Parapsychology at school

Dr Bernard Auriol

(Montalembert, Toulouse, 1974)

Hypothesis

Our starting point was the fundamental hypothesis that parapsychological faculties are present in each and everyone of us (H1).

It was thus our wish to attempt to verify if these parapsychological faculties vary according to the age or the academic level of the participants : would younger children necessarily have better results than older ones (H2) ?

Thirdly, is the existence of Skinner's operant conditioning (of the bio-feedback kind ?) possible even when working with a small number of tests (such as 60 draws divided into four sets of 15). Such a learning process was to be carried out using our " Reinforcement Device " which we hoped would be seen to work with a restricted number of tests between the individual participants (H3).

Finally, we believed that there could be an influence caused by "role changing"; as the subject, who was previously agent, becomes percipient and vice versa ("percipient order effect") (H4).

Protocol

With the kind co-operation of Brother Cazère (†), teacher at the Lycée Catholique Montalembert in Toulouse, and of Brother Thénoz (Headmaster), we were able to ask in each class (educational regulations permitting) for the pupils to come forward as volunteers. We asked the pupils to group themselves into pairs of "good pals", believing this suggestion would bring about more favourable conditions than if the pairs had been put together randomly. In fact, it is of popular belief that the testing of the presence of parapsychological phenomena, has more positive results when working with couples who are emotionally linked.

The two children were thus brought into a room that the school had put at our disposal. This area allowed for sufficient space and provided an environment in which "cheating" (or cumberlandism of) the transmission of information would be unlikely.

Two stations of the "telepathic biofeedback" apparatus were set up, with an additional bag of 60 marbles at the agent's station. These were divided into 5 games of 12 marbles, each game being distinguished by colours : blue, green, yellow, red and purple.

The two players were told that there would be a test of thought transmission and that they would be asked to take turns in "sending" or "receiving" a piece of information which would be chosen out of five possibilities.

The experimenter then instructed the agent to shake-up the bag of marbles, draws one, plug in the corresponding colour-key, put the marble back into the bag and then to knock twice as a signal to the percipient that it was his turn to play. At this, the percipient was to endeavour to guess the target piece of information and then likewise to plug in one of the colour-keys.

Whenever the information "emitted" an received" coincided, the 2 pupils would be rewarded by the lighting up of 2 light-bulbs at the moment when the percipient plugged-in his key. In the case of non-performance nothing would happen, and the agent would have to take the next turn.

Each child had to "emit" 60 times in four series of fifteen goes, each set of series separated by a break of a few minutes. The expectation of success was 3 per series and therefore 12 (4x3) for the entire 60 tests.

At the end of the 60 tests, the players changed places so that the agent was now the percipient and vice versa. The experimenter, as well as keeping count of the number of successes, noted anything that was striking to him or her in each series.

Results

We were able to test 59 pairs of pupils in this way, all of which were masculine (see enclosed file). Each pair of subjects was accorded a number in the file and an identifier in the record of the compiled observations. The name and date of birth was taken of the subject who was first to play the role of agent. Likewise, the same details were accorded to the subject who was first to play he role of percipient, after which we noted their common academic level. Next came the recording of the number of successes for each of the four consecutive series of 15 draws (subtest 1, subtest 2, subtest 3, subtest 4) and their total (Total 1 = subtest 1 + subtest 2 + subtest 3 + subtest 4). We also made note of any peculiarities presented in the behaviour of the children. We went on to find the same readings for the series carried out after the roles were reversed : the agent having become percipient and vice versa (Total 2 = subtest 5 + subtest 6 + subtest 7 + subtest 8).

Are parapsychological faculties present (H1) ?

The statistic study was carried out using the software Flash and NCSS. The histogram (fig.1) shows the number of successes in series of fifteen tests, all the subjects having been "changed-over" (from A to B and then from B to A).

Fig. 1

The question is, if these results are compatible with the zero hypothesis, i.e., are they compatible with that which would result from a random draw ?

The expectation that we foresaw of such a draw is 3 (on 15, i.e. one chance out of five). The average count for the overall observation is 3.35 successes out of fifteen draws. (The standard deviation being at 1.70. If we compare this average to the expectation of a random draw, we have T = 4.52, which is very significant (p < .0001) (fig. 2).

