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 Experimental Study of Survival

 - Robert H. Thouless -

          THERE ARE a number of topics in parapsychology which remain in the observational stage because no one has so far been able to devise experimental methods for their study. In others the evidence is mainly observational although attempts have been made to submit it to experimental control. Amongst the latter is one problem that has, from the beginnings of psychical research, been regarded as one of its key problems, that of the possibility of obtaining evidence as to whether human beings survive, in any sense, the death of their bodies. That there is survival in some sense is taught by all the religions; it is also supported by a good deal of observational evidence. Yet to many it has seemed so certain that human consciousness is a function of an intact nervous system that they have ruled out as impossible the idea that any consciousness could survive the death and corruption of the nervous system.

This is not a situation to be resolved by argumentation even on a philosophical level. Philosophical arguments for or against survival turn on the implications of language. It is not in this way that questions of fact can be settled. We cannot decide what is the case by considerations drawn from the socialised habits of thought which underlie the implications of our language; rather we must be prepared to modify our habits of thought and our customary ways of using language in order to make them conform to whatever facts we may discover.

It is obvious that there is one kind of empirical evidence that could provide incontrovertible evidence of survival; this is the actual experience of our consciousness going on after our own bodily death. But we must wait for our death before we can know whether we have that experience; we cannot have it in this life.

Some people do, however, claim to have in this life an experience somewhat related to the post-mortem experience of continued conscious existence. This is what is called the 'out-of-the-body' experience (Green, 1968). This experience may occur while the subject appears to be asleep or unconscious but it may also occur in the waking state. Its distinguishing feature is the fact that the subject appears to be elsewhere than in his body, and the body itself may be seen as an object within his field of vision. Such experiences may seem to one having them, not direct evidence of the survival of death, but convincing evidence that he is of such a nature that survival is not an unreasonable expectation. They are not convincing as evidence of anything to other people than those experiencing them, since they rarely give verifiable information that would indicate that they were not merely hallucinatory. To give generally acceptable evidence of survivability it would be necessary that such experiences should be brought under experimental control. Such control has not yet been achieved.

The observational evidence for survival derived from the appearance of ghosts and other apparitions has not much force as evidence that people do survive their bodily deaths. There is much evidence that seeing a ghost is a genuine but rare experience, but little ground for supposing that a ghost is the surviving element of some deceased person. Even when what is seen is the apparition of some relative or friend who has recently died, there are strong arguments against the view that the spirit of the dead person is there as a visual object.[1]

[1] Tyrrell, G. N. M. Apparitions (London, 1942).

More cogent evidence comes from communications ostensibly from the departed that are received through mediums. The medium is a woman or man who is considered to act as intermediary between a living human sitter and a once living communicator. The contact between the medium and communicator may appear to be direct as when the medium speaks by direct voice, that is, apparently with the voice of the communicator. More commonly there is another intermediary, ostensibly in the spirit world, called the control. There is some experimental evidence, obtained by Whately Carington[2], that the controls are secondary personalities of the medium, although they may claim to be Red Indians, Greeks, Persians, and so forth. The control usually reports the presence of deceased relatives and friends of the sitters, gives messages from them to the sitters, and reports their answers to questions put by sitters. The medium may be in trance or semi-trance and may communicate what is said by the control either by word of mouth or by automatic writing or by some such method as spelling out by pointing successively to different letters of the alphabet.

[2] The Quantitative Study of Trance Personalities, I, II and Ill, Proc SPR, xlii, 173-240, xliii, 319-61, and xliv 189-222, 1935, 1936 1937.

There is no doubt that statements and answers to questions purporting to come from the departed are given in mediumistic séances and no reason for supposing (except in rare and purely fraudulent cases) that they are conscious fabrications by the medium. It has indeed been abundantly shown (in the case of Mrs Piper and other good mediums) that the information supplied about relatives of sitters unknown to the medium has gone far beyond the medium's normal knowledge. If their source was not the ostensible communicators, it would have to be attributed to the extrasensory capacities of the medium. Without a suitably designed experiment, it is difficult to see how a rational choice between these two alternatives can be made.

