Roberts, F. Somerville. The Missing Factor in Thinking Super-Computers, JSPR, 56, 1990,  pp. 300-304
Reproduced with the kind permission of Peter Roberts.
List of papers by the author; archived at

The Missing Factor in Thinking Super-Computers

by F. Somerville Roberts

An assessment of anomalous behaviour in visual perception indicates that thinking super-computers operating on a purely physical basis can never exhibit all the phenomena of consciousness and that only some form of dualism can account for the fact that such anomalies can be observed in conscious activity.

 Much of the mental phenomena of parapsychology would be more acceptable to the mainstream of science if it could be shown that certain manifestations of consciousness could be readily explained on a dualist basis, whilst at the same time it could indisputably be shown they could not have arisen through cerebral activity alone. Previously it has been virtually impossible to establish this position because the second requirement could not be fulfilled, but the evidence discussed below could influence this situation.

The argument between materialism and dualism has remained largely academic because neither protagonist could produce sufficient evidence to establish its case. The materialist standpoint can be stated very simply. Consciousness is a purely physical activity and can be, or eventually will be, completely explained in terms of chemistry and physics. There is absolutely no need for any hypothesis claiming that the mind is something apart from the brain. A human being can therefore be regarded as nothing more than an exceedingly complex robot or computer endowed with the ultimate in technical perfection.

However, the dualist has sound arguments to refute this claim and recently Penrose {1} put forward a case based on the solid foundation of mathematical deduction, which has been summarised by Smythies {2} when he says: 'Penrose provides compelling arguments demonstrating that the theory that any computer of sufficient complexity must think and feel as we do, is wrong'.

Support for this approach is shown by an analysis of the limited capabilities of computer consciousness in the area of visual perception. This indicates that it is not possible for a robot-computer no matter how highly intelligent it may be, to exhibit all the characteristics of conscious activity. In the first place, every computer in existence has been designed and constructed, and above all programmed and operated, by an independent external intelligence in exactly the same way that dualists maintain that an independent consciousness works in harmony with the brain. For this reason the behaviour of computers as we know them, or of any future computers designed by human consciousness could never be regarded as simply the activity of a self-contained physical system, but must be considered from the dualist standpoint.

Nevertheless, just as it can be maintained that the brain reached its present standards by means of nothing more than the process of natural selection, so the materialist is entitled to claim that computers might, in some manner incomprehensible at present, evolve through generations of self-construction into thinking machines with the fullest capabilities and faculties of the human intellect, including the equivalent of the neural network necessary for learning.

The question then arises as to whether such an intelligent super-computer is capable of exhibiting the phenomenon of, say, the voluntary control of visual perception. As this activity forms the basis of the following argument, a few notes on its behaviour are being included. Essentially it is the ability to control perception, and more specifically visual perception, by volition, and this is a faculty which almost everybody possesses to a greater or lesser extent. It occurs in normal healthy people when the visual cortex is recording a certain image received from the retinal system but, as a result of the exertion of'will-power' the consciousness creates an entirely different image. This image, however, is the one we actually see. Examples of this phenomenon have already been given in the Journal, such as the way the rotating arms of a radar scanner or an anemometer can apparently be reversed at will and how the movement of a dim light in a dark room can be directed as desired, although, of course, it is stationary. {4}

The ability to control perceptual images depends largely on two factors, of which the first is the nature of the physical object being viewed. One of the easiest to control is a body rotating in the line of sight whilst the most difficult is probably a stationary object in an illuminated field of view. The second condition is the critical approach of the observer; the more uncritical he is, the greater his control of the image. Normally, the image of a car crossing the field of vision would be almost impossible to influence, but if the critical faculty of the observer is reduced, as by a suitable degree of hypnosis, then, on being informed the car is, for example, stationary, he will believe it is so and accordingly will see it there, obliterating the appropriate area of background, even though the car has actually moved on out of sight. All this suggests that we do not see what is there, although the visual cortex will, of course, be registering the physical field of vision, but instead we will see what we believe is there. It follows that the senses do not control our perceptions but only act as a guide which the consciousness can follow wholly or in part, or disregard entirely if it so wishes. This point was first put forward (as far as I can ascertain) by Tyrell, {J} and amplified by Roberts {4} who described the wide range of phenomena which can be explained adequately by this theory. This indicates that the belief, or 'idea', necessary for controlling visual perception can be engendered not only by the observer's own 'will-power', but by another person, and although he is able to control the observer's visual perception, this does not affect the basic principle that it is being controlled by conscious effort and not by the visual cortex.

