CIA-RDP96-00787R000400100021-6

Approved For Release 2001/03/07 :. CIA-RDP96-00787R090400100021-6
The Journal of
Parapsychology
VOLUME 35
MARCH 1971
NUMBER
REPETITION OF THE FRENCH
PRECOGNITION EXPERIMENTS WITH MICE1
By WALTER J. LEVY, JR., L. ALLEN MAYO, JR.,
EVE ANDRÉ, AND ANITA MCRAE2
ABSTRACT: Six series of exploratory precognition tests with mice and jirds
were carried out with the purpose of confirming the work of Duval and Montredon
in France. The experiment tested the ability of the animals to use their psi to
avoid jumping into whichever side of their cage would be electrified on each
forthcoming five-second trial. Out of 1,154 random-behavior trials, the animals
made a hit (avoided shock) 612 times (CR = 2.06; P <. 05, two-tailed).
Since the apparatus monitored the animal's position all the time it was in the
cage (in the French work there was only selective monitoring), it was possible
to reanalyze the data for evidence of psi in the animal behavior patterns. It
was found that the animals significantly avoided shocks when the just-previous
trial had been a nonshock trial (P <. 015). If the just-previous trials were
shock trials, however, the number of hits was not significant. Another test effect
was that trials following a period of low activity (fewer than three jumps back
and forth during the intertrial period) were also significant (P == . 015); trials fol-
lowing three or more jumps were not. When both successful conditions (non-
shock and low activity) were combined, they gave an even more significant
separation (P = . 006).
Mr. Levy is a second-year medical student at the Medical College of Georgia;
Mr. Mayo is a graduate student in zoology at Duke University; Miss André is
employed as a librarian in Victoria, Australia; and Miss McRae is a research assis-
tant at the Institute for Parapsychology .- Ed.
I. THE PRELIMINARY SERIES
In 1968, two French biologists, Duval and Montredon,8 published a
report on experiments which gave evidence that mice were able to
" This work was supported by the Institute for Parapsychology of the Founda-
tion for Research on the Nature of Man.
* The authors wish to express their thanks to Drs. J. B. Rhine and R. L. Morris
for valuable advice and encouragement and to Mrs. Dorothy H. Pope for much of
the work of compiling the paper. Miss Carol Schaber, Dr. Robert Brier, Dr.
Helmut Schmidt, and Miss Barbara Benton made important contributions at several
points.
* These are assumed names.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
2
The Journal of Parapsychology
C
use precognition to avoid shock as they jumped back and forth in
the two sections of a mildly electrified cage (1).
This work was outstanding in a number of ways. It won the
McDougall Award as an important scientific contribution; it came
from a respected biological laboratory; and it was based on work
with a species not hitherto tested by researchers, who had already
worked with cats, dogs, and pigeons. The French work was also the
first precognition test with any kind of animal and-what is more
important perhaps-it was the first completely automated test. At
least, it went further than any previous work toward eliminating
psi on the part of the experimenter which, as Rhine and Feather had
pointed out (3), stood in the way of a conclusive interpretation of
the earlier evidence for psi in animals.
From the point of view of methodology in psi research, however,
the most effective advance in the French research was its reliance
on the technique for selecting "random-behavior" trials, an idea first
introduced by Osis and Foster (2) but not applied as a primary
experimental device hitherto. This is a technique for sifting out those
trials in which it is obvious that ESP probably could not have been
operating. Osis and Foster had observed that when side habits dom-
inated the behavior of the cats being tested, the ESP scores were
lower than on the other trials. Duval and Montredon applied the
same principle to the mouse work; by eliminating those trials in
which the animal was obviously responding to non-psi factors-such
as shock-they obtained more significant results than when these
trials were included in the total. This procedure was reliable and
efficient, and it was an important step forward for parapsychology.
Over and above these specific points of importance in the French
work was the experimenters' own expressed hope that they had made
some progress toward discovering the "guinea pig" that is needed
for the greater effectiveness of the research program in parapsychol-
ogy. Their anticipation played a large part in motivating the authors
of the present work to attempt a replication. The automation in the
French tests seemed to have eliminated the experimenter as a source
of psi and to have centered on the animal as the producer of the re-
sults ; and it appeared reasonable that replication with other rodents
could be expected to follow.
.
:
.
1
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
1
:

