Can Science and Information Technology Replace Philosophy?


Dr Donald Cameron, BRLSI Convenor

5 April 2016


Taking an overview In philosophy, as in other subjects, a difficult question sometimes becomes a little easier when it is approached from another direction, when the question is formulated in a different way, or when everything is taken away and we start afresh from first principles. Perhaps a little blue-sky thinking might help! The information that seems to have been too little considered by philosophers is our knowledge of our evolution. This was excusable before 1859, but after that, it is not. If it is true that we have been formed from the most primitive life forms by the action of natural selection and no other organising force (and I have very little doubt that it is true) then that should tell us something about our theories of knowledge and our ideas of ethics.

[At this point, I conducted a poll. A asked everyone to show who accepted Darwin’s Theory of evolution as being more or less correct as an account of the origins of life and human life in particular. Almost everyone raised a hand. I then asked for those against and only one hand was raised and this was by a Church of England vicar.] No respectable academic today denies evolution. The majority of people who do, tend to be the religious who are defending a structure of belief which is important to them rather than being driven by the evidence. [At this point the vicar pointed out that he was perfectly respectable. I know him well and it is undeniably true. Perhaps he has disproven the first sentence of this paragraph while confirming the second!]

Yet it is astonishing that, despite almost total academic acceptance, evolution is ignored by many writers on philosophy. How could it not be important to understand how we know things when the apparatus that we use to know things has been constructed to a design coded in our genes? And these genes have been created by sifting and selecting from random mutations over millions of years by only one criterion: those that are more numerous in each generation are those whose parents left the greatest number of descendents. Predictions can seldom be made with confidence, even for short future timescales, but I can give you a prediction that will be true in one million years’ time. If any life remains on this planet, its genes will be being sifted and selected by only one criterion: those that are more numerous in each generation are those whose parents will have left the greatest number of descendents. It is also important for us to consider information flows in the manner of the computer systems designer; where different types of information could possibly come from and where they could be used.

Of course there is an unavoidable circularity. If we use a conclusion of science to inform us about how conclusions of science might be reliably achieved, the danger is obvious. Yet there is no better way. The alternatives of using gut feelings, or accepting the fashionable beliefs of the majority, or of trusting beliefs of the ancients more than our own direct observation, or of holding to the belief that makes us feel good, have all been convincingly demonstrated to be much less reliable. The problem for philosophy is that natural selection had no reason to manufacture a brain to discover truth about the world. Neither had it any reason to produce a brain which would make its user into a perfectly saintly altruist, or able to define a universal standard of right and wrong.

Decision Making Natural selection does not select according to any external criterion. It simply selects for the best qualities for survival and reproduction of the underlying genes. It did not make a machine to discover either factual or moral truth and it does not “care” about these things. It made a decision maker and programmed it to make decisions such that its user would improve its chances of transmitting the genes that formed it to the next generation. Decision making devices abound in nature and not all could be described as brains. The humble earthworm has modest information processing abilities, but these are enough to be useful. It has no eyes, but it can sense the presence of light. When it does, it produces the response of digging downwards to get away from it. This reaction has probably evolved because an earthworm on the surface has a much higher probability of becoming dinner for a bird. Of course, I do not suppose that the worm has any concept of birds or other predators. It does not need to. It decision making system contains the program that says when light is detected, dig. That is enough to greatly improve its chances of leaving baby worms to carry its genes forward in time. Many of our human impulses are no different. We just feel them without any thought of any ultimate purpose. Most humans today do not reflect that the many complex impulses that they feel have been produced solely to improve the chances of transmitting genes to the next generation in the ancestral environment.

Let us consider the essentials of a decision making device (DMD), as shown in the following diagram: Every DMD has the same basic layout. Information is taken by the senses from the environment. A processor uses this information to select from the range of possible outputs to give a result that is more “desirable” than a random action would be. What is desirable is defined by a utility function pre-programmed into the DMD and is internally stored. It is obvious that this must be pre-programmed; the DMD cannot decide how it would like the world to be, simply from information about how the world is. This is a principle well known to philosophy as Hume’s law or the naturalistic fallacy. Many examples of DMDs exist. Human and animal brains are an example, but there are others. The data processing system of a guided missile is a non-biological example. It takes in information and computes what is required to achieve its pre-programmed objective, which is to accurately strike on the target. The result of the computation is used to move the steering fins. The guided missile differs from most living DMDs because the goals programmed into it do not include self preservation! ]

