desire to remove it.

If we can’t appeal to quantum indeterminacy, is there any alternative theory, consistent with the laws of physics?

A satisfactory alternative theory would have to explain why pain is associated with harmful circumstances, and pleasure with beneficial circumstances, without the subjective quality of pain and pleasure having any effect on behaviour.

I mention, only to dismiss, the logical possibility that we classify sensations as painful just because they are associated with harmful circumstances, and vice versa for pleasure. This would account for the association, but it is clearly not the true explanation. We can and often do classify sensations as pleasant or unpleasant without knowing whether they are harmful or beneficial.

Any satisfactory alternative is likely to be a dual aspect theory. Under such a theory physical systems are conceived of as having both their normal physical properties and the qualities we experience as subjective sensations.

Many philosophers, and some scientists (including Schrödinger), have favoured a dual-aspect theory of some kind. These range from radical panpsychism, which maintains that all physical objects have a ‘mental’ side, to a more limited doctrine of emergent properties, which maintains that ‘mind’ is a rare and special phenomenon produced only by very complex systems, and in particular by brains (perhaps even only human brains).

For the present purpose the important distinction is not between degrees of radicalism among dual aspect theories, but between those theories which ascribe causal efficacy to the ‘subjective’ aspect as such, and those which only admit the efficacy of the physical aspect. A theory of the latter kind maintains that pain or pleasure is an aspect or property of a physical system (a brain-state) such that the physical system itself, operating by purely physical laws, produces the behaviour normally associated with pain and pleasure (e.g. cries of pain; avoidance of painful stimuli; over-indulgence in sweet-tasting foods when available, etc.)

Among the limited number of scientists who express any views on these issues, I think that most would only admit the efficacy of the physical aspect. For example, John Maynard Smith (The Problems of Biology, pp. 79-80) says: “As I sit writing, I decide to go into the kitchen and make a cup of coffee. To me, it seems that my felt desire for coffee is the cause of my going to the kitchen... But to say that my desire for coffee is the cause is not to deny that there are physical structures in the brain that are the cause, or that those structures are present because of my previous experience. The physical events in my brain, and the feelings that I have, are not alternative and mutually exclusive causes of my actions: they are simply different aspects of the same cause. Of course, I assume that you have feelings similar to mine, and I also assume that animals which are reasonably like us have feelings that are like ours.” Although Maynard Smith is not explicit on this point, he presumably sees the ‘physical structures in the brain’ as operating purely in accordance with physical laws. He recognises the existence of subjective feelings (in this case, a desire for coffee) as ‘different aspects’ of events in the brain, but he gives them no work to do. Everything in the physical world would happen in the same way if subjective feelings ceased to exist, while leaving the brain states as they were.

Despite my great respect for JMS, I don’t think this is a satisfactory account. It fails to explain the association between feelings and circumstances. It is not even adequate for the case of ‘desiring coffee’, let alone more serious pleasures and pains. JMS recognises that the desire for coffee feels appropriate to the action of getting up and making a cup, but he does not explain why the subjective feeling ‘desire for coffee’ is associated with the physical brain-states which lead, by purely physical laws, to his getting up and making coffee, rather than, say, making tea, or jumping out of the window.

I cannot recall any modern discussion of the problem, within a ‘dual aspect’ approach, which even recognises the need for such an explanation, let alone provides one. Back in the 19th century, a few thinkers did make some vague gestures in this direction. Herbert Spencer, for example, argued that pleasure is the natural consequence of a healthy and not overloaded nervous system, while pain is the result of disintegration or excessive nervous activity. Ernst Haeckel argued that pain was the subjective aspect of physical repulsion (e.g. of electrical charges) while pleasure was the subjective aspect of physical attraction. While these explanations are crude and naive, they do have the merit of attempting to make a systematic connection between the physical and subjective aspects, and offering some superficially plausible suggestions as to the nature of that connection.

Unfortunately, nothing we currently know about the brain gives any reason for believing that there is such a simple and satisfying connection between the physical and subjective aspects of brain-states. The basic units of nerve action seem to be much the same whether the result is a pleasure or a pain, a taste or a smell, a perception or a mental image. The qualitative aspect of brain states seems to emerge only at the level of many nerve cells interacting in complex ways. It is conceivable that if and when we begin to understand how nerve action gives rise to subjective sensations, we will see that there is a simple and satisfying connection after all, which reconciles a purely physical account of causation with our subjective intuitions. I hope that we will, but I see no reasons for optimism.