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
.
3
Repetition of French Experiments with Mice
The individual differences of the French mice were not a con-
spicuous factor in the results. Since different strains had been used,
it seemed likely that there would be no great experimental risk in
using various strains of American mice in the attempted replication.
Similarly, there were modifications in the construction of the ap-
paratus, which was built with the aid of information provided by the
French authors.
The first experiment was begun and largely completed during the
summer of 1969 at the Institute for Parapsychology, although a com-
plete analysis of the results had to be left for a later time, since the
two senior authors had enrolled in graduate school.
The general procedure of this experiment was to put the test
animal into a box which was divided by a low partition into two
equal parts. The floor grid of the box was wired electrically and
there was an electronic selector which randomly chose one side or
the other as the side of the box which would receive an electric charge
for each trial. The animal was free to jump from side to side. If it
chose the "wrong" one, it received a shock; if it chose the "right"
one, it did not.
An important aspect of the experiment, as in the French work,
was the elimination of those trials in which psi probably was not
operating. This "nonrandom" behavior was of two kinds: static
and mechanical. A "static" trial occurred when the animal, even
though shocked, continued to stay in the same side of the box, possibly
because it did not feel the shock, or because it was too tired or too
frightened to jump, or for some similar reason. If the animal was
shocked and immediately jumped in response, this was to be ex-
pected; it was therefore called "mechanical" behavior. The "ran-
dom" trials, on which the evaluations were based, were the ones in
which the animal jumped for no apparent reason. Any change of
side after the five-second shock period qualified the following trial
as random.
The selection of random-behavior trials in the present experiment
differed from that in the French work in the following way: the
French monitored the animal's position only at the start of each five-
second trial. If the animal jumped once after the trial, their equip-
ment recorded a change of sides ; but if it jumped twice (or any even
0
.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6

€
Approved For Releast 2001/03/07 : CIA-RDP96-00787R090400100021-6 .
4
The Journal of Parapsychology
number of times) before the next five-second trial, it would be in
the original side of the cage and would appear not to have jumped at
all in the interval between trials. Our apparatus recorded the animal's
position continually and therefore made it possible to sift out many
more random-behavior trials from the total data; and if the scoring
rate was high, a greater incidence of psi would be indicated in the
data.
A further difference caused by their selective monitoring was that
their animal was in the dark except immediately before and during
the five-second trial. With our apparatus, the light beams, which
were the animal's source of light, remained on constantly.
Apparatus4
The testing apparatus comprised six components : a shock box,
a position indicator, a random number generator, a logic circuit,
a shock control, and a recording apparatus.
A. The shock box, or testing cage, was a clear, bottomless lucite
box measuring 612" x 61/8" x 6". It rested on an electrically wired
grid. A low barrier 11/8" high and 1/4" thick divided the cage into
two halves (A and B in Figure 1), each half wired independently.
(See Figure 1.)
B. On those sides which were at right angles to the barrier,
there were mirrors, used in conjunction with a light source and a
photocell in each half of the box to indicate the animal's position.
If the reflection of the light beam in the mirror was cut, the animal
was considered to be present; if it remained intact to shine on the
photocell, the animal was considered not there. The shock box, with
the position-indicating apparatus, was in a dimly lit room separate
from the remainder of the equipment. The separation varied from
an adjacent room, at first, to one that was two floors distant later
in the experiment.
C. The random target generator was an electronic device which
randomly picked side A or B as the target side for each trial. There
was one target selection (and therefore one trial) every minute. The
random number generator produced a rapidly oscillating sine wave
‘The main work of designing and constructing the apparatus, including the
random number generator, was done by Allen Mayo at the Institute for Parapsy-
chology.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
Repetition of French Experiments with Mice
5
Barrier.
Mirrors
Shock Grid
Light Source
Photocells
Side A
Side B
FIG. 1. Diagram of testing cage.
which caused a flip-flop circuit to go from state A to state B, or the
reverse, with the start of each full cycle it produced. Thus, when a
target was called for, the flip-flop stopped in either state A or B.
The randomness came from the fact that the error of the timing de-
vice calling for a target was tens of thousands of times greater than
the period of the wave. Thus, the side at which the flip-flop stopped
was not predictable other than by a chance distribution.
The randomness of the random number generator was checked
before the first series started. The results of several thousand trials
showed only a chance variation (CR = . 7), with slightly more tar-
gets for side B than side A. A further randomness check was carried
out later on the actual targets generated during the experiment. The
CR of .87 indicates that the distribution between the A and B sides
was well within the range of chance variation, with slightly more
targets on side A than side B.
D. The logic circuit integrated the information from the target
generator and the position indicator and sent a shock to the proper
side if one was needed. If the position of the animal and the selected
target side matched, a five-second shock was sent to the side the
animal was in at the initiation of the trial, and no shock was sent
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6

