Robert Thouless

Dr. Robert H. Thouless

Educated at Corpus Christi College, Cambridge, where he was awarded his PhD in 1922. He then went on to become a Lecturer in Psychology at Manchester, in Glasgow, and again in Cambridge. Here he became the Reader in Educational Psychology. He was President of the British Association's Psychology Section in 1937, and published a number of books connected with this subject. President of the SPR from 1942 till 1944.

The Pattern of ESP

 - Robert H. Thouless -

          IT HAS already been suggested that the main purpose of experimental research into ESP is not to provide more convincing evidence that ESP exists but rather to discover what can be found out about the nature of ESP. If ESP is a reality, the study of its events will be found to fall into a pattern. When we know enough of that pattern, we shall be able to make a guess at the rest of it, to construct a paradigm which is the conceptual system and set of rules within which ESP will fit so that it will seem no longer to be exceptional or surprising but to belong to the familiar order of things which we call 'Nature'.

We are far from that position yet. Bits of the pattern have emerged but they remain isolated bits of a pattern of which the whole is not yet perceptible. It is as if we had a large jigsaw puzzle of which only a few parts fit together but these parts remain isolated; we cannot see the larger pattern into which they all fit. Our task then is to try to fit together new bits of the pattern, in the faith that when enough little bits have been fitted together, the pattern of the whole will emerge.

Bits of the pattern are being all the time suggested by various researches which are published in the technical journals. A sample of these researches will be discussed in the present and following chapters. I shall not be attempting to cover the whole field but rather to give a representative sample of the experimental work that has made a soundly based contribution to still very incomplete knowledge of the total pattern of ESP. A conclusion can only be regarded as soundly based, either if it rests on data of high statistical significance which have been obtained under rigid experimental conditions, or if it has been amply confirmed by independent researches. We can, of course, feel more confidence in conclusions that satisfy both of these criteria.

There is no sharp dividing line between researches aimed at proving the reality of ESP and those directed towards finding out about its nature. This is especially true of the earlier experimental work, and it would be difficult to decide into which of these categories one should place the work of the Duke Parapsychology Laboratory reported in J. B. Rhine's first book (1934). In some respects the experiments took the typical form of experiments intended to prove the reality of ESP; there were numerous repetitions of observations obtained under identical conditions until values of P had become extremely small. At the same time, there were also variations of conditions, and the experimenters were alert to observe changes of response which occurred either spontaneously or as a result of these changes of conditions. As a result, there were a number of observations reported at this time which have been borne out by later work and which make a significant contribution to what we know of the general pattern of ESP responses.

It seems to be the case that some (but not all) of these early experiments were carried out under experimental conditions that were imperfect by modern standards. Certain defects in method were pointed out by critics both within and outside the laboratory, and where imperfections in method were seen to exist these were rectified. There is no good ground for supposing that any imperfections of method in early experiments did lead to spurious indications of success, but there is an element of doubt which makes the results of these experiments of uncertain value as evidence for the characteristics of ESP if they stood alone. They remain of great value as indicators of possible characteristics of ESP so far as these have been afterwards verified by later experiments carried out under more rigid conditions. There will be mentioned here a number of characteristics of ESP first suggested during the early Duke experiments which have received ample confirmation from the common experience of later experiments and researches carried out by the more rigid methods which became standard afterwards.

The first and most important of the early discoveries about the nature of ESP was that of the fact that looking at the target card by the experimenter (or agent) was not, as had been generally previously supposed, an essential condition for successful guessing by the percipient. One could, therefore, no longer refer to successful guessing by the percipient as due to 'thought-transference' or 'telepathy', and the more inclusive term 'extrasensory cognition' was substituted; later the more inclusive and less definite term 'psi' was suggested and largely adopted.

