Terms of Reference

WHEN a theme has become the subject of prolonged discussion certain arguments from both sides tend to become rather stereotyped and taken for granted. It then becomes necessary for such arguments to be re-examined, both for their relevance and for the conclusions to which they lead. Even the constructions that have come to be put on the expressions employed in the discussion need, from time to time, to be carefully reviewed. The present paper is an attempt at such a re-examination of the traditional approach to the question: What are the implications of any similarities and differences that there may be between living and lifeless machines?

The Traditional Arguments

The controversy about this question is philosophically important because of its bearing on the question: Whether living matter is or is not controlled by something that is (a) distinct from it, and (b) non-material. Those who say that it is so controlled give such names as Life, Mind, Soul, Entelechy, Elan Vital, Diathete to the influence claimed to exercise control. In doing so they attribute the structure and performance of living things to the aided action of matter on matter, for they claim that this non-material influence aids the course of material events. By contrast, those who deny that living things are controlled by any sort of non-material influence attribute their structure and performance to the unaided action of matter on matter.

Believers in aided action are sometimes called dualists, while believers in unaided action are called monists. But rather than become involved in a discussion about the justification for these names I shall avoid using them here.

Traditionally, believers in unaided action argue as follows: 'A machine does not contain any substance that one could describe by such a term as Life or Entelechy. The living organism is in all significant features a machine. Therefore there is no reason to postulate any such substance for the living organism.' Recently a number of mechanical devices have been specifically quoted in support of this school. Among them are digital and analogue computers, apparatus so designed that it maintains some quantity at a pre-determined value, battery operated trucks encased in a good imitation of the shell of a tortoise, and other ingenious devices. Each is said to support the theory of unaided action in respect of one of the performances of a living organism, the inference being that a sufficient number of such devices, sufficiently varied, would between them cover the whole of life's activities.

Traditionally, believers in aided action reject the view that living organisms are essentially machines. They enumerate sundry accomplishments of which living organisms are and machines are not capable. So it has come to be accepted without question by both sides that to say that a living organism is a machine amounts to saying that it results from the unaided action of matter on matter, and that the proper way of refuting this theory is to prove that it is not a machine.

What is a Machine?

This seems to me most odd and I hope that it will not be thought paradoxical but merely obvious when I say that protagonists on both sides ought to use exactly the opposite arguments. Believers in aided action ought to insist that living organisms are machines, and believers in unaided action ought to insist that they are not. The error is, I think, that no attempt is made on either side to think out what characterises a machine. For one cannot decide whether a living organism is a machine, nor what conclusions to draw if it is one, unless one has a criterion by which to distinguish between what is and what is not a machine. To say that a living organism is a machine means, if it means anything at all, that it differs in some significant way from a structure that is not a machine. What is this difference?

It would be correct enough, but quite inadequate to describe a machine merely as a structure in which energy is converted from one form to another. For that holds true of any arbitrary collection of bits and pieces, from a motor car down to a piece of iron that rusts, a rock face that crumbles, or the surface of the sea when the wind disturbs it. It is equally true of any part of a motor car, or of an assembly consisting of the wheels, together with a portion of the road, half the house that the car is passing and the boots of a pedestrian. I bring in this absurd illustration in order to make clear, firstly, that the conversion of energy must be in some specific form and, secondly, that there must be a criterion by which to define the boundaries of anything that one would call a machine.

When those who wish to prove the theory of unaided action use an illustration they never choose rusting iron, crumbling rock faces, or disturbed sheets of water; they never choose a random collection of bits and pieces. They always choose something that performs a wanted operation, something of which one can define the purpose, be it only to maintain a pointer in a predetermined position, something that an engineer too would call a machine, something teleologically determined. If they did not choose something like this the illustration would not be at all convincing.

The Input and Output Functions

A characteristic of all man-made machines is that they serve as instruments of control. They are manipulated at more or less frequent intervals by an operator, who uses handles, levers, pedals, push buttons, thumbscrews, and similar controlling devices; and their performance is not a random one but the one that the operator contemplates while he is exercising the control. This characteristic has caused engineers to speak of the input function and the output function of some of their machines, of servo-mechanisms for instance. They would be quite precise if they applied the same terms to every machine they worked with. Both an input and an output function can be distinguished for digital and analogue computers, servo-mechanisms, guided missiles set to home on a target, toy tortoises, in short, for every one of the devices quoted in support of the theory of unaided action. And the terms input and output function, be it well understood, do not refer to energy. The input energy and the output energy are distinct from. the input and output functions. They have different origins, different destinations, and follow different paths.¹

To say that the performance of a machine requires an input function is to say that specific things must be done to the machine for it to give its characteristic performance. And every man-made machine is equipped with special devices by means of which an operator can do those things to it. In a motor car they are called the controls. Corresponding devices are an essential constituent of analogue and digital computers, of any machine that an operator so adjusts that it will maintain a pointer in a predetermined position, of a 'robot' aeroplane, of the mechanical tortoises said to be analogous to living organisms. They are what I have just spoken of as the controlling devices. They serve as receptors for the input function.