An interesting example illustrating this situation (and somewhat similar to the group control of autokinesis referred to by Gregory) {6} occurred in Ireland some months ago. It was reported by several people that a statue of the Virgin Mary had actually been seen to move and within weeks reports of similar sightings were coming in from many parts of the country. The usual sceptical explanations were made that the observers were drunks, publicity seekers and so on, yet many normal people of the highest integrity claimed, although usually very reluctantly, to have seen the phenomenon. There is little doubt but that they actually 'saw' the statues moving, although, of course, they were rigid. It is likely that these observers were conditioned by various reports into believing the statues moved, and irrespective of what their retinal systems and visual areas were registering, their perceptive images were of moving statues. Claims were made that it was all a matter of 'suggestion' but this was no explanation as it is simply another way of describing external influence on the area of conscious belief which eventually creates perceptual images.

A phenomenon somewhat similar to voluntary perception appears to take place with alternating figures such as a Necker cube, when either voluntarily or involuntarily, the perspective dimensions of the figure will suddenly reverse and then change back again. This indicates that the two images could be two separate creations of consciousness based on a single pattern in the visual cortex but the mind brain identity theorist can always claim that the image is a fixed pattern in the cortex and that a further stage of brain interpretation is simply testing the alternatives available. This is an operation which can readily be carried out with existing computer equipment.

With the radar scanner, however, there can be little doubt that whilst the cerebral configuration is one 'image', the image we actually see is a creation of consciousness. On looking at the left arm of the rotating scanner, or of the anemometer, it will be seen at its maximum size when fully extended across the field of vision and will decrease in size as rotation carries it further away from the observer. However, that identical arm in the volition-created image, in reversed direction, is approaching the observer and will be seen to increase in size, showing that this image is only suggested by the visual cortex, but is actually created by the observer's consciousness and what it believes. It is very likely that the same is true of the Necker cube alternatives, but since in this case it cannot be proved, as with the radar scanner, it leaves the position open for the alternative interpretation.

When considering whether the voluntary control of visual perception could ever be a function of a super-computer, all controversy on the exact consitution of a visual image can be avoided by regarding it as simply an electronic pattern in that part of the computer's circuitry corresponding to the visual cortex. Unless it can be created by the electronic equivalent of'will-power', computer consciousness cannot be regarded as being on a par with human consciousness in this respect.

The voluntary control of perception necessitates the existence of an area of activity within the computer, separate from its 'visual' area, which initiates the 'willed' perceptual changes, unless it is accepted that the 'visual' area itself is able to create, control and modify the electronic patterns of which it is composed. However, irrespective of where it operates, the agency originating the changes in the 'visual' pattern is not a 'sensory' stimulus arising from an external source, and accordingly whatever its nature it must be endowed with some remarkable properties. In the first place, the circuitry must be such that it is aware that an act of volition will be required, and it must have the ability to create the circuit changes necessary to bring it about. The computer is a closed physical system and materialists claim that volition is a physical act. It is a law of physics that in a physical system any action must be the result of a prior action (except perhaps with radio-active emission or some quantum reactions). Such prior action must therefore be the result of a preceding action, and that of an earlier one still, and then another, and because no external agency is involved, the sequence continues into the absurdity of an infinite regress. It is thus clear that a computer could control perception only by breaking one of the basic laws of science. This situation does not arise with the dualist theory because an independent mind would be capable of originating any such activity, and it is obvious it would not arise in a computer if it were granted the co-operation of an independent consciousness, such as that of its designer.

Even assuming that the laws of physics do not necessarily apply to computers (and it is a rash assumption indeed!) and that by some extraordinary means the computer's intelligence was somehow able to develop the electronic pattern necessary for an act of volition, this pattern would have to ensure that its signals were able to traverse the computer circuits to reach the 'visual' area, and such signals must know exactly which parts of it they must affect on that specific occasion, and which they must not. They must be able to supersede the effects of any external 'sensory' stimuli already operating there, as they must be able to introduce new patterns of the agent's own construction, obliterate the existing 'sensory' configurations and impose their own designs. They must know the exact moment to relinquish their command of the situation and allow the 'sensory' activities to resume. The agent must be able to create exact duplicates of 'sensory' images although not provided with the information carried by 'sensory' stimuli for their construction. These are a few examples of the faculties a computer must possess if it is claimed it could emulate the various observable phenomena of voluntary perception, and it must operate them without the informatory 'sensory' signals which initiate the computer's normal perceptual activities. Some of these activities can, of course, be carried out by electronic equipment of present day standards, but this has all been designed and programmed by an independent intelligence and what is under consideration is a computer acting entirely on its own without dualistic connotations.