This was not, it is true, the first time that experiments had been done under the condition of the experimenter not knowing the right response, or the first time that success had been reported under such conditions. Success under clairvoyance conditions had, however, received little attention and the standard form of paranormal cognition had been regarded as that of one mind communicating with another ('telepathy'), although it was generally admitted that there might be an additional mode of paranormal cognition by 'clairvoyance'. Very often those who accepted mind-to-mind communication, rejected the possibility of clairvoyance. Experiments were generally done under conditions which could be explained by telepathy since the experimenter knew always what was the target object. The experiments at Duke made a revolutionary change in this situation. Since subjects were found to succeed equally well in ESP tasks in which the experimenter did not know the target as with those in which he did, the way was obviously open to consider the possibility that such experiments did not involve two paranormal cognitive capacities 'telepathy' and 'clairvoyance', but only one which might be revealed under different experimental conditions.

If this surmise were correct, it would not follow that all subjects would be equally good under both conditions; even trivial changes of conditions of experimenting may produce differences in performance. It cannot, therefore, be regarded as evidence against the possibility of the equivalence of telepathy and clairvoyance that Soal's subject Shackleton was reported to be able to succeed when the agent looked at the target card and not when he did not. If this were a real inability to succeed under clairvoyance conditions, it might be an individual preference of the subject (or of the experimenter) for the telepathy condition of experimenting. On the other hand, it may have been merely an unfavourable reaction to the introduction of a new condition.

It cannot, of course, be properly said that the early Duke experiments proved the equivalence of what had been distinguished as telepathy and clairvoyance; they showed that the distinction was less firmly grounded than had been supposed, and they opened the possibility that both were the same paranormal capacity operating under different conditions.

Another ingenious modification of experimental conditions was made about the same time when some percipients tried to give wrong responses instead of right ones. The percipient Pearce, for example, succeeded in getting a score of 17 in 225 guesses when he was trying to score low (28 below mean chance expectation), whereas he was reported to be averaging about twice mean chance expectation when he was trying, in the usual way, to score high (p. 40).

The same variation in method of responding was tried with another percipient Stuart (p. 70). This subject scored 81 above m.c.e.* (260) in 1,300 guesses with the usual aim of getting right, but only 182 hits (78 below m.c.e.) when he was trying to guess wrong.

* m.c.e = mean chance expectation.

There is nothing surprising about the fact that, if a subject can guess cards right, he can also, if he wants to do so, guess them wrong. What is interesting about these results is a fact pointed out later, that they show the subject having a much greater capacity to name the cards wrongly than to name them rightly (Thouless, 1935). It is true that both percipients showed about the same deviation below expectation when they were trying to guess wrong as they showed above expectation when they tried to guess right, but this is not the result that would follow if both kinds of guesses were being made with equal efficiency.

Stuart's 341 hits when he was trying to guess right can be accounted for by supposing that he knew the target card by ESP in 101 of his 1,300 guesses and got one-fifth of the remainder (1,119) right by chance. That would give him the observed number of hits (101 + 240 = 341). But his deficiency of 78 below mean chance expectation cannot be accounted for on the assumption that he knew 101 of the cards and could, therefore, deliberately guess them wrong, while for the remainder he guessed by chance one-fifth right and four-fifths wrong. A simple calculation will show that that would only give him a deficiency of 20 hits below mean chance expectation, about a quarter of the observed number. Further calculation shows that a deficiency of 78 would require that the percipient knew by ESP the nature of the target well enough to give it wrong about 390 times in his 1,300 guesses. Then if he gets right by chance in one-fifth of the remaining 910 guesses, this will give the observed number of 182 hits when he is trying to guess wrong.

What seems to be indicated then is that the successful subject knows by ESP a certain number of the target cards well enough to name them correctly, but a much larger number of them are less accurately known and the subject can perform the easier task of naming one of the four symbols that the target card is not. In Stuart's case, it seemed that the latter task could be performed nearly four times as often as the other. One cannot feel very strong conviction that this is a genuine ESP characteristic; it does not seem to have been confirmed by later work but this may simply be because no later experimenter has thought it worth while to carry out card-guessing experiments in which the percipient is trying to guess wrongly. It is not intrinsically a very unlikely result and it seems unlikely to have been produced by any deficiency in the experimental conditions of the Stuart experiments. It must now be noted as a possible characteristic of ESP; whether it is important or not can perhaps only be judged when the whole pattern has emerged.