Approved For Releast 2001/03/07 : CIA-RDP96-00787R000400100021-6
.
1
.
6
The Journal of Parapsychology
to the other side. If the target and position did not match, no shock
was sent. Thus, the animal could be shocked only if he was in the
target side at the start of a trial, and only as long as he remained in
that side. However, at times the animal might have his body on one
side and his tail on the other, or he might be changing sides, so that
his position was ambiguous. In such a case both beams or neither
beam would be cut and no target would be chosen. When such a
"mistrial" occurred, no shock would be sent to either side. This part
of the apparatus provided safeguards against the possibility that
mechanical failure, such as a burned-out bulb, might give spurious
results. It was difficult for an animal to jump into a shock, since he
would have to jump at least twice within the five seconds to land in
the side of the cage that was given a charge.
E. The shock-control device determined the intensity of the
electric charge given to the animal. The shocking voltage was indi-
vidually adjusted for each animal at each run to give a voltage suffi-
cient to make him move away but not react strongly. During the
first part of the series the shock grid was charged by alternating cur-
rent; later, the voltage was supplied by a battery in order to provide
better control of the intensity of the shock.
F. The recording device was a Grass two-channel polygraph, one
channel of which continuously recorded the animal's position and the
other, his position at the time of the trial as well as whether or not
a shock had been sent to the animal. From this information the tar-
get chosen could be determined.
The lower channel on the chart was the position indicator. (See
Figure 2.) Side A of the cage was recorded at the bottom of the
channel; side B, just above it. The position of the pen indicated the
animal's position. For a move of the pen to count as a change of
position, it had to go at least 1 cm. past the midline into the opposite
half of the area. Each minute, the pen automatically centered for the
five-second trial period, and in checking the data this allowed a ma-
chine judgment (rather than a human judgment) to be made about
what position the animal was in at the initiation of the trial and
whether a jump occurred during the trial period.
The upper channel was a shock indicator. A full-scale deflection
of the pen indicated that the animal had received a shock; a half-
·
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
·

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
Repetition of French Experiments with Mice
ANIMAL POSITION SHOCK INDICATOR
3
1
4
6
2
-.
5
Side B Side A Side B Side A
1. Shock Indicator : "Shows which side of the cage the animal is on at the time of
the trial and whether it was shocked or not.
2. Animal Position Pen : Records the animal's position at all times, except during
the 5-second trial period. The pen moves from side A to side B and back as the
animal moves. During the 5-second trial period the pen moves to the center to
indicate the trial.
3. Miss : When the animal makes a miss, it is shocked and the shock is recorded on
the Shock Indicator by a long stroke of the pen to the side where the animal
is located at the time of the trial.
4. Hit: When the animal makes a hit, there is no shock and this fact is recorded
on the Shock Indicator by a short stroke of the pen to the side where the animal
is located at the time of the trial.
5. Mechanical Behavior: Results when the animal jumps only in immediate re-
sponse to a shock.
6. Static Behavior : Indicates no jumping activity between trials.
7. Random Behavior : Results when the animal jumps from side to side without
an apparent reason; i.e., not in immediate response to shock.
8. Mistrial : Indicates that at the time of the trial either both light beams reached
the photocell unbroken or that neither reached it. The animal may have been on
the barrier in this case, and no shock was sent to either side of the cage.
ANIMAL POSITION SHOCK INDICATOR
4
4
3
Side B Side A
2
--
7
®
Side B Side A
FIG. 2. Facsimiles of two polygraph charts.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
-
7
7