Those who say that the structure and performance of a living organism, are controlled by a non-material operator distinct from its body say, in effect, that its characteristic output function occurs only when there is or has been a corresponding input function, just as it does for a man-made machine. As with this, they claim, something specific is only done by the organism when something specific has been done to it. And the living machine, one should expect them to insist, must be equipped with devices analogous to the controls of a motor car, with receptors for an input function, by means of which the assumed non-material operator provides the control or, as I prefer to put it, the diathesis. In other words, one should expect them to insist that a living organism is essentially a machine.

Yet it is their opponents who do this. These opponents, the believers in unaided action, deny that a living organism is controlled by any agent analogous to the operator of a man-made machine; they deny the need for any mechanisms in a living machine that would correspond to the controls of a motor car. And yet it is they, and not the believers in aided action, who quote in support of their theory the resemblance of living organisms to certain machines.

This is why, as I remarked earlier, the position of the contesting parties is inverted. Those who ought to argue that living organisms are machines do not and their opponents do.

Anyone who believes in aided action ought, if he thought the matter out, to argue thus: 'I support the hypothesis that a living organism does not exhibit its characteristic output function when subjected only to random forces. I claim that it must be controlled by some influence that is distinct from itself. I claim that, in technical terms, there is an input function whenever the organism exhibits its characteristic behaviour. The same is true of a machine. So the lifelike performance of this mechanical tortoise tends to support my hypothesis in so far as similar output functions justify the assumption of the reality of an input function for both.'

Anyone who believes in unaided action ought, on the other hand, to argue thus: 'I support the hypothesis that a living organism does exhibit its characteristic output function when left entirely to the random influences of its environment. I do not believe that it is controlled at all in the sense in which a man-made machine is controlled. I do not accept the view that there is anything that an engineer would recognise as an input function. In this respect every living organism differs basically from any device that human ingenuity can produce. This mechanical tortoise, for instance, does not work if left entirely to the random forces of its environment. It only exhibits its characteristic output function when I manipulate it in certain ways, when I operate certain controls and make certain adjustments. It would be quite wrong to jump to the conclusion that the lifelike performance of this ingenious device proves the need for an input function for a live tortoise. The distinction between man-made and living machines is basic; it is that there is an input function for the former and none for the latter.'

This, I repeat, is how both sides ought to argue. But instead believers in aided action deprecate the machine analogy. And believers in unaided action insist on it. They attempt to prove that the living organism does not need an operator distinct from itself by pointing to machines that do. From the resemblance of a living organism to a machine that has an input function they infer that the living organism has none. Surely it can only be in the Colleges of Unreason, described by Samuel Butler, that the traditional arguments can have originated. It is only there that one might expect so much attention to be given to the output function, which is barely relevant to the controversy, while the input function, which is very relevant, is being assiduously ignored.

The Hypothesis that Living Organisms are Automatic Machines

A bad advocate may yet have a good case. So let us consider next what facts and reasoning would support belief in unaided action.

To justify such belief one would clearly have to prove that there was no input function to a living organism. And to say that there is no input function is to say that the living organism is loo per cent automatic in the sense of being a self-constructing, self-operating, self-repairing, self-maintaining machine. This is what believers in unaided action must and do say even when they try to prove it by means of experiments and demonstrations with machines that are definitely not self-constructing, not self-operating, not self-repairing, not self-maintaining.

Some man-made machines are more automatic than others. And the degree to which a machine is automatic is a measure of the amount of control and adjustment that must be applied to it. In a very automatic machine the number of separate controlling devices is small and these only need to be operated at rare intervals. The driver of a very automatic engine can leave it alone for long intervals after he has started it up, and many parts of the engine may work satisfactorily for years before they require any attention whatever. The driver of a motor car equipped with automatic gear change need do less things with his hands and feet than if the gear is not automatic. A robot could be left without attention for long periods. The input functions might be reduced to the occasional adjustment of a few thumbscrews and some words spoken into a microphone. A self-winding watch keeps good time if adjusted only every few years.