The next observation made in these early experiments is of undoubted importance, both in practice and in theory. This is the tendency of positive ESP scores to decrease with time. This has been called the 'decline' effect, or 'chronological decline'. This term does, in fact, cover more than one characteristic of the psi-response. There is first the type of decline within any single occasion during which a subject shows a lowering of score as the occasion proceeds, or during some division of the occasion, as, for example, between the first and the last half of each run through the pack of twenty-five cards. Let us call all such declines within a single occasion 'episodic declines'. It is characteristic of episodic decline that it is temporary; there is almost complete recovery of performance at the beginning of the next occasion or of the next run through the pack.

There is also, however, a more permanent decline which is found over a long period of experimenting. An experimental subject who showed promising positive ESP results in his early experiments very commonly shows declining scores which may reach chance level within a period of months. He may not recover his former ability to score positively even after a long period without experiments. This characteristic of ESP may be called 'long-period decline'.

It is not easy to give a date for the first discovery of the decline effects although they were first singled out as a significant feature of the ESP response by Rhine in his 1934 book Extra-Sensory Perception. They had, however, been noticed earlier. Of the Creery sisters, for example, it was reported that 'the average of successes gradually declined' (Gurney et al., 1886). A similar decline was also pointed out by Estabrooks in an early study of ESP (Estabrooks, 1927). Since then, decline effects (both episodic and long-period) have been found by so many workers that one must regard decline as one of the best attested and most often repeated observations in ESP research. It is true that a certain number of subjects (as, for example, Soal's subject Basil Shackleton) have gone on scoring over very long periods without measurable fall-off of score, but such cases are exceptional. The more general rule is that initial ESP success is followed by decline.

We commonly use the one word 'decline' both for the decreases in score that may take place within an experimental occasion and for the more permanent decreases that take place over a long period of experimenting. This must not be allowed to hide the fact that there are important differences between these two effects. While long-period decline is most inclined to attract attention because of the experimental inconvenience that is caused by a successful subject ceasing to be successful, episodic decline is perhaps of more theoretical interest because it is related to other discernible bits of the pattern of psi. It may, in fact, be regarded as a special case of the more general phenomenon of 'position effects' (Cadoret and Pratt, 1950).

The general principle of position effects is that a successful ESP subject may show tendencies to score in some parts of every experimental occasion while he may show chance results or even negative scoring in other parts of his session. Episodic decline may not only take place over the whole session, but also be a characteristic of the results of each run of twenty-five guesses; the first half of each run may be consistently better than the second half.

More detailed examination of run scores shows in many cases that decline is not the only relation of scoring to position within the run. There is often found also a tendency to an increase in scoring during the last five calls of the run so that, if the average score per segment of five calls were plotted on graph paper, it would show a u-shaped curve with the second arm of the u generally going less high than the first arm. This effect was called terminal salience, and a formula may be used to evaluate its amount. The index so given is called the salience ratio.

The run is an arbitrary division made within the experiment by a discontinuity produced at the end of the calling of a complete pack of cards. Both the decline within the run and the terminal salience of the run must therefore be attributed to an effect on the subject's psi-response of a structuring which has been made within the experiment by the experimenter's way of carrying it out. One naturally asks, therefore, whether any other arbitrary division within the experiment would produce its position effects. This is found to be the case. When the percipient writes down his own guesses on a form in which each segment of five guesses is clearly marked off by a line from the adjacent segments, then (with adults, but curiously not with child percipients) the mean results of the segments showed significant terminal salience (Rhine, 1941).

Position effects were also first noticed early in parapsychological experimentation. In Rhine's first book Extra-Sensory Perception (p. 137 f.), graphs are shown demonstrating position effects within segments in Pearce's calls. The final mean segment curve shows a marked drop between the results for the first two calls of the segment and the last two. The curve is not u-shaped since the average scores for the fourth and the last call are the same. It looks as if both segmental decline and terminal salience may have been present, with the former predominating.