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
8
The Journal of Parapsychology
scale deflection indicated no shock. If the pen deviated toward the
center line (downward) it meant that the animal was on side A
at the beginning of the five-second trial; if the pen deviated away
from the center line (upward) it meant that the animal was on side
B. From this information it could be deduced which side had been
the target. During a mistrial no deflections of the upper pen oc-
curred, since no target and shock decisions were made. This also al-
lowed a machine judgment as to the occurrence of a mistrial.
Except for the recorder and the testing cage, the equipment was
enclosed in a box surrounded by an inch-thick layer of insulation.
The animal species used in the experiment were 14 white mice
(Mus musculus Swiss Webster) and 9 jirds (Meriones unguicula-
tus)." They were initially housed in homemade cages; but as work
progressed, they were moved twice into new cages. They received
as much personal attention by the experimenter as possible and were
frequently handled. A 12-hour light-12-hour dark cycle was used
except for the first two series, during which a 15-hour light-9-hour
dark cycle was used.
General Procedure
There were six series in the experiment, with 25 trials in each
run. Since the work was mainly exploratory, there was no preset
length for the series ; they varied from 7 to 41 runs. The breaks be-
tween the series were arbitrarily set by such events as relocation of
equipment and changes in personnel, but the series were basically
similar in design.
The experiment was terminated partly by the departures of the
senior authors and partly by the dismantling of the equipment-a
decision which was made by someone other than the experimenters
and which therefore eliminates the question of optional stopping.
The scoring of the data was completed during the winter, but the
results were not evaluated at that time. The following spring, when
the authors were again free to resume work on the experiment and
before the results were tabulated, the decision was made to combine
the series into a unit because of the lack of any clear basis for de-
" This animal is commonly but erroneously referred to as a gerbil. True gerbils
are closely related but different species.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
Repetition of French Experiments with Mice
9
marcation and to examine the degree to which the results gave pre-
liminary indications that we could replicate the French work.
Before each experimental session, the experimenter checked the
apparatus to be sure it was functioning properly. He then selected
an animal and recorded on the polygraph sheet the experimenter's
name, the date, time, and the animal's number. He then put the
animal in the testing cage and adjusted the shock level to the
animal's tolerance-just enough to make the animal move to avoid
the shock. The shock level was recorded on the polygraph and the
apparatus was run until one trial had taken place. Then the poly-
graph sheet was marked and the experimenter left the room for
at least 26 minutes in order to minimize the possibility that his own
psi might affect the results. (Trials in excess of 25 were not to be
included in the experiment.) After the run was completed, the ex-
perimenter removed the polygraph sheet from the machine and num-
bered the trials from 1 to 25, including mistrials. The sheet was then
placed in a file.
Later, an independent checker removed the chart from the file
and scored the data, transcribing the information onto another record
sheet. He noted which of the 25 trials indicated random behavior.
(See Figure 2.) Hits and misses were determined by the length of
the shock-indicator pen stroke. A miss (shock) was indicated by a
long vertical stroke; a hit (nonshock), by a short stroke. The num-
ber of hits and misses was noted, the score was recorded, and the
polygraph chart was returned to the folder to be rescored eventually
by a second checker who was unaware of the first checker's results
and who scored the data directly on the chart itself. Both sets of
scores were then compared by a third independent checker.
Results
Altogether there were 140 runs of 25 trials each, a total of 3,500
trials. There were 416 occasions when a mistrial occurred or the light
source failed and the animal's position could not be monitored. This
left a total of 3,084 trials, of which 1,154 were random-behavior
trials. When evaluated by chi-square, the results were not significant.
:
However, the experimenters were interested in finding out
whether there was any evidence that ESP was present or operating
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
..------
:

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
.
F
10
The Journal of Parapsychology
Table 1
GENERAL RESULTS
Series
No.
Trials
in Series
No. Random-
Behavior
Trials
No. Hits
on Random-
Behavior
Trials
Percentage
Random-
Behavior
Hits
1
478
201
104
51.7
2.
986
423
227
53.7
3.
437
210
105
50.0
4
224
139
78
56.1
5.
165
62
33
53.2
6
794
119
65
54.6
Total
3,084
1,154
612*
52.9
*CR = 2.06; P <. 05 (two-tailed).
at all in these series. Therefore an overall CR was computed. (See
Table 1.) In 1,154 random-behavior trials where mean chance ex-
pectation was 577 hits there were 612 times when the animal avoided
the shock (made a hit). This gives CR = 2.06, (P <. 05, two-
tailed).
Discussion
The experiment reported here was an exploratory attempt to
repeat the findings of the French work. It was terminated when
the authors returned to school in the fall semester, and as a first
effort, with suggestive results, it is a sufficiently successful beginning
to warrant the further studies, reports of which are to follow.
One of the most crucial questions in this research is the random-
ness of the targets-whether or not there was some favoring factor.
The empirical check preceding the experiment and the check on the
targets actually generated during the experiment showed only a
chance variation between the two target sides, thus eliminating non-
randomness as a factor in the results.
The likelihood that the results could be accounted for by psi on
the part of the experimenter appears to have been at least partially
reduced by the use of automatic equipment and the fact that the ex-
perimenter was out of the room and otherwise occupied while the
trials were being run. However, the theory of psi could easily be ex-
tended to allow it to work in the absence of the experimenter; and,
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
.....

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
· Repetition of French Experiments with Mice
11
unlikely though it seems, it can still be considered a conceivable
alternative. This reasoning suggests further measures to put it to
test. The second section of the paper goes further into this problem
of making a study of the differential responses of the animals that
produced the successes.
II. ANALYSES OF THE EFFECT OF THE TEST SITUATION
ON THE TEST ANIMALS
The analyses which follow are based on the results of Section I.
They were carried out by J. L. to explore the effect of the test pro-
cedure upon the ESP responses of the test animals.
Obviously, only a very inefficient demonstration of psi ability
was given by the animals, as the pooled totals show. The 1,154 ran-
dom-behavior trials gave a CR of 2.06, which represents a psi quo-
tient (PQ) of 3.68 as a measure of efficiency, whereas the com-
parable figure for the Duval-Montredon evidence is a PQ of 29.93.
It seemed reasonable to suppose that there might be differences
in the distribution of the small scattering of hits in our data that
would reveal a more meaningful pattern. This, after all, was the
type of reasoning that led to the original French findings and the
principle of random behavior as a method of concentrating the suc-
cesses in the analysis of the data. The selection of random-behavior
trials before the checkup was made was intended by the French
authors to eliminate trials on which psi was not to be expected under
the conditions of the test. This selection concentrated the successful
trials and produced the high PQ mentioned above. However, the at-
tempt to repeat the French experiments, with all the subtle variables
involved in such attempted replication, may have introduced dif-
ferences in method that weakened the confirmation, and it seemed
possible that there might be some areas of the test situation that were
more conducive to the operation of psi than others.
In following this line of thought, J. L. saw an opportunity for a
more complete analysis of the animal behavior than had been possible
in the French study. It has been pointed out that our apparatus
monitored the behavior of the animal continuously and gave readings
of jumps back and forth between trials that would not have been
recorded by the French apparatus, which monitored the mouse only
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6

Approved For Release_2001/03/07 : CIA-RDP96-00787R000400100021-6
12
The Journal of Parapsychology
at the start of the five-second trial. On inspection, our records gave
indications that the relative jumpiness, or nervousness, on the part
of the animals was noticeably greater right after a shock trial;
that is, a trial in which the animal made a miss and therefore re-
ceived a shock. Considering the probable psychology of the animal,
it seemed likely that right after a shock trial, his sensory and motor
systems would be more dominantly activated and that this might
interfere with the spontaneous operation of his precognition. J. L.
planned, on this basis, to divide the trials following shock from the
trials following nonshock. This breakdown makes up the first step
in the present study.
Comparison of Results Following Shock and Nonshock Trials
The after-shock vs. after-nonshock breakdown was made by two
checkers, who rechecked the original polygraph charts. They totaled
the number of times a shock trial was followed on the next trial by a
hit (avoidance of shock ) or a miss (shock) ; and the number of times
a nonshock trial® was followed by a hit or a miss. These data were
later rechecked by computer. The results are shown in Table 2.
Table 2
COMPARISON OF RESULTS FOLLOWING SHOCK AND NONSHOCK TRIALS
Condition
No. Random-
Behavior
Trials
Hits
% Hits
CR
Pa
After-shock trials
After-nonshock
trials
506
257
50.79
.36
.72
648
355
54.8
2.44
<. 015
.Two-tailed.
There were 1,154 random-behavior trials, 648 of which followed a
nonshock trial. (See Table 2.) Of these, 355 were hits (54.8%)
with CR = 2.44 (P <. 015, two-tailed). The PQ = 9.19. The
remaining 506 random-behavior trials were trials which followed
a shock. They contained only 257 hits (50.79% success ) ; PQ = . 25.
" A nonshock trial could be preceded either by a hit or a mistrial, as defined in
Section I.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6

Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
.
.
.
Repetition of French Experiments with Mice
13
Analysis of Intertrial Jumps
Encouraged by the results of the after-shock vs. after-nonshock
analysis, J. L. undertook a further breakdown of the same results
on the basis of the apparent "nervousness" of the animals as indi-
cated by the number of times they jumped back and forth during the
period between trials when the two compartments, of course, were
not electrified. This was a mere behavioral count of activity as re-
corded by the apparatus.
Since the animals could jump into a shock only with difficulty
(by jumping at least twice within a five-second period), one could
say they were reinforced for jumping. In jumping they eluded shock
almost exclusively; hence it was a beneficial behavior pattern for
them to adopt. As with the random-behavior selection, one might
expect those cases where the animals jumped less to show more ev-
idence of psi, since low-jumping is an activity less accountable for
by non-psi factors. Also, one would expect less jumping after non-
shock than after shock. So this was logically complementary to both
the "random-behavior" principle and the shock-nonshock idea.
From a rough inspection of the same original data, it appeared
that the best separation of the total trials would be to group those
in which the animal jumped back and forth not more than two times
between trials and compare them with those in which it jumped three
or more times.7 This analysis was carried out on the entire group
" The choice of this point to break the low-jump vs. high-jump trials was, in
addition to the one given, a mathematical one stemming from the difference be-
tween selective monitoring (the French method) and continuous monitoring. As
pointed out, the selective monitoring recorded the animal's position only at the
time of the trial. Thus, if the animal jumped once between trials, a position change
was recorded; but if he jumped twice, he appeared not to have moved at all.
All trials with even numbers of jumps, therefore, were lost and only trials with
odd numbers of jumps were counted. Again, since mechanical jumps were ex-
cluded, any trial after a shock, whether the number of jumps was odd or even, was
excluded. It was apparent that, while the scoring in our experiment gave mar-
ginally significant results on the basis of continuous monitoring, it was at chance
on the basis of selective monitoring. This suggested the occurrence of an uneven
distribution of psi; i.e., that more psi hits occurred in trials with certain types of
behavior than with others. Since the trials that are excluded in selective monitor-
ing necessarily have an even number of jumps, it would follow that in continuous
monitoring the trials with two, four, etc. jumps must be the ones that contributed
the positive scoring.
It was assumed that as the number of jumps increased, the number of trials
with that number of jumps would decrease. Thus there would be fewer trials with
four and five jumps than with one or two. This, together with the reasoning about
.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6