In the mean time, I think that the balance of evidence points to the conclusion that the subjective aspects of brain-states do influence behaviour in ways not explicable by purely physical laws. Since the classical physical laws purport to give a complete account of physical events at the macroscopic level (including the behaviour of organisms), the causal efficacy of subjective sensations would actually conflict with the physical laws (though probably only at a small scale level within the brain). I am not happy with this conclusion, and I hope it is wrong.

I take some comfort from the fact that the physical laws are demonstrably incomplete with respect to the emergence of subjective sensations themselves. None of the known laws of physics is capable, even in principle, of explaining, for example, the sensation of colour (Schrödinger, op. cit., pp.165-68). Ultimately, all the laws of physics reduce to equations governing the position of entities in space-time. They can no more lead to conclusions about qualitative sensations than a set of propositions about dogs can lead to conclusions about cats.

It will be evident that I have not resolved the latent contradiction in the scientific world-view which I described in my first post on the subject. The relation between mind and body is one of the toughest problems in all of science and philosophy. Haeckel described it as the Welträtsel - The World Riddle - and I am not quite vain enough to suppose I can solve it. But some modest progress is made if the existence of the problem is recognised.

(Joke for British readers:) All right, I’ll get me coat.

I hope that this post has implicitly responded to many of the comments on the earlier posts. Some responses to more specific points are in the continuation.


In responding to comments, I won’t mention particular individuals, as I prefer to summarise the arguments in my own words, and I don’t want to be accused of misrepresenting people.


I have deliberately avoided using the word ‘consciousness’, because it has been used with so many different meanings, most of them obscure, that it is worse than useless. As a rule of thumb, if someone starts talking about consciousness, expect them to commit a fallacy within about five sentences.

Some people claim that ‘having a sensation’ and ‘being conscious of a sensation’ are different mental states. As far as I can judge from my own experience, there is no such distinction. I can attend to a sensation, and, as a language-user, I can refer to a sensation, either in public speech or in my ‘interior monologue’ of imagined speech. Unless this is what people mean by ‘being conscious of a sensation’, I don’t know what they do mean.

Then there is ‘consciousness of self’. Like David Hume, I find the ‘self’ elusive. I have perceptions and other sensations of my body, I can remember some of its previous experiences, and I can use the first-person singular pronoun ‘I’. Analytical philosophers are inclined to put language in the central role, so that the ‘self’ is the complex of linguistic uses associated with the words ‘I’, ‘me’, and so on. Language-using humans have a richer and more complex set of beliefs, concepts, and intentions about themselves than is possible for non-human animals and humans who have not acquired language (such as small children and deaf-mutes in olden times). This is important for many issues, but not for pleasure and pain. Will anyone argue
that, before she acquired language, Helen Keller was unable to feel pain?

Writers also often refer to the unity of consciousness. We normally have many different sensations at the same time (of sight, hearing, smell, mental images,etc.) which are all somehow within the same ‘field of attention’. Usually one mode of sensation is in the focus of attention while others are in the background, but we can switch our attention to any of them in an instant. They are all accessible to our powers of thought and verbal expression.

This is a genuinely interesting, important, and puzzling phenomenon.
Unfortunately, it often leads to the fallacious assumption that any mental
events that are not within the field of attention are ‘unconscious’, or that they do not exist as qualitative sensations (qualia) at all. But this does not logically follow from the evidence: it is like saying that because we have a lot of furniture in our main living room, we have no furniture anywhere else in the house! And the evidence from split-brain phenomena, blindsight, hypnosis, etc, strongly suggests that we can have sensations which are not accessible to the ‘verbal’ part of the brain but which otherwise have the same properties as those that are (see Chapter 8 of David A. Oakley (ed.) Brain & Mind). (There is also the neglected example of dreams. In one sense these are not ‘conscious’, but it would be absurd to say that they do not contain qualia.)

Man and other animals

I deliberately did not say much about non-human animals, because there has
been so much controversy about them (e.g. Donald R. Griffin, The
Question of Animal Awareness
). My arguments are expressed primarily in relation to humans, but by implication extend to other animals who resemble humans closely enough.

Since, with dubious exceptions, animals do not have language, they cannot
tell us what (or whether) they feel. There are however three main arguments
for believing that some non-human animals feel pleaure and pain in much the
same way as humans:

a) they have similar brains, nervous systems, and perceptual apparatus to humans

b) their behaviour in response to the stimuli that humans find pleasant or painful is often similar to that of humans; and

c) if pain and pleasure has an adaptive value, then we would expect it to have such value for many animals other than man.

(The last argument obviously will not be persuasive to those who do not accept that pain and pleasure has an adaptive value.)