Another extremely simple DMD is the steam-engine governor. In this device a couple of weights are rotated by a drive taken from the engine. As they speed up, they rise and close off the supply of steam and as they slow down they descend and admit more steam to the engine. The governor is a simple DMD which has a fixed response to the input speed of rotation. As with other DMDs it cannot derive its goals from its inputs from the environment; in this case these come from the engineer who designed it. And, of course, it has no understanding of why it is doing it, it only says “if I am going faster then less steam; if I am going slower then more steam”. It does not even know that there is a steam engine. The principle of goals or utility functions being internally programmed into DMDs is universal. The criterion for decisions must be built into the DMD somehow. It cannot be obtained from the input information. The simple DMDs we have discussed abound in nature. They produce a fixed response to a recognised stimulus and they have evolved because they are useful to their owners in the universal goal of life – to survive and reproduce. They involve no understanding of why they are there, but that is not necessary to fulfil their function. Many, although not all, of the parts of our own human brains and nervous systems work on this principle.

But more sophisticated versions are possible. When memory is added the DMD can achieve a greater result in terms of its inbuilt objectives. The results of previous actions can be remembered and the successful ones repeated. In its simplest form, this is simply the conditioned response as studied by Pavlov, but its value for survival and reproduction is easy to understand. If an animal finds that food supplies are often found in a certain locality it will pay to revisit it. Equally if something bad, such as an encounter with a predator, occurs, it would pay to avoid that place. A further level of mental capacity can be seen in humans and in many other animals; that of model building, or forming theories. These are explanations or understandings that are, in effect a mental model of reality which can be interrogated to give information. This information is not immediately evident from the observations, but is nevertheless derived from them. It can greatly improve the performance of the DMD in terms of its built-in objectives. Our inbuilt mental abilities – theory of knowledge We tend to take our model-building abilities for granted. A young child, or even a dog, knows that when an object disappears behind an obstacle that it still exists and can be found by looking behind it. We make a mental model of the obstacle and the moving object.

Part of it is deductive logic. We just know that if all leaves are green and what I have here is a leaf, then it must be green. This form of logic is correct, even though, in this particular example, the first premise does not happen to be true. We just know, without testing, that a square peg will not fit a round hole. We just know that a statement cannot be both true and not true at the same time. Deductive logic is something of which we are completely confident and yet no one attempts to prove it. When logic is formally studied, its basis is not proven from any prior axioms. The rules of logic are themselves the axioms which can be explored and errors in their use highlighted, but their origin is taken for granted. Yet this logic is a function of our brains. It is not there by chance; other things such as stones, trees or cans of baked beans cannot do it (well, perhaps the trees can do it in a very limited sense). But the only basis we have for assuming it right is to hope that evolution has installed the correct rules in our heads. Then there is the skill of induction. We just know that if every swan we have ever seen is white, then probably the next swan we see will be white. Induction is not completely reliable (and, in fact, it is true that there are some black swans). But it is often true that if A has been followed by B every time in my experience, then when I see A again, I should think it probable that B will follow. Of course, there is no logical guarantee that it will, but it remains a most important element in our intelligence.

In everyday life we use it all the time and we would not survive for long without it. There are many counterexamples. The one I like best is about the turkey that forms the reliable opinion that the farmer is a nice caring man who brings food at regular intervals. Then one day in December, he kills it and sells it to make someone’s Christmas dinner! Scientists studying the nervous system have shown that we use induction much more than we might suppose. The eye has a “blind spot” where the optic nerve enters the retina and the brain’s visual system “fills in” missing parts of the picture so that we are mostly unaware of it. We do it in our recognition of other things too. For example, if we have a text with a few letters missing or changed, we can usually understand the meaning as long as the degradation is not too extreme. I am reminded of the three old ladies taking a walk out of doors. One says, “it’s windy isn’t it?” The second replies, “No it’s Thursday”. The third then says, “So am I, let’s go and have a drink”. With poor hearing, they minds were having to apply induction. Even where certainty or near certainty is not possible we make our decisions based on a subjective evaluation of probability. Our abilities here are probably sufficient to cope with situations in which most of our evolution occurred, but the modern mathematics of probability and statistics has shown that we are far from perfect.