Approved For Releast 2001/03/07 : CIA-RDP96-00787R000400100021-6
14
The Journal of Parapsychology
Table 3
COMPARISON OF RESULTS FOLLOWING HIGH AND
Low NUMBER OF INTERTRIAL JUMPS
Condition
No. Random-
Behavior
Trials
Hits
% Hits
CR
Pa
Three or more jumps
Two or fewer jumps
$
395
759
199
413
50.5
54.5
.15
2.43
.88
.015
.Two-tailed.
of data, again following essentially the same procedure as before,
first by hand and then by computer.
Out of 1,154 random-behavior trials, there were 759 which were
preceded by two or fewer jumps before the trial. (See Table
3.) On that next trial there were 413 hits (54.5% success) with
CR = 2.43; P = . 015, two-tailed. PQ = 7.78. The remaining 395
random-behavior trials, which were preceded by three or more in-
tertrial jumps, gave only 199 hits on the next trial. The scoring rate
has a CR = . 15 and a PQ = . 06.
The Two Analyses Combined
Finally, on the discovery that both of the breakdowns did show
a meaningful concentration of the successes in a way that clearly
improved the efficiency level as measured by the PQ of the selected
category, the obvious next step was to combine the two analyses for
the still further improved concentration that would have to follow.
There were 422 trials which were preceded by a nonshock trial
and before which the animal had jumped only once or twice. Of
these, 239 were hits, 56.7% success, and a CR = 2.73 (P = . 006,
two-tailed). This has a PQ = 17.66. Most of the overall scoring
was from this type of trial.
jumping activity and shock mentioned above, made it seem that trials with one and
two jumps were likely to be scoring better than those with three or more. The one-
jump trials' positive scoring would be brought to chance by the scoring of three-
jump and five-jump trials, giving the result obtained with odd-jump trials. Since
the two-jump trials greatly outnumber four-jump trials, the overall result with
even-jump trials would be positive, as seen. This division of one- and two- versus
three- and more jump trials also gave roughly equivalent numbers of trials in
each group, and it was therefore predicted that the data would separate along those
lines.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
of
4

Approved For Releast 2001/03/07 : 'CIA-RDP96-00787R000400100021-6
Repetition of French Experiments with Mice
Table 4
RESULTS OF TRIALS FOLLOWING NONSHOCK
TRIALS AND TWO OR FEWER JUMPS
After Shock
After Nonshock
Total
Three or more jumps
169 trials
226 trials
395 trials
83 hits
49.1%
116 hits
199 hits
51.4%
50.4%
Two of fewer jumps
337 trials
174 hits
422 trials
759 trials
239 hits
413 hits
51.7%
56.7%
54.5%
CR = 2.73*
Total
506 trials
257 hits
50.8%
648 trials
355 hits
54.8%
1,154 trials
612 hits
52.9%
*P = . 006 (two-tailed).
As can be seen from Table 4, the other three conditions (after
shock, three or more jumps; after nonshock, three or more jumps;
and after shock, two or fewer jumps) gave lower rates of scoring,
as was expected.
Discussion
This is, in effect, a pilot study in which the method itself was
used for the first time and the decisions concerning certain points had
to be made with some arbitrariness. The comparison of trials fol-
lowing shock and nonshock left little latitude for choice of procedure
unless the breakdown might have been based on the results of two
trials (instead of one) following a shock trial vs. two trials follow-
ing a nonshock trial. This alternative method might, in fact, be
worth trying in the future. The selection of the dividing point for
the number of jumps made by the mouse between trials had to be
an arbitrary one, but there was little choice that would have left a
suitable division of numbers of trials for the two groups.
Furthermore, unless generalizations are confirmed in the analyses
of other data, they are very tentative and soon lost sight of. For-
tunately, in this case additional data were immediately available in
the work of other staff members of the Institute and in the results of
further tests in the present series. While it is permissible to say that
these later analyses confirmed the findings reported here, further
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
:
15

Approved For Release 2001/03/07 : 'CIA-RDP96-00787R000400100021-6
16
The Journal of Parapsychology
details must await publication of the other researches on which the
evaluations are based.
-
Not all the questions raised by these results can be very fully
discussed at this stage. Some will have to wait for the later con-
firmation, especially those having to do with the principles involved
in the bases for discrimination used in this breakdown. It does look
as though shock recency is a deterrent to the exercise of psi, and it
does look as though the hyperactivity of the animal, which may owe
something to the recency of shock itself, is a sign of unreadiness for a
psi response. One view of this situation that might be relevant is
that this immediacy may be a rough analogue of consciousness in
man and that these results may suggest some parallels between the
unconscious control of psi in man and the expression of psi in
animals.
If these points add up in a rational way, and especially in an
experimentally demonstrable way, the summing-up would run like
this: if the French workers have found a test animal which, with
automation and random-behavior selection, can produce a moderately
efficient and fairly reproducible type of psi demonstration, and if,
in addition, the selection of trials following nonshock in which the
animals are less activated to jump around enhances the psi effect,
there would appear to be some advance toward the desirable objec-
tive of dependable psi test performance.
This leaves the problem of what role the experimenters have
in the experiment. With each of the steps summarized in these
analyses the segment of the data in which psi is found is re-
lated more to the animal's situation and less to the experimenter's.
Conceivably the experimenter could affect the apparatus controlling
the frequency of shock in some such way as must be happening in the
experiments described by Schmidt (4) when he had subjects try to
influence the frequency of the lighting of lamps which were determined
by a similar selector system. The complexity of the operations is
comparable in both these experiments, and the question is still good
enough to deserve attention, as indicated earlier in the paper. A suc-
cession of steps such as the ones in these analyses removes the ex-
perimenter very considerably from the suspicion that he may be in-
volved. It was not the experimenter who experienced the shock and
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
.