I haven’t studied in depth the evidence on non-human capacities for pain, etc., (but have re-read relevant parts of Stephen Walker, Animal Thought, Griffin, op. cit. and Oakley (ed.) op. cit.), so I don’t have strong view on where the line should be drawn between those animals that feel pain and those that don’t. Mammals, yes; birds, yes; other vertebrates, probably (see Walker, p. 224); invertebrates - difficult to tell. With some exceptions, such as cephalopod molluscs, invertebrates have very small brains, and much more rigid, instinctual, behaviour patterns than vertebrates. So it wouldn’t surprise me if, say, an ant has no subjective sensations at all. But there is intriguing evidence that earthworms produce internal opiate secretions (endorphins), which in vertebrates are associated with pain control, so I don’t think we can be dogmatic about it. Fairly clearly organisms without a nervous system - plants, protozoa, bacteria - don’t have sensations, unless some form of panpsychism is true.


In commenting on my list of apparent exceptions to the adaptiveness of pains and pleasures, several people mentioned music (or the arts generally) as a case of pleasure-giving circumstances without obvious biological benefit.

I agree that ‘aesthetic’ pleasure is a genuine evolutionary puzzle - in fact, it is a puzzle regardless of the mind-body issue. Several thinkers (e.g. Geoffrey Miller, The Mating Mind) have sought an explanation in sexual selection, but I don’t find this convincing. I think it is more likely that aesthetic pleasure is a kind of reward for perceptual problem-solving, and that the human arts are a by-product of this pre-existing adaptation. (I think this is roughly Ramachandran’s theory.)

Does sensation have added value?

This leads me to the most interesting criticisms of my arguments. The main counter-argument (putting it in my own words), runs on the following lines:

A. We know that organisms can be attracted to favourable circumstances (e.g. food) and repelled from unfavourable ones (e.g. predators) without sensations of pleasure or pain. Even bacteria show such reactions, and nobody will suppose that bacteria feel pleasure or pain. Sensations are therefore not neccessary for adaptive reactions. Since they are not necessary, they have no additional adaptive value. Therefore they are not the result of natural selection. (And the rest of my argument collapses.)

A variant or supplement to this runs as follows:

B. The majority of the animal kingdom gets by perfectly well without ‘consciousness’ (or without ‘subjective sensations’, in my terms). ‘Consciousness’ appeared late in the course of evolution, with the higher vertebrates, or perhaps only the higher primates. But if it had the kind of adaptive advantage I claim, it would have been been useful much lower down the evolutionary scale, and we would expect it to be more widespread than it is. Therefore it cannot have this kind of adaptive advantage, etc, etc....

The first thing to say about these arguments is that neither of them directly challenges the premisses of my own argument. They do not deny the correlation, to which I have drawn attention, between pain/pleasure and harmful/beneficial circumstances, nor do they explain it.

I accept that adaptive reactions are possible without subjective sensations.
Even in humans, some reflex reactions protect us from harm without requiring subjective sensations, because they do not involve the brain.

That said, I don’t find either argument (A) or (B) compelling. The obvious non-sequitur in (A) is the sentence ‘Since they [sensations] are not necessary, they have no additional adaptive value’. Just because some adaptive reactions don’t require sensation, it doesn’t follow that this is true of all of them, or that sensation does not fulfil certain needs more effectively. Simply to assert that ‘unconscious’ tropisms can do everything that is done by sensations of pleasure and pain is to prejudge the whole point at issue. I hasten to say that no-one who commented on my posts committed such an explicit fallacy, but I think that if an argument of the kind (A) is spelled out in full, some such premiss is required to make it run.

Argument (B) rests heavily on the factual assertion that most animals don’t feel pleasure and pain, and for reasons given earlier, I dispute this. If the factual assertion is correct, the argument does have some force. As noted earlier, if pain and pleasure have an adaptive value, then we would expect them to have such value for many animals other than man. But even if the assertion is correct, it is not conclusive. It could be that the emergence of subjective sensations requires a large brain, and only the higher vertebrates (and maybe cephalopods) have evolved brains large enough. A notable feature of these large-brained animals is their capacity for learning and flexible behaviour. If argument (B) were sound, one could substitute ‘learning’ for ‘consciousness’ in the argument and conclude that learning has no selective advantage, which presumably no-one will accept. Indeed, it may be that subjective sensations themselves are an important part of the learning process, so that the correlation of large brains with both learning capacity and subjective sensations would then be a natural one.

Finally, I hope it will be agreed that this is not an area for over-confidence or

Posted by David B at 04:48 AM