A machine that was 100 per cent automatic would, by definition, not even require that much input function. Whether such a machine could be produced in a factory is not relevant. One relevant question is whether it is possible in theory, whether the concept of a machine that is ioo per cent automatic is consistent with physical science. This is an interesting question but I do not propose to discuss it now.

The question that I do propose to discuss is the more concrete one, whether observation of living organisms tends to support or to refute the hypothesis that they are ioo per cent automatic. For this is, be it remembered, a hypothesis and as such in need of support. Moreover, it is fundamental in the dispute and has not been discussed in the literature. The present position is that both disputants can at present have recourse only to faith: on the one side to faith in the existence of non-material influences and on the other side to faith in the powers and accomplishments of matter. If this is not always appreciated it is because of a human weakness for describing one's own faith as facts and one's opponent's as mere hypothesis.

On the question, automatic or not automatic, I find again that the traditional view, blindly accepted by both sides, is the very opposite of the view to which facts of observation point. It always seems to be taken for granted that only those performances of living organisms that are subject to conscious control can conceivably be non-automatic. All other, and particularly purely vegetative processes, are usually declared to be 100 per cent automatic both by believers in aided and in unaided action. And yet no facts of observation support the hypothesis that living organisms are 100 per cent automatic; on the contrary, evidence available to everyone, as I shall show in a moment, strongly supports the hypothesis that living organisms are nearly, though not quite, 100 per cent non-automatic.² Before we come to this evidence let us consider the very minimum physical constituents of any machine, be it living or lifeless.

The Essential Constituents of a Machine

There are, and always must be, three indispensable constituents, namely: moving parts, fixed parts, and a source of energy.

That some things move in any machine is obvious. In moving they transmit forces, they exert pushes and pulls. It is equally obvious that there must be something to push and pull against. This is the frame. So any machine, living or lifeless must contain some parts that are fixed relative to others that move.

Among the moving parts in a motor car engine are pistons, connecting rods, shafts, wheels, valves, levers. These are guided and restrained by portions of the frame. This latter includes the bedplate, holding-down bolts, bearings, cylinders, cylinder covers, casings, struts, tie-rods, containers for fuel and lubricants. The frame of a motor car engine must be adapted to its function, which means that it must be strong, rigid and durable. So must each individual moving part. It is designed so that it may not bend or buckle, break or tear under the imposed stresses. Surfaces subject to wear must be hard and smooth.

The properties required of the three main constituents are often mutually exclusive. Joints between moving parts must be loose; they are often hinged. Joints between parts of the frame must be rigid; they are bolted, riveted or welded. Moving parts must be light and most of them have a small inertia; a well-designed frame is comparatively heavy. So far as possible the moving parts and frame are made of substances that do not burn easily; but this is exactly what the fuel must do. It lacks the property of durability essential for the moving parts and the frame, and has the property of inflammability lacking in them.

The Basic Difference between Living and Lifeless Machines

A living engine must, as I have said already, also consist of moving parts, frame and fuel. And I want to point out how basically the constituent parts of a living machine differ from the corresponding constituents of any man-made one. One might think at this stage that such a comparison is trivial and of no great philosophical significance, and that a true philosopher would ignore it in favour of other and bigger distinctions, such as capacity for thought, capacity for reproduction, a sense of values, higher things. But perhaps the notion will obtrude, as I proceed, that the real reason for reluctance to give attention to these problems of mechanics may be that they introduce some awkward questions.

In a man-made machine it is easy enough to distinguish on a casual inspection between the three essential constituents. One would not mistake the petrol for the bedplate nor a rivet for a cam. One should expect, therefore, when viewing a living machine to be able to tell quickly which parts were moving ones, which the frame and which the fuel. The properties of the materials and the shape of the items ought to provide an easy clue. In a man-made machine the materials used and their shapes have to be distinctive enough if the engine is to have a good thermal efficiency, to be light, to work without backlash, to last for a sufficient time without repair.

Let us list then first the moving parts of the human body. They include, of course, the legs, the arms, the hands, the fingers. The head is another moving part; it can nod and turn. The upper and lower portions of the trunk can each move relative to the other. In fact one cannot name one major division of the human body that is not a moving part.