The organisers of professional betting schemes and lotteries would not find it so easy to separate the gullible from their money, if we had all studied probability theory. We use these abilities to build “models” or explanations of reality. We can claim to understand a topic when we have built our observations into a mental structure from which we can make predictions. We then test our prediction against outcomes and, if it works, we use the model to make further predictions. Although we sometimes get it wrong, this capacity is immensely useful to us in achieving our objectives. But we can never escape from uncertainty. Most theories are the product of induction and the best we can ever say about those is that they have not yet been falsified. And on top of the weakness of induction, we have to hope that the rules of logic installed in us during our millions of years of evolution are correct. Science is no more than an organised attempt to avoid error. Its methods, in particular to look directly at the evidence, to be sceptical of unsupported testimony, to experiment and to collect data systematically have been spectacularly successful. Revision is often required in the light of new observations, or new model building, but that is its strength compared to those who cling to ancient dogmas.

So there is our theory of knowledge (or epistemology, if you are partial to pretentious words). It is not perfect, but at least we know that we are fundamentally limited in the certainty that we can claim about what we know. That is as good as it is going to get, so we may as well get used to it. Values – goals – ethics –utility function But we must not forget that we are not knowledge-discovering machines; we are decision-making machines. Our ability to acquire knowledge only evolved so that we could use it, imperfect as it may be, to take actions which would make it more probable that we would achieve more in terms of our built-in objectives. So let us now consider these objectives. One of the failures of philosophy is that it gives consideration to ethics, our purposes in relation to others, but little attention to our aims for ourselves, the actual purpose of our lives. It simply assumes that our desires for personal benefit are too obvious to merit enquiry.

I am reminded of W.H.Auden who said, We are here on earth to help others; what the others are here for, I don’t know. Our consideration of the information flows in all decision-making devices leaves little doubt that the objectives cannot come from the environment – they must be built in. And for living things, their objective can only be what is selected by natural selection itself – to survive and reproduce. It is easy to visualise that natural selection would install certain values in an animal’s brain. A desire to eat when the stomach is empty; a desire for sex; a fear of death or injury; a love of nearest and dearest, a desire to live successfully in a community; all of these we observe in ourselves and other animals. Of course, a decision maker can have multiple goals, but in any given circumstance it can only produce one decision. This means that there must be a way of deciding which of the many sub-goals will take priority at any time or what the relative weighting will be between them. We observe in this in ourselves too. And, of course, an animal does not need to understand about any underlying reason for these instincts; it is not aware that the ultimate purpose of the instincts that it feels is to do what it takes to get its genes naturally selected. And this too, we can observe in ourselves.

This leaves us in a position where some may find it uncomfortable. It means that your purpose in life is no more than to be a machine which exists to survive, reproduce and transmit your genes into the future. All your hopes and aspirations, all your feelings for family and friends, all your loyalties to causes, all your appreciation of the arts, all your spiritual experiences, all of your ethical feeling towards others are no more than the playing out of instincts that have been installed in us to promote the survival and reproduction of your genes during our evolution. Despite the inevitability of this fact, many react to it with denial, because it does not correspond to how we feel. A typical reaction is “but surely there must be more to life than this?” This is an example of the logical form “if I do not like it, then it cannot be true”. The answer, sadly, is no, there is not more to life than this – just get over it. If it is true that all our complexity has been generated by the force of natural selection (and I am as sure as I can be that it has) then there is no other possibility. Yet, in Darwin’s words, there is “grandeur in this view of life”; there is dignity in being the first form of life that really understands what if going on. The fact that we do not feel that this describes our purpose is not surprising.

Evolution has programmed us to have certain goals which serve the overall purpose of transmitting our genes to the future (or at least did so in the conditions of the past) but it was never necessary for us to be aware of that goal. It is no more astonishing than the case of the earthworm which is programmed to get away from light, without understanding anything about birds that might want to eat it. Even the best professional philosophers have trouble coming to terms with this. For example, John Harris, a very able professor of philosophy at Manchester University, has produced a useful book on medical ethics called The Value of Life (1985). He only mentions evolution once in the entire book, with the words “Indeed if one or other of the versions of natural selection describe a natural evolutionary progress for human beings, then again while it may be evolutionarily successful, one could hardly describe the survival of the fittest and its corollary, the destruction of the weakest, as a humane (albeit a human) arrangement.” This is a more sophisticated version of the bad logic which says, I don’t like it, therefore it is not true. After some other confused speculations, he continues to draw his moral premises from his gut feelings. He completely fails to spot the fact that natural selection and natural selection alone is the only information input that has formed our value sentiments.