That ESP results may depend in regular ways on the structuring of the experimental task should not seem very surprising if we remember a very similar observation in memory experiments. If a subject learns a passage by heart, by the 'whole' method in which he repeats the whole passage many times in succession, it is found that, before learning is complete, the part best remembered will be at the beginning of the passage, the part next best remembered will be at the end, while the part least effectively learned will be in the middle. Thus the reproduction, when learning is incomplete, will be a u-curve of the same shape as that for ESP. I have little doubt that, if the passage were split up into verses, one would find optimal learning also at the beginning and end of each verse, corresponding to segmental terminal salience in ESP experiments. I do not think, however, that this experiment has been carried out for memorising tasks.

When position effects were first discovered, the interest felt in them was less in the fact that they revealed something about the pattern of psi than in their possible usefulness as an indication of the presence of ESP or other psi-process alternative to the direct evidence which would come from a significant positive deviation. It is obvious that position effects are evidence for some cause at work making the subject's responses correspond to some extent with the targets, since a merely random set of right answers would show nothing but accidental and non-significant internal consistencies of this kind. All position effects can be used as evidence for the presence of psi provided they were predicted before the particular set of results in question was examined for position effects. This prediction could be made as the result of a pilot experiment or from expectations aroused by some other set of experiments. It would be quite erroneous to notice a position effect in an experiment and then to work out its significance, and to regard that position effect, if significant, as evidence of ESP; that would be an example of the 'crumpled paper fallacy'. No competent modern experimenter would be guilty of such an error.

Episodic decline has a certain usefulness to the experimenter since it may provide him with indirect evidence of the presence of ESP; he is more inclined to regard long-period decline as just a nuisance since it prevents him from carrying out as many and as varied experiments as he would like with a gifted subject he is fortunate enough to have found. It is certainly a severe limitation to experimental work, and it would be easier to obtain answers to the unresolved problems of ESP if we had a reliable way of overcoming the tendency to long-period decline. This tendency is, however, also in itself of considerable theoretical interest. It is not what we should expect, because the more general rule in psychological experimentation is to find that the repeated performance of an activity leads to improved performance. This is what we call 'learning'. It is true that, within a single experimental occasion, we may find a falling off in performance which is attributed to 'fatigue'. This is parallel to the episodic decline in ESP experiments, but the long-period decline of ESP is obviously something different since it does not show the feature characteristic of fatigue that a period of rest leads to recovery of the lost ability.

The normal psychological activity to which long-period decline seems to be analogous is that of 'inhibition' in which there seems to be an active process within the organism preventing the repetition of a response, as, for example, when the response is coupled with some stimulus disagreeable to the responding organism. There does not, however, seem to be any obvious reason why a successful ESP response should be inhibited. Its results are not disagreeable to the person producing them. On the contrary, the percipient wants to be successful and is pleased when he is successful. At the conscious level, the situation seems to be typically one in which there should be reinforcement of the successful response which should lead to its becoming better over a period of time. If there is an inhibiting mechanism it must be an unconscious one; we must hope that future research in ESP will enable us to say more than this about it. Understanding of long-period decline will have both the practical advantage that it may lead to increased control and also the theoretical advantage of giving a new insight into the nature of ESP.

One possible theoretical implication of the tendency of ESP to become inhibited in the course of an initially successful series of experiments is that it seems to give some support to the speculation that ESP is a primitive form of cognition normally suppressed in favour of the more recently developed and more efficient perceptual system provided by the sense organs and the central nervous system. A successful ESP experiment would then be a situation in which this normal suppression has, in some way, been short-circuited; decline may be the automatic reinstatement of the normal suppression of the primitive psi-function. This speculation may give some hint as to the relation of the decline effects to the total picture of psi-functioning.

There are two further points about long-period decline on which the experimental records provide strong indications although neither has yet been made the subject of full-scale research. These are: first, the possibility that such decline may not be spread over all ESP performances but may be to some extent specific to the particular task which the percipient has already been carrying out; secondly, it seems very likely that long-period decline may not be merely something that happens to the percipient; it may also be that the experimenter is declining in his capacity to get positive results.