A glance at the skeleton shows a design well adapted to this. Joints are hinged and not riveted, bolted or welded, as they would be if mobility were not essential. Similarly soft tissues are designed to yield. When one examines them one finds that they are designed for mobility. And when one examines the constituent parts of any soft organ one finds that these, too, are moving parts. The tongue, the heart, the walls of blood vessels, individual muscle fibres, nearly all components, right down below microscopic dimensions, are well adapted to the function of mobility.

Where then is the frame? Is the living engine a curious kind of machine built of nothing but moving parts? Certainly not. If it were it could not work. It would collapse. By the principles of mechanics there must be a frame. Where is it?

There is, of course, no mystery. At any moment certain parts of the body cease to be moving parts and, for that moment, develop the function of the frame. They are altered to serve the function. They then take up the pushes and pulls exerted by the parts that are, at that moment, moving ones. Whenever this happens the shape of each component is temporarily changed. In other words, a significant difference between a living and a lifeless machine is that for the latter the frame is permanent, while for the former it is temporary. It is an ad hoc frame. While one takes one step the stresses within the body are constantly changing and those parts used at the moment to take up these stresses are changed in shape accordingly. Components that are tie-rods at one moment become struts at another and connecting rods at yet another. While one word is being pronounced the tongue and lips change their shape and function many times. It is the same with all internal parts down to each tiny cell. Every component of a living organism is a machine that works in a strange way indeed. There is continuous redesigning of all its minutest component parts.

The Principle of Continuous Redesigning

This redesigning explains the high weight efficiency of living substance. The theme is a big one and I regret that there is not time now to discuss it adequately. Let me be content merely to point out that a man-made walking machine with a permanent frame would have to be so designed that this frame would take up stresses when the heel was touching the ground as well as those others that occurred when the toe was doing so. But the living machine is such that while the heel is touching the ground there is nothing capable of taking up those stresses that will be imposed at the next moment as the toe comes down. The frame will be correctly redesigned at the right moment for this and will then no longer have anything that would resist the stresses imposed by the posture just abandoned.

Similarly a motor car must be permanently fitted with a brake as well as an engine. But an animal has no brake while it is running. It is redesigned so as to acquire one when it is coming to a stop; and at this moment it ceases to have an engine of propulsion, an engine designed for the function of running. Any boxing machine would have to be equipped both with thrusting and parrying devices. A human boxer has each when needed and not at other times. An arm is a device for delivering a blow at one moment and a shield at the next.

Thus the conversion of the living body from one kind of machine to another is as radical as the conversion of swords into ploughshares. And it goes on so long as the organism is alive. In engineering such redesigning would call for a new set of blueprints, a new set of calculations many times a second. Can anyone seriously think that digital and analogue computers, toy tortoises, guided missiles come anywhere near to being a true analogy? For a true analogy it would not suffice that the toy tortoises were made of soft materials that the operator could easily bend and twist into new shapes. Devices for doing the twisting and bending would have to be incorporated in those soft materials. In other words many subsidiary machines would have to be provided. Their function would be to control the shape of the tissues from moment to moment. And the subsidiary machines would, in turn, have to be made up of sub-subsidiary machines and so forth down to molecular dimensions. The operator of a toy tortoise that was even remotely like a live one would be kept busy to bring about all the changes in the shape of the component parts that occur in the live one.

I have noticed that a fact of common observation is often not properly appreciated until it is given a name. So let me find one for this weight-saving method by which living organisms do not embody any component parts that are not required for the function of the moment. I shall call it the Principle of Continuous Redesigning. It is one of the most basic of biological principles and deserves, indeed, much more discussion than there is time for on this occasion.

To the illustrations of this principle already given, further ones could be added without limit drawn both from large scale and micro-physiology. But even if time permitted, additional illustrations would do little to make the Principle of Continuous Redesigning more self-evident. After all, anyone can easily find an unlimited number of further illustrations for himself.

The Principle of Continuous Repair and Maintenance

Let us now turn our attention to the fuel. In a motor car this is petrol and it is stored in a tank. In the human body it is chiefly glycogen and is stored in the muscles, having been converted from glucose in the liver. The chemical processes during muscle activity include the combination of muscle protein with sodium. This protein is therefore another part of the fuel. So both the glycogen and the protein serve the double function of being fuel and being constituents of those muscle fibres that are at one moment moving parts and at another components of the frame. The living body is analogous to a motor car in which the chassis, brakes, cylinders, pistons, connecting rods, valves and bearings all contained combustible material, some of which was burnt whenever the driver placed his foot on the accelerator. By engineering principles such an engine could not work. Depleted of their substance supports would weaken and collapse, tie-rods would break, struts buckle; valves and pistons would cease to be a tight fit; every working part would develop backlash. A motor car built to the principles adopted by life would fall to pieces before it had covered a mile. Indeed an engineer could only say of the living body: 'This is not built to last. This is built to wear out.'