Without understanding this, a complete philosophy is impossible, although much of the refinement of gut feelings that he makes is both competent and useful. And the fact that most of our actions today do not maximise the survival and reproduction of our genes is not counter evidence. Our brains evolved over millions of years of low population densities of tribes of hunter-gatherers. Our instincts and social habits are well attuned to that and they may have been modified by 10,000 years of agricultural settlement. But our modern world of large cities, where we are surrounded by all kinds of human inventions and are almost totally divorced from a natural habitat, has existed for too short a time for natural selection to have acted very much. Evolution of Morality If natural selection is the only non-random information source that has built our minds, then it must be true that our wishes not to die, to eat, to have sexual relationships or even to acquire wealth and status, could have evolved in this way. But when we consider our moral instincts, less than perfect as they are, many reach for Alfred Lord Tennyson's words, Nature, red in tooth and claw, to describe what they would expect to have come from a struggle for survival of the fittest.

But a moment’s reflection is enough to put aside that quick rejection. Probably for most of our ancestry, even before our divergence from the other great apes, we have lived in social groups. It is quite unrealistic to suppose that a totally selfish and violent individual would leave more offspring than a person who was good at cooperating with his or her neighbours. The famous “prisoner’s dilemma” of game theory can be formulated as a table of rewards between two players. Player 2 Cooperate Defect Cooperate 8 / 8 0 / 10 Player 1 Defect 10 / 0 2 / 2 In this formulation of the game, if both players cooperate, each will receive a reward of 8 points, but if both defect they will only receive 2 points. There is, however, a temptation to defect, because, if Player 1 defects when Player 2 cooperates, then Player 1 will get 10 points while Player 2 will be a sucker and get nothing. So what would you decide, if you are playing this game? If your opponent cooperates, your best move is to defect – you will get 10 instead of 8. If your opponent defects, your best move is also to defect – you will then get 2 instead of 0. So no matter what your opponent does, your best move is to defect. Both players are making the same calculation, so the inevitable outcome is that they will both end with 2 points when they each could each have had 8. In a single game, assuming the players cannot communicate, there is probably no way of achieving cooperation, but in repeated trials, there might be. This very simple piece of game theory seems to encapsulate the very essence of cooperation. It often happens in the real world that restraint and trust is necessary to achieve cooperation, yet cooperation brings reward in the longer term. There is a big payoff, in terms of natural selection for creatures that can find a way to cooperate, so, if the possibility exists, we should expect it to have been evolved. I have carried out computer simulations which show that a population of simple creatures, which have a tendency to cooperate or not and can remember encounters with identified individuals, can in the right circumstances evolve an instinct to cooperate.

I described this project at the BRLSI a few months ago (The Nature of Ethics 7th January 2014). I was by no means the first to attempt a simulation of this kind and many other computer projects using different structures and assumptions have reached a similar conclusion. Researchers have identified five mechanisms by which evolution would favour altruistic or cooperative behaviour: Kin selection – it is obvious that altruistic behaviour towards one’s children and other relatives could lead to a genetic trait being selected. Direct reciprocity, both partners can gain by finding a way to cooperate. This is the problem illustrated in the simple game theory example above. Indirect reciprocity , or “display altruism”, occurs where an individual is altruistic without hope of return, but thereby improves his reputation in the hope of receiving cooperation from others in future. Organisers of charity events understand very well the tendency of people to give more generously when they are being watched. Spatial Selection occurs when populations are relatively static, as human populations were until the invention of the steam locomotive. They will probably share some kinship with others in the area and groups of cooperators could arise. Group Selection could occur when there is a greater altruism within a group allowing whole groups to prosper more than others. This may, or may not, be coupled with less altruism shown to outsiders. Group selection has not been universally agreed among biologists, because its effect may not be enough on its own to prevent the growth of freeloaders. These mechanisms have produced a degree of morality in many species. For example, lions eat antelopes, but tend not to eat other lions. Of course, when a new alpha male lion takes over a pride of females, he tends to kill the cubs sired by his predecessor. This is less than moral in human terms, but it is perfectly moral in lion terms and is equally a behaviour formed by natural selection.