On the first of these points, a number of experimenters have had the impression that, if a subject has been successful in an ESP task but has dropped to chance results, he may start being again successful if he is given a new task. For example, in an experimental series with myself as subject, I carried out three tasks in card-guessing which differed in the degree to which I was accustomed to them (Thouless, 1949). On each of 144 occasions, one pack was guessed under 'down-through' conditions, one pack was guessed with the intention of the guesses corresponding with the order of the pack after a cut at a randomly determined point, while another pack was also guessed precognitively but in a more complex manner since the pack was thoroughly shuffled as well as randomly cut between the guessing and the check.

The first of these tasks was one that I had practised for a good many years with some degree of success, the second task was one that I had met at Rhine's laboratory about a year earlier and had used during that year again with some success, the third was a form of precognition task that I had never used before and which I thought would prove impossibly difficult. My expectation was that I would succeed in the familiar first task, probably succeed in the second more difficult task, but I felt sure that I could not succeed in the third task since this would require that I should foresee the result of the double process of shuffling and randomly cutting.

I made 25 guesses under each of these three conditions on each of 144 occasions. The results were at variance with all of these expectations. Under the first condition, I scored 62 less than the mean chance expectation of 720 hits, a significant negative deviation (P = 0.01). Under the second condition, the number of hits exceeded mean chance expectation to only an insignificant extent. In the new, and supposed impossible, third task, I obtained only chance results on the first 36 occasions but after that I began to score positively and ended with a significant positive deviation of 57 above mean chance expectation (P = 0.02).

A likely explanation of these results would seem to be that the first task, being familiar, was thoroughly inhibited to the point of negative scoring. The second, less familiar, task was also inhibited but to a lesser extent, while the third task, being novel, was not inhibited. The situation has changed now from that of twenty years ago; I am also now, I think, completely inhibited with respect to this third task.

When an experimenter has carried out a long series of successful experiments with a particular experimental subject and begins to find that he is scoring at chance level, he is inclined to say that the subject is suffering from long-period decline, and he begins to look for another percipient. This interpretation of the decline may, however, not be the true one, or it may not be the whole truth. It may be the experimenter who is showing long-period decline. This would not be easily apparent in the above situation; it might only show itself in the fact that the experimenter did not find it easy to get another subject to score successfully with him, and he might well attribute this difficulty to a shortage of gifted percipients, not to a change in himself.

Although I know of no systematic investigation of the relative parts played by the experimenter and the percipient in long-period decline, there are very numerous indications in the experimental literature which point to the experimenter as one factor in decline. Many experimenters who reported positive results in the early days of their researches have found that, after a few years of experimental work in this subject, they no longer have highly successful results, even perhaps no successful results at all. The most dramatic example of this was in Rhine's Parapsychology Laboratory at Durham, North Carolina.

In Extra-Sensory Perception, Rhine reported that the frequency of successful subjects and their rates of success were much higher than had been found anywhere else. This astonishing success rate seems to have continued during the first decade of the laboratory's work, but their later record is of success rates as disappointingly low as are found elsewhere. Critics of ESP sometimes suggest that the experimental conditions were lax in the early days of experimenting at the Duke Laboratory and that when these conditions were tightened up the early successes disappeared. If this were the case, it would be supposed that the early successes were due solely to the lax conditions, but the actual facts do not fit this simple explanation. It is true that, as experimenting went on, more rigorous conditions were introduced but the falling off of scoring did not coincide in time with the imposition of these stricter conditions. The Pratt-Pearce experiment, for example, took place in 1933-4 under the stricter conditions but it showed a high rate of success (over 12%). The high rate of success seems to have gone on for some years after this. The later fall to levels of success only slightly above mean chance expectation appears to be, not a result of any change in experimental conditions, but an example of the setting in of the general inhibition of ESP scoring which we have called 'long-period decline'. If this is its explanation, it must have been the experimenters who were becoming inhibited since the experimental subjects were not the same individuals during the period.