The essential requirement when considering whether a computer could control perceptual activity is a theory to explain how the computer acquired this faculty. It might be claimed that it could be programmed into the construction of a highly advanced computer but this is unacceptable, as programming by an external agent is a tacit admission of dualism, whilst self-programming implies that the computer would be able to devise a suitable program before it was aware of the requirements of such a program, which is a situation hardly acceptable to science.

It might also be claimed that the computer is able to learn the ability to control perception, as computers at present may learn more elementary functions, and so establish the faculty as one of its attributes. This is not so. Even the supercomputer has to be programmed to be able to learn. It cannot just 'decide' to acquire a faculty, the very existence of which it is completely ignorant. It would be just as likely to 'decide' spontaneously to learn Chinese without ever having heard of such a language. External design of the program is essential, which inevitably brings us back to dualism.

The only alternative to dualistic design and programming is that the faculty must have come into existence through the equivalent of natural selection, through endless generations of developing self-construction, in the same way as it is assumed its human counterpart must have done. However, such a situation defies the principal basis of evolution as at present accepted, i.e. that an evolved organ or faculty must be advantageous to the entity of which it is part, and must be completely reliable. Any distortion of the data which the computer receives from the outside world, such as that created by the anomalous operation of voluntary effort, is clearly a disadvantage to it, and, moreover, renders invalid any 'learning' acquired by the computer for its future development.

It follows that if the development of voluntary perception in computers by natural selection is virtually impossible, it is necessary to consider how and why evolution succeeded in creating it as a faculty of consciousness. The objections and difficulties which apply to computer evolution apply equally forcibly to organic evolution. All physiological faculties have originated through natural selection and they have developed because of the advantages they have bestowed on their parent body, but this very prerequisite invalidates the hypothesis that voluntary perception originated in this way. On the other hand, there is no denying the fact that it does exist. There may be other organic forms of self-development, quite unknown to science, but the paradox can be resolved by one simple and logical deduction, which also explains other anomalies, namely, that the voluntary control of visual perception is not a physiological function and therefore could have come into existence independently of natural selection. There is then no need to assume that it must be an acquired faculty but is simply an inherent characteristic of consciousness.

If this theory is correct then, in addition to the apparent impossibility of its evolutionary origin, the faculty of voluntary perception should exhibit characteristics clearly distinguishing it from all physiological functions, and this it does in several ways. For example, ther is no aspect of voluntary perception which can indisputably be described as a physiological charcteristic, and this appears to be a unique situation. This claim might also be made for those stages of sensory perception following brain activation, but even so the sensory process has an organic origin whilst controlled perception operates without a sensory stimulus. Again, voluntary perception can, in certain circumstances, exert control over the functioning of the visual cortex, and this in spite of the fact that the retinal system is the well-tried and thoroughly reliable end-product of a long period of evolutionary development, and this clearly indicates that the agency behind voluntary perception is more potent than physiological motivation.

Or, again, voluntary perception does not appear to deteriorate or change in any way with the passing of time, as do all physical processes. Any faculty arising from a mutation must remain in constant and vigourous usage, as otherwise it will atrophy and die away. Birds which do not exercise their ability to fly soon become a flightless species. There are few, if any, records in history of the use of controlled perception, and even today few people are aware of it and fewer still have ever exercised it. It has been lying dormant for countless generations, presumably without any deterioration, as practically everybody has the ability to use it. It is unlikely that any physiological function could be subjected to such drastic treatment and emerge unchanged, and this strongly suggests it is not subject to physical constraints.

There are thus many valid reasons for rejecting the idea that some kind of super-robot/computer could exhibit the perceptual anomalies of human vision, and this fact satisfies the second requirement necessary for the acceptance of dualism. It follows, also, that if human beings were nothing more than organic computers they, too, would be unable to control perception. However, they are demonstrably able to do so. Surely the explanation is that human perception involves some factor lacking in a computer, however technically advanced and perfect it may be, and that that factor is a conscious activity which in some respects is independent of the brain and which is generally accepted as being one of the main foundations of the mental phenomena of parapsychology.

Ashford, Kindlestown Hill
Delgany, Co. Wicklow, IRELAND


{1} Penrose, R. The Emperor's New Mind, Oxford University Press 1989.

{2} Smythies, J. R. JSPR, 1990, 56, 229-234.

{3} Roberts, F.S. JSPR, 1989, 55, 309.

{4} Roberts, F.S. JSPR, 1989, 56, 216-221.

{5} Tyrrell, G. N. M. Apparitions, London, SPR/Duckworth 1953.

{6} Gregory, R. L. Eye and Brain, Weidenfeld & Nicolson, 1966.