Any man-made machine in which surfaces of the frame and moving parts are subject to much wear requires correspondingly frequent attention. Wear is corrected by replacing what has been worn away. So it is with the living body. If matter is worn away at a rate vastly exceeding that in any man-made machine, it is replaced at a corresponding rate. To the Principle of Continuous Redesigning one must add the Principle of Continuous Repair and Maintenance.

A Permeating Input Function

What is true of control in time is equally true of control in space. Large portions of the volume of any man-made machine never need to have anything done to them. This applies, for instance, to the inside of a shaft or a bedplate. But for a living machine one cannot say that the inside of any morsel of tissue is not subjected to controlled change. Even bones are the sites of much and varied activity. The tiniest membrane in a cell must conform to this rule. There is, in my terminology, a micro-diathesis in all living substance. In more colloquial terms, the living machine is analogous to a motor car for which the controls are not limited to a few items, such as the steering wheel, pedals, clutch handle and so forth, but occur throughout the structure. If one could make a toy tortoise that truly resembled a live one, the input function would not only have to be continuous; the controlling devices that enabled the operator to supply the input function would have to be legion. And they would have to be distributed throughout the tissues. The things done to such a device would have to be permeating as well as unremitting. An operator who had to cope with such a machine would surely not say that it was loo per cent automatic ! One would have to go to the Colleges of Unreason to meet that interpretation.

Summary and Conclusion

To summarise. The available facts point to a more basic resemblance between living and lifeless machines than has generally been recognised on either side of the dispute. Both kinds of machine are distinguishable from any structure that is not a machine in two basic features. Firstly, as frequently stressed by others, both exhibit characteristic output functions. Secondly, though less often pointed out in the past, the facts suggest that input functions are necessary for both. The most fundamental difference between the two types of machine is in the nature of the input function. For a man-made machine it is applied only occasionally and at a few places. For a living machine one must assume, at least until opposing facts have been found, that the input function is (a) continuous (b) permeating, and (c) operating on a scale too small for any material manipulator, indeed on a scale at which a Clerk Maxwell demon would work.

Thus certain hard facts tell against the hypothesis that living organisms are 100 per cent automatic. And these facts suggest that the control to which they are subjected is not applied only to the organism's conscious activities, but to all of them including even vegetative ones. The activity of what I prefer to call a diathete is of a nature that one cannot grasp with one's imagination. Is there any need that one should?

It may be argued that these hard facts will one day be interpreted in a way consistent with the hypothesis of unaided action. Perhaps. But if it is to be so the hard facts must be faced and an interpretation that will save the hypothesis deliberately sought. Until it has been found the hypothesis of unaided action cannot be maintained with the support of, but in spite of, observable facts. Its basis must continue to be pure faith.

Whether this conclusion will appear attractive or unattractive I do not know. But I think one major obstacle to its ready acceptance may be difficulty in finding a satisfactory answer to the question: Where is the operator that is assumed to be in control of this continuous and permeating input function? It is therefore worthwhile to point out that this question cannot have a meaning. For if the facts oblige one to assume the activity of a non-material influence in control of any living substance, the facts also force one, for reasons that it would take too long to explain here, to assume that such an influence must be without location.

* Lecture given to Postgraduate Students at the London School of Economics and Political Science, 2nd October 1953

References
1   As a term that would cover both input function and output function, that would cover indeed every process calling for control, guidance, selection, discrimination, timing, I have suggested elsewhere the word 'diathesis' (see Science v Materialism, London, 1940, and Mind, Life and Body, London, 1951). This term makes it possible to define a machine as a device for converting diathesis from one form into another. But I do not need to use the term here. The more familiar terms 'input function' and 'output function' serve the present purpose equally well.
2   As the hypothesis has recently been supported by some distinguished scientists that properly designed machines could both think and reproduce their kind it may be advisable for me to make it clear that I propose, on this occasion, to discuss only observable performances. And I do not propose to discuss either capacity for thought or capacity for reproduction.

 

Acknowledgement
This paper and any subsequent discussions and rejoinders are reproduced from the British Journal for the Philosophy of Science with the kind permission of the Oxford University Press.
Website home page: http://bjps.oxfordjournals.org/

Top of Page