No doubt the development of the inhibitory process was accompanied by, perhaps caused by, changes in the motivation of the experimenting team. Possible changes affecting motivation would be such factors as the falling off of the first enthusiasm which accompanied the conviction that they were breaking into a new area. Side by side with this falling off of the early enthusiasm, there may have been increasing boredom with the monotonous and repetitive task of routine experimenting, increasing ego-involvement with results, and so on. These are merely speculations as to possibilities; no systematic research has yet been done into the emotional factors determining experimenters' long-period decline.

Although episodic and long-period decline could not have been predicted as properties of ESP if they had not been experimentally demonstrated, there is nothing particularly surprising about them; we have other examples in experimental psychology of performances which fall off through fatigue or inhibition. The next discovery reported in the early Duke experiments was a much more startling one, that percipients pressed to continue in card-guessing after their scores had dropped to chance level, might begin to get fewer hits than the number one would expect by chance.

The first example reported of this was with the experimental subject A. J. Linzmayer (p. 62). After a session in which he was reported to have scored about 40% hits, he had dropped to about chance level but Rhine urged him to go on with the experiment which he reluctantly agreed to do. In successive series of 100 guesses, he scored 14, 16, 20, 17, 14; that is, 6, 4, 0, 3, 6 below mean chance expectation. No conclusion could, of course, be drawn from this single observation; the result might be accidental and not due to a real tendency to score below chance. Obviously one may, by chance alone, get groups of results below mean chance expectation as one can, by chance alone, get groups of results above mean chance expectation. The observation was, however, noted as an indication of a problem for future research. Later research has clearly shown that below-chance scoring is a genuine phenomenon and not a mere statistical accident; that such negative scoring effects may be real characteristics of ESP performance is shown both by statistical tests and also by their repeatability. It has also been found that negative scoring is not restricted to situations in which the percipient is urged to go on experimenting when he wants to stop, although some degree of negative attitude towards ESP seems to be a factor in some cases. Negative scoring has also been noted in some cases in which sets of scores have been compared in group experiments or between different conditions. It also seems as if some experimenters under some conditions habitually obtain significant negative scores. The matter of negative scoring is by no means fully understood although it seems clear that it is a genuine effect. The general term 'psi-missing' was used by J. B. Rhine for this and closely related effects (Rhine, 1952).

What makes psi-missing a somewhat surprising effect is the consideration that a percipient can only consistently make wrong guesses if, in some sense, he knows what the right guess ought to be. This strengthens the suggestion that an inhibitory process may be at work in psi-guessing, in this case not merely inhibiting the ESP response but converting it into a direction opposite to that consciously intended by the percipient himself. This would seem to imply that, in a group of experimental subjects of whom some are scoring at about chance level while others are scoring below the level expected by chance, it will be those who have the lowest scores who are showing most ESP activity. It is to be expected, therefore, that these low scorers will prove to have more in common with high scorers in ESP tests than with those who merely score about chance level. This is a problem which has not been much explored, but there are some experimental indications that this expectation is fulfilled.

There is another reported characteristic of ESP which, if it proves to be a genuine one, would seem to be related to psi-missing. This is the 'consistent-missing effect' reported by Cadoret and Pratt (1950). This is the apparent tendency in some subjects consistently to guess a wrong target, e.g. to guess 'cross' when 'circle' is target. This is not an unlikely effect and might indeed be the mechanism of psi-missing. The evidence for its reality is not, however, very strong. The only one of the experimental series they examined which showed the effect at all strongly was one carried out under defective conditions in which the percipient saw the backs of the cards at the time of guessing. Such indications need confirmation by experiments carried out under satisfactory conditions. Some of the experimental series found by Pratt and Cadoret to show the effect were carried out under perfectly satisfactory conditions, but none of these was at a sufficient level of significance to inspire much confidence. The consistent-missing effect may be, and most likely is, part of the pattern of ESP responses, but this needs confirmation. If the habit of showing the results of card-guessing experiments in a (5 x 5) matrix (showing all calls on all targets) became usual amongst parapsychologists, any published series could be examined for consistent missing; then we could quickly find out whether the effect exists and how common it is.