We can therefore rule out the notion that these are the results of mere chance ! Out of all of the pairs of children who were tested in our protocol we can see that the trend of their answers, would imply a certain communication between them.

Histogram of “Total”

fig. 4

 

bins
1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
0-1.5	1.5-3	3-4.5	4.5-6	6-7.5	7.5-9	9-10.5	10.5-12	12-13.5	13.5-15	15-16.5	16.5-18	18-19.5	19.5-21	21-22.5	22.5-24
1	0	0	1	6	3	15	9	27	11	25	4	7	4	4	2
fig. 4 bis

 

Age Effect ? (H2)

Our wish was to attempt to verify if the parapsychological faculties varied according to the age or the academic level of the participants (H2).

We studied two parameters according to age (or class), i.e. the total of successes (out of 60 goes : "Tot") and the gradient (Gt) of the least squares line (which was established from the four series of fifteen goes each).

Our protocol shows that there is in fact no linear relationship between age and class of the two participants and their performance, or indeed their believed learning gradient in the four series of fifteen goes each.

In fig.5 and fig.6 we have displayed the number of successes obtained by each pair of pupils who participated in our experiment. This has been put together according to their class. In a random draw the average number that we would expect would be 12 (out of 60). We can see that the 4^th, 5^th and 6^th classes seem to have better results than the others. Should this fact be considered important ?

It at least calls out for further investigation.

Fig. 6

 

Classe	T	1	2	3	4	5	6	7	8	9	10	11
Median	9	11	13	9	14	14	16	11	13	12	14	11
Average	9	10.5	13.3	9.5	14.3	15.2	15.9	12.1	13.6	12.6	14.1	12
s.d.	3	1.8	2.1	1.5	3.3	4.6	3.8	4.5	1.3	3.2	3.3	5.7
Number	2	6	8	2	10	9	16	10	8	18	18	12
Fig. 6

 

Learning Effect ? (H3)

It has already been shown [1] that a multi-repetitive use of our " Reinforcement Device" allowed for significant improvement in the performance of two motivated people. What we wanted to do in this instance was to test the hypothesis according to which, a similar effect could be recorded even by a very brief use; even though the proportions would be much weaker, the number of subjects would perhaps compensate the fact that the sessions were brief.

If this were the case, we would then be able to note better performances in the last (and second last) series than in the first (or the second).

The study of the average value of each series according to its position in the set, shows an insignificant tendency upwards (average of the series 1 and 2 = 3.29; average of the series 3 and 4 = 3.42).

To give us an idea of this phenomenon, we can draw up a graph showing the number of successes in the series in question, according to its position in the order (fig. 7). Our results are compatible with the hypothesis of a zero gradient of linear regression. In fact, the correlation between the ordinal number of the test and its value is only 0.03 (probability = 0.47).

No. of individuals : 476	s.d. : 1.7	Test = 0.05 x No + 3.2
Ord. Nb. of tests :  1..4	Dispersion : 50.8	Correlation : 0.03
Test : 0..8	Average : 3.4	Test of hypothesis : zero gradient (No significant)
Fig. 8

 

Let us study each series more precisely.

However these reading convince us all the more of our belief that the zero hypothesis (i.e. H1 = 0) should be rejected, since the live representing the series can be seen to rise upwards. The rejection of this chance hypothesis, which has little predominance in the first series (p < 0.09), can actually be seen to occur at the second series (p < 0.05) and even more so at the last two (p < 0.02).

Furthermore, by studying the variances of the group we can see that the first series is noticeably distinguished from the second (p < 0.05) but not necessarily from the third or the fourth. The second is significantly different from the fourth but not necessarily from the third. There is however no real distinction between the third and fourth series [2] . When we look at the results collectively, we can see that they suggest certain variations in the sending (agent) or the receiving (percipient) of information during the four series.

For each set of four series (from the first to the fourth), we were able to determine a certain grading that we will refer to as the "gradient of this pair" (the linear regression calculated from the four representative points of the four series of this pair). Does the average value of this gradient [3] differ very much from 0 ? No !