To suppose that the choice is simply between extra-terrestrial communicators and the medium's psi-capacities is to make the alternative too absolute. It is possible that both factors might enter into mediumistic communications as well as such other factors as the medium's own habits of thought and also the habits of thought and extra-sensory capacities of the sitters. If there is any factor of communication from the communicators, this may be diluted to any extent by such other factors, normal and paranormal. An answer to the key question of whether there is any evidence for communications from the departed in mediumistic material depends on the investigator's ability to filter out such an element from all such possible irrelevancies. If the question as to the reality of this element is answered affirmatively, there remains the further quantitative question of how large it is in comparison with other factors. A satisfactory answer to these questions demands more refined techniques of enquiry than mere attendance at mediumistic séances.

There may be real communication from the departed which is very imperfect as a result of dilution from such sources as the thought habits of the medium, the expectations of the sitters, and so on. The situation from the point of view of the communicator may well be as described by (ostensibly) F. W. H. Myers communicating by automatic writing through Mrs Holland: 'I appear to be standing behind a sheet of frosted glass - which blurs sight and deadens sounds - dictating feebly - to a reluctant and somewhat obtuse secretary'[3].

[3] Johnson, Alice.  On the Automatic Writing of Mrs. Holland, Proc SPR, xxi, 166-91, 1909.

In the same way, when Mrs Willett was the ostensible communicator through Geraldine Cummins, she referred to the communications as a 'mixed grill' composed of memories of the medium and the communicator, and spoke of herself as being in a sense compelled to select from the memories of the automatist (Cummins and Toksvig, 1965). If this was communicated by Mrs Willett, it is of special interest since her own experience in this life as a medium might have been a help to her in understanding the problems and difficulties of communication.

The evidence from communications received through mediums has appeared convincing to many people. Two lines of evidence may be distinguished: evidence by recognition and evidence by information.

The characteristic of evidence by recognition is that the sitter is convinced of the reality of the communicator by recognition of such personal characteristics as ways of speaking and even (in direct voice communication) of characteristic intonations. The feeling is that of being in direct contact with the deceased personality. Such experience of recognition seems understandably convincing to the sitter experiencing it. It can carry no conviction to anyone else, particularly when they remember how illusory experiences of recognition can be even in everyday life.

More commonly the evidence that convinces sitters of the reality of their communicators is their ability to give some identifying information. If someone were telephoning to us and we were uncertain whether he was the person he claimed to be, he might assure us as to his identity by producing such an identifying bit of information: 'I'm John Smith, you know, who used to ride a piebald pony called Peg.' If we are in doubt as to the identity of ostensible communicators, they may attempt to satisfy our doubts in the same way.

Not all information given in a séance is useful for identification; that part which can be used as evidence of identification is of very uneven value for that purpose. The informational content of a séance can be classified as follows:

1. Information that has no value as evidence because it is on matters that could be within the normal knowledge of the medium.

2. Information that is of little evidential value because, although true of the communicator, it is true also of many other people. Such information, for example, as 'Your grandfather died as an old man after a long illness' can be shown to be acceptable to a large proportion of sitters. On credulous sitters, it may have an effect disproportionate to its real value as evidence.

3. Information as to some particular fact known to the ostensible communicator which can be confirmed by one of the sitters or by some other person. This is the kind of information that obviously points strongly to the ostensible communicator being the person he claims to be. The evidence is not, however, coercive if we admit the possibility that the medium can get such information from the minds of the sitters by ESP. Information of this type is, of course, evidence of some kind of paranormal acquisition of knowledge, but not necessarily from other-world sources.

4. Information as to some fact (subsequently confirmed) which is known to the ostensible communicator but to no other person. This most striking kind of evidence does not often occur spontaneously although there have been cases. It is more easily arranged experimentally, as when a communicator tries to reveal the contents of a package he sealed before his death. In the early days of psychical researches such information was regarded as of very high evidential value since the early psychical researchers were inclined to regard telepathy as the only form of ESP to be considered. Since, however, we now know that psi-knowledge of an external fact is no more beyond the power of a gifted percipient than is psi-knowledge of another person's thought, this evidence is of no more force than is that of the third kind unless the possibility of the medium having acquired the crucial information by ESP has been in some manner excluded. This exclusion of the possibility of ESP can never be complete in a spontaneous mediumistic communication; it will be later considered how it can be achieved in an experimental situation.