Approved For. Releas& 2001/03/07 : CIA-RDP96-00787R000400100021-6
Repetition of French Experiments with Mice
17
it was not the experimenter who was jumping back and forth be-
tween sections in the apparatus; but these two considerations mark
off almost exclusively the area in which the efficient performance
was produced. Perhaps the experimenter is as much exonerated here
as is possible at this stage of psi research. However, the question
may be reconsidered when the further analyses are reported. Perhaps
there will be other indications of how to distinguish man from animal
in the psi test.
REFERENCES
1. DUVAL, P., and MONTREDON, E. ESP experiments with mice.
J. Parapsychol., 1968, 32, 153-66.
2. OsIs, K., and FOSTER, E. B. A test of ESP in cats. J. Parapsychol.,
1953, 17, 168-86.
3. RHINE, J. B., and FEATHER, S. R. The study of cases of "psi-trail-
ing" in animals. J. Parapsychol., 1962, 26, 1-22.
4. SCHMIDT, H. PK experiments with animals as subjects. J. Para-
psychol., 1970, 34, 255-61.
-1
Institute for Parapsychology
College Station
Durham, N. C. 27708
.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6
.

Approved For Release 2001/03/07 : 'CIA-RDP96-00787R000400100021-6
LOCATION OF HIDDEN OBJECTS BY A
MAN-DOG TEAM1
By J. B. RHINE
ABSTRACT: The experiment was carried out with two German shepherds and
their trainer who, as a team, attempted to locate empty land mine cases that were
buried in a straight line four inches deep in sand under 6 to 12 inches of water.
Each five-yard section, made up of five one-yard units, contained one buried mine,
the location determined by a random number table. All surface signs except sec-
tion and unit locations were removed by raking.
The test site was out of visual range of the man-dog team during the planting
of the mines. The dog was led over each section as a single trial, continuing to
retrace until it indicated a location by sitting. The experimenter who buried the
mine remained silent and invisible to the team during the test, but from a screened
position he was able to observe and record the team's responses. When an assis-
tant was present, the recording observer did not know the target locations.
Both dogs gave significant results. The presence of steady crosswinds, surface
ripples, and strong side currents in the water gave good assurance against ol-
factory cues. Special tests with colored liquids and streamers were made to
verify the adequacy of these controls.
Two hundred and three underwater trials gave 38.9% hits where 20% is ex-
pected (P <<. 001), but a decline set in with both dogs, even with no change of
personnel or overt conditions. Possible psychological factors are recognized that
suggest an essential link in the man-dog relation .- Ed.
The report given here describes experimental tests of the ability
of two man-led dogs to locate underground objects made to simulate
military land mines. The purpose of the investigation was to pro-
vide conditions that afforded no sensory guidance to the dog or man
but allowed the possibility of extrasensory perception.
The experiments were conducted by the Parapsychology Lab-
oratory of Duke University in 1952 at the request of, and under
contract to, the Engineer Research and Development Laboratories
(ERDL), Fort Belvoir, Virginia. The reason for the delay in pub-
lication of the report is that the material was classified and that
efforts to obtain declassification were unsuccessful until 1969.
The work was part of a larger program of investigation into the
psi capacity in animals. Some of the other studies have been re-
1 This paper was first presented on Jan. 3, 1970, at the Winter Review Meeting
of the Institute for Parapsychology.
Approved For Release 2001/03/07 : CIA-RDP96-00787R000400100021-6