Humans, equipped with speech, have developed complex social mechanisms to a greater extent than other animals. We have many systems to reinforce cooperation and prevent defection. Fences, locks, laws, police forces, contracts, codes of conduct, are all ways of preventing cheating. We have an instinct to promote cooperation in the society in which we live. We also have an instinct to disingenuously preach a higher level of morality than we follow ourselves and, of course, we have an instinct to look after number one. It is sometimes supposed that morality is not a built-in set of values, but it is somehow a thing that is “emergent” from society. This is a fallacy and the reasons why it is so wrong must be clearly understood. Let us think of the information flows. It is clear how goals could be set by natural selection acting on a population of animals with brains, but how could this information originate out of the flows between one individual and another?

But we still have to explain why the ethical code of any society changes with time. If our purpose is built in by evolution, how could it change in a timescale of less than one lifetime? Meaningful information cannot “emerge” from a data set that is totally random. When it appears to do so, it may be reflecting non-random information that has been put into the system, or it may be specified by non-random properties of the system itself. Monkeys at typewriters would produce a great deal of random rubbish before one piece by chance generated a work by Shakespeare. What is actually happening in human society is that its members are in constant negotiation with each other. We have an instinct, which has also been created by natural selection, to propose moral principles to each other and to reach agreement of what is fair. We have an instinct to negotiate a social contract. By this means the ethical norms of a society are changed over time, but the process is one where individuals are trying to create circumstances in which their built-in, evolved values will be better served. An individual will, at least publicly, absorb the norms of society. The bad selective consequences of not doing so are obvious. This instinct to accept the society’s norms is yet another value that has been built in during our evolution as social animals. Different societies have developed different moralities and for the individuals within them, it may be impractical to depart too far from accepted practice. But it can still be worthwhile to try to “improve” the morality of one’s society.

This could be for the benefit of all, including the improver and his or her kin. Societies in different places have been more or less successful in building ethical systems which benefit its members. Within countries there are “nice” areas where a high standard of morality prevails and “sink estates” where a bicycle parked for a few moments will disappear. Some countries are stable, while others are “failed states” where groups expend much of their resources in fighting each other, yet within each warring militia, a moral code always exists. There is an instinct, also plausibly created by natural selection, to favour the group that one is in communication with above outsiders. Part of the change, or progress, that we have seen is a widening of the groups that are treated as insiders and this may be due to the changes in communication possibilities that we have, from the invention of printing to the internet. The relationship between the formation of codes of behaviour and communication technology is an interesting topic for research. It is a massively complex system that has grown up and it is not surprising that philosophers have completely failed to understand it.

Such suggestions as “the greatest happiness of the greatest number” or duty-based rules, or a belief that we should be following rules set by a big invisible magic bloke who lives up in the sky are so simplistic that they are only reflections of naivety. The many different approaches to ethics outlined in any introductory text, and the various luminaries who have expounded them, show the confusion. We can now understand that our sense of right and wrong has evolved as an adaptation to our living as groups of cooperative hominids. We are programmed to develop moral codes with our neighbours to mutual advantage. And our genes have created vastly complicated brains and nervous systems, all devoted to one simple objective: to transmit copies of themselves to future generations. The objective is simple, but the means that have evolved to achieve it have reached immense complexity. But there is no other information input that has contributed to our sense of purpose or morality. Before the evolution of life, there was no purpose, no morality. These have been created by natural selection. But our understanding of how our life purpose and our morality have come about is not the full answer to the problem. We may understand why we have an urge to act in a certain way, but it is not obvious that this means that we ought to do so. Indeed it has been a clear rule of philosophy that you cannot derive and “ought” from an “is” and this rule makes a great deal of sense. As Hume said, reason can only be the servant of the passions, but we know now where these passions have come from.

So we are left with what should be the central problem in moral philosophy. Either our complex, animal based, evolved system of morality is right, or else it is meaningless because there is no other source. Our cave-man instincts no longer achieve their original aims in the modern world. Should we adapt them to do so? Or, if we should serve some other purpose, then what should that be? It is no use saying, “Oh I just feel we should work to make a more just and caring society”. That is just an example of the evolution-installed instinct to negotiate a code of behaviour with others, to better serve our evolution–installed needs. The same will apply to anything else we drag out of our subconscious. This cannot be solved by interrogating our gut feelings. It needs more careful thought than that. We have to stand back and try to look at what is going on, as if from the outside. I can give no final answer, but I am sure of one thing: many of those who aspire to the title of philosopher nowadays simply don’t get it. For all their pretentiousness, they have failed to understand the basics. You can get nowhere in philosophy without taking into account paths taken by the information flows. And you certainly can get nowhere in philosophy without taking into account the origin in evolution of our reasoning powers and our moral sentiments.