Although it is, in principle, possible for authentic information of the third and fourth kinds to be explained by the sceptic as due to the medium's own extra-sensory powers, this explanation may appear, in the most striking cases, a highly improbable one since it assumes an ESP capacity in the medium greater than there is any other evidence for. The least that can be said for good spontaneous mediumistic evidence of survival is that it creates a strong prima facie case for some part of the communication coming from the ostensible communicator. Of recent cases which contribute strongly to this evidence, I should be inclined to pick out the 'Palm Sunday' case[4] and the communications received by Geraldine Cummins ostensibly from Mrs Willett[5].

[4] Balfour, Jean, The 'Palm Sunday' case: new light on an old love story, Proc SPR, Iii, 79-267, 1960.
[5] Cummins, Geraldine, and Toksvig, Jean. Swan on a Black Sea, London, 1965.

If the most striking mediumistic utterances do seem to show evidence that they may come in part from an other-world communicator, it remains true that they seem to be diluted to an unknown extent by the psychological this-world factors already mentioned. In the less striking cases these factors may predominate to such an extent that one can feel no conviction that anything is coming through from the ostensible communicator. In all cases, correct information, from whatever source, is mixed with much misinformation, and evidence by recognition is confused by an admixture of uncharacteristic responses. While such misinformation complicates the problem of evaluating mediumistic evidence, it is not inconsistent with real activity of a communicator if it is considered that the communication is, at best, a composite product in which such factors as the thought processes of the medium may play a part. It must also not be forgotten that the communicator himself may suffer from defects of memory and some changes of personality after his bodily death.

It is clear that there are enough uncertainties in spontaneous mediumistic material to suggest the desirability of an experimental attack on the problem of survival. Either evidence by recognition or evidence from information may be made the basis for experiment. The first of such experimental enquiries to be considered here is the attempt of Whately Carington to reduce to experimental terms the problem of identification of communicators by recognition[6]. The essence of his method was to use the responses of communicators to psychological tests as a means of identifying them instead of relying on the subjective impression of recognition by the sitter. The intention was to discover whether what was ostensibly a single communicator would give, through different mediums, responses which were more similar to each other than were the responses of ostensibly different communicators. If it were found to be so, this would give reason for supposing that a communicator x coming through one medium was the same individual as communicator x coming through another medium. It would not, of course, prove that he was the individual he claimed to be, but Whately Carington had also the idea of a more extended experiment which would provide evidence also on this question[7]. He had a plan of doing psychological tests on living individuals, and after their death to see whether, as communicators, their responses to the same tests showed a significant resemblance to their responses during their life. This experiment was, however, never carried out.

[6] Whately Carington, W. The Quantitative Study of Trance Personalities, I, II and Ill, Proc SPR, xlii, 173-240, xliii, 319-61, and xliv 189-222, 1935, 1936 1937.
[7] Whately Carington, W. Survival of death, Chambers's Encyclopedia, xiii, London. 1950.

In the more limited experiment on the resemblance of communicators coming through different mediums, Whately Carington thought that his case was proved. Further study of his numerical results showed, however, that this was over-optimistic; his positive results turned out to be the result of serious errors in Whately Carington's use of statistical methods (Thouless, 1937). When correctly evaluated, his results showed no more tendency for communicators to resemble themselves than to resemble other communicators. They provided no evidence that the different communicators were distinct personalities.

[8] Thouless, R. H. Review of Mr. Whately Carington's Work on Trance Personalities, Proc SPR, xliv, 223-75, 1937.

Whately Carington's experiment was a courageous and praiseworthy attempt to solve a difficult problem. There is no reason for taking its negative result as final. Other workers may repeat Carington's laborious plan of research by more fruitful methods and they may get more satisfactory results. No one seems as yet to have tried to follow the path laid down by Whately Carington; there seems here to be a field open for experimental research but it is not an easy one.

More commonly, parapsychologists have preferred to do experiments on identification of communicators by information. These have involved choosing the item of identifying information, generally by the use of a package containing information and sealed in the communicator's lifetime. Before discussing these 'sealed package' experiments, however, something should be said of another attack on the problem of survival which is, at least, semi-experimental. This is the system of 'cross-correspondences' which was reported in the Proceedings of the Society for Psychical Research from 1906 onwards.