There were two other inhibitory effects noted in the early experiments at the Duke Laboratory which may be called the 'witness effect' and the 'change effect'. Both have been so fully confirmed by the common experience of later experimenters that it is not necessary to enquire whether the original experiments by means of which they were first detected were sufficiently safeguarded.

The witness effect was first reported in the case of the percipient Pearce (p. 76). The results of the introduction of visitors on 9 occasions in 1932 and 1933 are shown. These ranged from William McDougall, Professor of Psychology at Duke University, to Wallace Lee, a magician. In each case, Pearce showed positive scoring before the visitor was introduced, a drop of score, sometimes to chance level, after the visitor came in, with a later recovery to a satisfactory level of scoring. A typical example is his reaction to a visit by McDougall on 2 February 1933. In 350 guesses before the visit he had scored 132 hits (38%, i.e. 18% over mean chance expectation of 20%). During the first 125 guesses in the presence of McDougall, the scoring dropped to 33 (6% over m.c.e.). The next 250 guesses, however, still with McDougall present, showed 105 hits (a rise to 22% over m.c.e.).

What this observation suggests is that the introduction of a new person to the experimental session is likely to lead to a reduction or total disappearance of positive scoring, but that this is not a permanent effect since the percipient may become adapted to the presence of a new witness after a period of time. I know of no later systematic research on this effect, but it is constantly the experience of parapsychological experimenters that an observer who comes to look on has the disappointing experience of seeing no success. It may, therefore, be regarded as a finding that has had sufficient confirmation by common experience if not by systematic research. These early experiments also suggest that if the disappointed observer stays quietly looking on until the percipient has become adapted to his presence, the inhibition set up by his coming is likely to disappear and any former level of success to reappear in his presence. It is a common opinion amongst experimenters that such inhibition is particularly likely to occur if the observer is hostile to the experiment or to the percipient. There seems to be no certainty as to whether this hostility must be overtly expressed to be effective, or whether a concealed hostility also inhibits the phenomena. There seems to be no systematic research on these problems, but they are of both theoretical and practical interest.

Another inhibitory effect first reported by Rhine (1934) was that resulting from a change of experimental procedure. This may be called the 'change' effect. It was reported (p. 77) that any change in the experimental methods adopted with the percipient Pearce were likely to cause a drop in scoring rate unless the changes had been suggested by himself. Again the drop was temporary; after a period of low scoring, the subject recovered the ability to score under the new conditions. For example, the agent started looking at the target card after experiments carried out entirely under 'clairvoyance' conditions. In the first 175 guesses under this changed condition, Pearce scored 42 hits (a success rate of only 4% over m.c.e.). In the next 175 guesses when he had got used to the new condition, the success rate went up to 34% over m.c.e. (95 hits). In six other changes of task, a similar drop in scoring rate followed by recovery was recorded.

This reaction to experimental change has also been confirmed by later experimenters with other changes of task. Tyrrell, for example, in experiments in which the percipient (Miss Johnson) indicated which of five boxes would be lit up on opening, introduced a new mechanism (Tyrrell, 1935). Miss Johnson had been succeeding with an earlier and simpler type of machine, but her scores dropped to chance level when she started with the new one. But after a period during which she had only chance results with this new machine, she seemed to become adapted to it and scored 373 hits in 1,271 trials, an excess of 118.8 (or 9%) over mean chance expectation. This excess is highly significant.

She was, however, still unable to score above chance level when a commutator was introduced into the circuit to make the choice of targets independent of the experimenter. In 500 trials under this condition, she scored 101 hits which is almost exactly mean chance expectation. Even when Tyrrell mixed up batches of experiments with the commutator in action with others in which it was not, Miss Johnson scored in those experiments in which the commutator was not in circuit but failed in those in which it was. A less persevering experimenter than Mr Tyrrell might well have concluded that the random choice of targets by the commutator eliminated the possibility of successful scoring, and might have inferred that the experimenter's choice of the target was essential to the percipient's right guessing.