Mean - Average	0.96
Lower 95% c.i.limit	0.82
Upper 95% c.i.limit	1.10
Std.error of mean	0.07
Fig. 9

 

Percipient Order Effect ? (H4)

Haraldsson laid great emphasis on the possible existence of a "Percipient Order Effect". This notion was drawn from a number of experiments during which two subjects would take turns playing the role of agent and percipient. "The first percipient tended to obtain ESP scores below the mean chance expectation, the second percipient above, with the resulting difference in ESP performance between the first and second percipients being statistically significant".

This effect, if it so exists, could be indicative of a learning process taking place : the couple in the second phase could benefit from the experience gained during the first; for example, the fact of having played the role of percipient would allow the new agent to better "convey" his message.

Equally, a subconscious relational phenomenon could also be at work such that the second percipient in view of the fact that he had already played the active "role" in the preceding phase, would allow himself to be that bit more confident in his role of percipient. If this were the right explanation, the effect would be inconsistent and would depend firstly on the subconscious dynamics of the participants and secondly of the way in which their roles were distributed and so on.

In Haraldsson experiment, the order of play would be determined by flipping a coin; one of the subjects would be asked for his preference, then the experimenter would flip the coin and the game would be played out according to heads or tails whether it was the subjects'choice or not.

In our Montalembert experiment, the choice was left up to the player, while the experimenter was content to simply indicate the roles to be chosen.

Descriptive Statistics       (Detail Report)

Multivariate Analysis

The Principal Component Analysis of the results considering each of the series of each pair (four + four) as one observation, enables us to note that the Totals of direction {A => B} and of {B => A} appear as orthogonal and thus are independent (at least as regards to the first two axis). We can also note that the first two axis can be explained almost entirely by these Totals (even used as supplementary variables), which results in a strong coherence between the four series belonging separately to each direction. The successes in the four series when A is agent and B percipient explain the first axis (X) whereas the successes in the four series of direction B to A explain the second axis (Y).

 

 

We can compare the total of the first direction (Total 1 : direction A B) with the total of the second direction (Total 2 : direction B A). Each one differs very significantly ( p < 0.003 for Total 1; p < 0.009 for Total 2) from the expectation ( = 12).

If we compare Total 1 (direction A B) in parallel to Total 2 (direction B A), the value of T is 0.47 which is not significant (probability = 0.64). The two totals have a coefficient correlation of 0.12. The difference of their averages is 0.3 in favour of Total 1 (direction A B) (that is the reverse of the predictions of "Percipient Order Effect").

We can also compare the gradients; Gt1_(A-->_B) = - 0.02 while Gt2_{( B--> A)}= 0.15. This implies a gradient which would be seven times higher in the second direction than in the first. This however is not important (F-ratio testing group variances : 1.01; probability = 0.48).

Discussion

1) We started out with the fundamental hypothesis that faculties were present in each and every one of us (H1).

This hypothesis, we were able to confirm via a selection of pupils from an academic establishment in our area; while a random draw would have given 3 successes out of 15 draws, we were able to obtain 3.353. This result, though small in appearance, is nevertheless statistically very significant (T > 4.5; < 0.0001).

2) We also wanted to verify if these faculties vary according to the age or academic level of the participants, i.e. would the younger children have better results than the older ones ? (H2).

We did not find any indication that this was the case.

3) Was the existence of Skinner's conditioning possible when working with a small number of tests ? (H3).

We were not able to fully establish this theory but we did note that the various series were not equal. This would seem to imply the idea that the variations that were observed on a small number of tests, in several series, could vary not only according to a larger capacity to communicate information, but perhaps also on a shifting attitude to this very capacity. For example, certain pupils did well, others systematically avoided doing well, while most of them seemed unsure and went from one attitude to another.

4) Finally, we believed that there could be an influence caused by "role changing"; as the subject, previously the agent becomes percipient and vice versa (Percipient Order Effect).... (H4).

This last hypothesis was not confirmed; at least not in its very simple beautiful form as is shown in the work of Haraldsson ! The observation made in H3 can also applied to H4 : while the series are structured differently, the results themselves do not vary significantly.

Conclusions

With our protocol (although perfectible…) ESP is widespread even despite its limited expression (poor signal/noise ratio).

This faculty cannot easily be improved on through a small number of tests (60) even when working with an appropriate " Reinforcement Device".

It doesn't seem to vary enormously according to age. The existence of an influence of the order of the roles in a pair of subjects taking turns to play agent and percipient ("Percipient Order Effect") was not able to be proved.