If this was an experiment, it would appear to be one that was arranged from the other side of the grave. The general idea was that separate items of information were given through different mediums, items which were without significance when considered separately but fitted into a coherent whole when taken together. This feat was taken to be beyond the telepathic powers of the mediums concerned, and therefore to be evidence of a single mind acting on all the mediums, which mind had the necessary knowledge to arrange the connection between the items. The principal identifying characteristic of the items supplied was the amount of classical knowledge that was required for fitting them together; this knowledge was certainly possessed by the ostensible communicators: Verrall, Butcher, and Myers. I have not myself sufficient classical knowledge to judge the value of the evidence from the cross-correspondences, but some competent judges consider that they do demonstrate the unifying activity of the surviving mind of a classical scholar.

If this was an experiment devised by these scholars on the other side of the grave, I think it must be judged to be a badly designed experiment. It has provided a mass of material of which it is very difficult to judge the evidential value, and about which there are varying opinions. It reproduces, in fact, the defects of spontaneously gathered mediumistic material in a somewhat intensified form. A successful experiment should give a more clear and unambiguous answer to the question it is designed to answer than does spontaneous material; otherwise the experiment is not worth while. When judged by this criterion the cross-correspondences would seem to fail as an experiment.

The sealed-package experiments were attempts to make experimental the problem of identification by information. If the deceased person has put something into a package whose contents are known to no one else, then we have a specified bit of information which we may hope to recover from him through a medium after his death. A message was so sealed by F. W. H. Myers before his death, but the nature of the message was not afterwards discovered through a medium but only by opening the package. Sir Oliver Lodge left behind him a similar package but although there were many subsequent séances in which Lodge ostensibly communicated (in some of which I was the sitter) it proved impossible to get a plain statement from him as to what was in his envelope. When this was finally opened it proved to contain an account of a finger habit acquired from the five-finger exercises of his childhood which remained with him all his life and was, he said, known to nobody else.

It is not, perhaps, surprising that the communicator may find it difficult to remember the particular bit of information that he put into his sealed package. In this life, we use our material brains for the purpose of remembering. If there is a disembodied communicator, he has lost this material brain which he left behind in his grave. He no longer possesses this convenient mechanism for the storage of information; the brain may not be essential for remembering and yet the loss of it may interfere with effective remembering. Some of us are in the habit of remembering appointments by writing them down in a pocket diary; we could, no doubt, have remembered them without the diary, but, since we have formed the habit of relying on it, we shall find it difficult to remember our engagements if the diary is lost. It may be so also when we lose our material brains at death. The task of remembering a specific memory may be a difficult one; if anything of the sealed-package type is to be attempted, it is important that the experimental material should be simple, striking, and easy to remember.

It is odd too that, in the experiments of Myers and Lodge, better success was not obtained by the clairvoyant powers of the medium. Too much weight should not be attached to this failure, but, so far as it goes, it suggests that the psi-capacities of mediums may be less formidable than they are commonly assumed to be.

It is true that certain writers have claimed that neither the Myers package test nor that of Lodge were as complete failures as they were supposed to be, since the message delivered through the medium, although not identical with that enclosed in the package, had some relation to it[9]. In the Myers package, for example, both the sealed message and the séance reproduction of it had something to do with love but one was concerned with the house where lived the lady whom Myers loved, and the other was concerned with Plato's Symposium. The resemblance is somewhat remote and one cannot say with confidence that it is not accidental.

[9] Salter, W. H. F. W. H. Myer's Posthumous Message, Proc SPR, Iii, 1-32, 1958.

The mere fact that there can be such discussion means that the experiments have failed in their main object, for the evidence they have produced has the same sort of uncertainty as that produced in ordinary non-experimental sittings with mediums. The point of doing an experiment is to do something better than this; it should produce evidence which approaches the ideal of an unambiguous proof that communication was received that could only be accounted for on the hypothesis of a disembodied communicator.

The failure of the Myers and Lodge sealed-package experiments should be considered as a challenge to the psychical researcher to design something better.

There are a number of defects in these two sealed-package experiments. First, there is the fact that, if they had been successful, there would have been no way of ensuring that this success was due to communication of information from another world and not to the extra-sensory capacities of the medium. Arguments could, no doubt, be put forward on both sides, but there is no experimental discrimination between the two possibilities. Secondly, the experiment is not repeatable; once the package is opened, the experiment is over. Those in charge of the experiment may be doubtful when to open the package. If it is opened too early, the experiment may fail although the communicator could have got through the correct message in a later sitting. If the package has been opened and examined there can be no later sitting. What one needs is to devise a task for which there can be an indefinitely large number of verifications without spoiling the experiment.