This inference would, however, have been wrong since, after a holiday in September 1935, Miss Johnson became adapted to the experiment with commutator, and in 4,200 trials under this condition, she scored 178 hits over mean chance expectation (Tyrrell, 1936). This is highly significant (P < 10-10) and corresponds to a success rate of 4.2 per cent, which is somewhat less than that recorded in the simpler conditions of experimenting but is sufficiently striking.

That change of task may produce a temporary inhibition of ESP response is only in apparent conflict with the finding already reported that, when a percipient has become inhibited with respect to an often repeated task, a change of task may result in a resumption of successful scoring. Both sets of observations may be included in the statement that either successful adaptation to a task or inhibition of ESP success in a frequently repeated task may be disturbed by the introduction of a new task. Both of these effects may be observed in succession in a single experimental series. In my own comparison of three psi-tasks already described, the new (complex precognition) task showed chance results on the first 36 occasions but significant positive scoring on the next 108 occasions.

The possibility of the newness of a task causing the disappearance of an ESP response is unfortunate for the experimenter in this field since it may mislead him as to the effectiveness of a change. It is a common device in experimenting to make some alteration in conditions which, on some hypothesis the experimenter wants to test, may be expected to eliminate the effect that is being studied. If the effect then disappears, the hypothesis is regarded as confirmed.

Although this method of experimenting is indispensable in psychical research, it must there be interpreted with caution. It may be the mere fact of change and not the character of the change that produces an alteration in scoring rate. When, for example, Soal introduced a few experiments under clairvoyance conditions into his series of telepathy experiments with Shackleton, he found that Shackleton dropped to a chance level of scoring and concluded that this percipient could not succeed under clairvoyance conditions. This may, however, merely be an example of a misinterpreted 'change effect'. To be sure that it was not so, it would have been necessary for Soal to persist with the clairvoyance condition of experimenting as Tyrrell did in his attempts to adapt Miss Johnson to the condition of working with a commutator in circuit.


CADORET, R., and PRATT, J. G. (1950),'The Consistent Missing Effect in ESP', Journal of Parapsychology, xiv, 244-56.

ESTABR00KS, G. H. (1927), 'A Contribution to Experimental Telepathy', Bulletin Boston Society for Psychical Research, v, Boston.

GURNEY, E., MYERS, F. W. H., and PODMORE, F. (1886), Phantasms of the Living, London (abridged edition, New York, 1962).

RHINE, J. B. (1934), Extra-Sensory Perception, Boston.

RHINE, J. B. (1941), 'Terminal Salience in ESP Performance', Journal of Parapsychology, v, 183-244.

RHINE, J. B. (1952), 'The Problem of Psi-missing', Journal of Parapsychology, xvi, 90-129.

RHINE, J. B. (1969), 'Position Effects in Psi Test Results', Journal of Parapsychology, xxxiii, 136-57.

THOULESS, R. H. (1935), 'Dr. Rhine's Recent Experiments on Telepathy and Clairvoyance', Proceedings of the Society for Psychical Research, xliii, 24-37.

THOULESS, R. H. (1949), 'A Comparative Study of Performance in Three Psi Tasks', Journal of Parapsychology, xiii, 263-73.

TYRRELL, G. N. M. (1935), 'Some Experiments in Undifferentiated Extra-sensory Perception', Journal of the Society for Psychical Research, 52-71.

TYRRELL, G. N. M. (1936), 'Further Research in Extra-sensory Perception', Proceedings of the Society for Psychical Research, xliv, 99-168.


Related Material

The Experimental Study of Survival: Dr. Robert Thouless

The Problem of the Reality of ESP: Dr. Robert Thouless

The Critics of ESP: Dr. Robert Thouless

The Beginnings of Psychical Research: Robert Thouless

Anecdote and Experiment: Dr. Robert Thouless

Home | About Us | Latest News | Biographies | Articles | Experiments | Photographs | Theory | Online Books | Links | Books | Contact Us | Glossary


Some parts The International Survivalist Society 2004