It was in an attempt to remedy these defects that I devised what I have called the cipher test[10]. This was intended to fulfil the following requirements:

[10] Thouless, R. H. A Test of Survival; Additional Note on a Test of Survival, Proc SPR, xlviii, 253-63 and 342-3. 1948.

1. That the information to be communicated should be simple and easy to remember.

2. That it should be of a hit-or-miss type; the answer should be either clearly right or clearly wrong.

3. The likelihood of hitting on the right answer by accident should be negligibly small.

4. It should be possible to try out any number of wrong answers without spoiling the experiment as it is spoiled by the premature opening of a sealed package.

5. The design should be such that a discrimination can be made between success through the medium's ESP and that which must be attributed to communication from the departed.

For the purpose of this test, I have prepared in cipher two passages in the Proceedings of the Society for Psychical Research.[11] The passages are: INXPH CJKGM JIRPR FBCVY WYWES NOECN SCVHE GYRJQ TEBJM TGXAT TWPNH CNYBC FNXPF LFXRV QWQL and BTYRR OOFLH KCDXK FWPCZ KTADR GFHKA HTYXO ALZUP PYPVF AYMMF SDLR UVUB. The methods used for encipherment are described in the articles already referred to. The object of the test is to see whether, after my bodily death, I can communicate the keys by which the passages can be deciphered. It will not be my intention to communicate the content of the enciphered passages; that I have already forgotten. What is to be communicated is the keys required for decipherment: a passage from literature in the first case, and a pair of words in the second. The methods of encipherment are such that these keys cannot be discovered by any rational process.

[11] Vol. xlviii, pp. 258 and 342. There is also an earlier passage on p. 253 which is now withdrawn from the test because its method of encipherment proved not to be unbreakable by rational methods.

This test seems to fulfil the first four of the required conditions mentioned above. It does not matter how many wrong keys are tried out; the passage will not be read by means of them and the subsequent reading of it by the right key is not prevented by these failures. The test may not indeed entirely get over the difficulty that imperfections of post-mortem memory may make it difficult for my disembodied spirit to produce the required keys. But this difficulty is made as small as possible by the simplicity of what has to be remembered and by the fact that any number of attempts to remember can be tested.

The design of this experiment also makes it possible to fulfil the fifth requirement, that of discriminating between explanation by the medium's ESP powers and by a surviving communicator. This possibility arises from the fact that any number of erroneous attempts at the key can be made without spoiling the test. Attempts can, therefore, be made to get the keys through mediums while I am still alive. Several such attempts have been made and all have been unsuccessful. If, therefore, the keys are correctly given by a communicator purporting to be me after my bodily death, the fact that they are so obtained and could not be obtained during my lifetime, will be a strong (though obviously not conclusive) ground for supposing that that communicator is indeed myself.

In order to carry much weight, this experiment should be carried out by a number of persons. There are difficulties in the preparation of enciphered material and, so far as I know, only one other person has made a similar test. It is satisfactory that Dr lan Stevenson has devised a test on similar lines involving simpler material[12]. What is deposited in Stevenson's test is a combination lock, on which the combination has been changed by the person intending to communicate to a number known only to himself. The matter to be communicated is the number to which the lock must be set before it can be opened. This form of test has the advantage that the material is easily available and requires less labour in preparation than does a cipher. Already a number of individuals have contributed locks for this experiment.

[12] Stevenson, I. The Combination Lock Test for Survival, Journal of the American Society for Psychical Research, Ixii, 246-54. 1968.

All such experiments, derived from the original sealed-package tests, can go no further than establishing experimentally the fact of survival. The further problem as to the nature of what survives will be the next experimental problem in this field when the fact of survival is demonstrated.

Obviously the experimental study of survival has still a long way to go.

There are promising leads as to possible directions of experiment. The path opened by Whately Carington, for example, still awaits further exploration. It would be a big breakthrough if anyone could devise methods of communication that did not involve a human medium, but there is no hint yet of how such a development could take place.


"From Anecedotoe to Experiment in Psychical Research" by Robert Thouless (London: Routledge & Kegan Paul, 1972).


More articles by Robert Thouless

The Problem of the Reality of ESP
The Critics of ESP
The Pattern of ESP
The Beginnings of Psychical Research
Anecdote and Experiment
Can the Experimenter be Trusted?

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


Some parts of this page 2012