CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
60
Interim Report- - Objective E, Task 9
December 1986
188
MEASURING REMOTE ACTION INFLUENCE ON THE
VERTICAL COMPONENT OF DUNALIELLA VELOCITY
By:
EDWIN C. MAY
BEVERLY S. HUMPHREY
SRI International
C. M. PLEASS
University of Delaware
Prepared for:
PETER J. McNELIS, DSW
CONTRACTING OFFICER'S TECHNICAL REPRESENTATIVE
SRI
International
333 Ravenswood Avenue
Menlo Park, California 94025 U.S.A.
(415) 326-6200
Cable: SRI INTL MPK
TWX: 910-373-2046
pproved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
r
L
SRI Internationa
C
Interim Report -- Objective E, Task 9
Covering the Period 1 October 1985 to 30 September 1986
December 1986
MEASURING REMOTE ACTION INFLUENCE ON THE
VERTICAL COMPONENT OF DUNALIELLA VELOCITY
By:
EDWIN C. MAY
BEVERLY S. HUMPHREY
SRI International
C. M. PLEASS
University of Delaware
Prepared for:
PETER J. McNELIS, DSW
CONTRACTING OFFICER'S TECHNICAL REPRESENTATIVE
SRI Project 1291
D
Approved by:
ROBERT S. LEONARD, Executive Director
Geoscience and Engineering Center
SRI
International
®
333 Ravenswood Avenue . Menlo Park, California 94025 . U.S.A.
(415) 326-6200 . Cable: SRI INTL MPK . TWX: 910-373-2046
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

H
į
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
:
ABSTRACT
The College of Marine Studies, Marine Biology Laboratory, of the University of
Delaware has been conducting experiments with Dunaliella for a number of years. The
researchers there claim that individuals are able to change the velocity of single algae cells
significantly. SRI International has formulated a different hypothesis to explain their putative
effect, i.e., individuals initiate experimental runs at a time during which the algae will naturally
swim at the required velocity. The ability of the individual to initiate data collection at the
opportune moment is called Intuitive Data Sorting (IDS). This report contains a historical
overview of the effort at the University Delaware and a detailed outline of a proposed test of
the IDS model.
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
I INTRODUCTION
As part of the general program at the College of Marine Studies at the University of
Delaware, Dr. C. M. Pleass developed a system to measure the swimming velocities of
single-cell marine alga. Having had a long time interest in psychoenergetic studies, Dr. Pleass
initiated a remote action (RA) investigation with the marine alga, Dunaliella, using specialized
hardware. As a consequence of his early effort and his promising results, SRI International
asked Dr. Pleass to conduct specific experiments to test a model of psychoenergetic
functioning that we call Intuitive Data Sorting (IDS). SRI let a two-year subcontract to the
University of Delaware to achieve this goal. The first year's task was to develop appropriate
statistics and protocols for a formal experiment that will be conducted during the second year
of the contract. This report contains the results of the first year's effort.
If we assume that a valid psychoenergetic phenomenon exists in Dr. Pleass' experiments,
then there are two heuristic models that are discussed below, RA and IDS, that might describe
the mechanism. (The proposed protocol will allow us to determine from the data whether this
assumption is valid.)
A.
Remote Action
One hypothesis proposed by Dr. Pleass is that (RA) accounts for the data. (The cells
are "forced" to conform to the "intent" of the participant in the usual cause-and-effect way.)
This hypothesis supposes that consciousness interacts with matter and, in particular, with living
systems. As a plausibility argument included in his report to SRI, Dr. Pleass invokes one side
of a controversy that suggests that physical systems (micro and macroscopic), in the absence
of measurement, exist as mixtures of all their possible configurations simultaneously. Although
this can be demonstrated to be true with quantum systems, two basic assumptions are required
to be true to extend the idea to the Dunaliella cells:
· Consciousness is a contributing factor in the quantum measurement
process.
· Living systems exist in "indefinite" states and thus qualify as quantum
systems.
1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
The prevailing thought in physics today is that neither of these assumptions are valid.
Although there are a few respected physicists who question the role of consciousness in the
physical world, virtually none of them believes that macroscopic bodies exhibit measurable
quantum mechanical effects. (There are circumstances in which macroscopic bodies exhibit
quantum effects -- superconductivity for example, but these are very special cases.) A
different hypothesis, therefore, will probably more appropriately account for the data.
B.
Intuitive Data Sorting
At SRI, we have been constructing a model to explain psychoenergetic data from a
different perspective. Based upon an interpretation of an experiment that we conducted in
1979,1* we propose that an informational process, rather than a causal one, is responsible for
certain putative RA phenomena. It is beyond the scope of this report to present a detailed
description of the model; rather, we will provide a broad overview.
We propose that humans can make decisions (by psychoenergetic means) to take
advantage of the natural and unperturbed fluctuations of a system. In the context of the
Delaware experiments, suppose that an individual is asked to "make" the Dunaliella swim
faster. Rather than "causing" the cells to swim faster, we suggest that the participant has
simply initiated the trial by anticipating when the Dunaliella were going to swim faster as part
of their natural fluctuation in velocity. Thus, the participant has capitalized upon natural
events, rather than "causing" anything to occur. We call this ability IDS.
We have been able to design the Delaware experiments in such a way as to distinguish
an IDS-mediated result from a causally mediated one. By definition, causal effects, on the
average, will effect the Dunaliella on a cell-by-cell basis. For example, they are phototropic
(i.e., each cell is attracted by light). The net effect of such a causal relationship is that
velocity averaged over a large number of cells will produce a very large statistical effect.
Because any informational processes (i.e., IDS), by definition, do not perturb systems (at least
classically), averaging over a large number of cells can only reduce any observed statistical
effects. Therefore, if the effects that Dr. Pleass observes are causal (i.e., RA), then averaging
over a large number of cells will produce a strong result. If the data are weaker as a function
of cell average (but remain significant), then the effect is likely to be informational (i.e.,
IDS).
Dr. Pleass' FY 1986 progress report is contained in Appendix A. The complete details
of the apparatus, analysis, protocols, and early results, provided by Dr. Pleass, can be found
*
References are listed at the end of this report.
2
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
in the annex to Appendix A. A brief summary of the past experimental details are provided
below.
3
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
II METHOD OF APPROACH
A.
Hardware
The Dunaliella are contained in a small test tube in a temperature-controlled, darkened
room. Their velocities are measured by a standard laser Doppler technique, which is sensitive
to the velocities of individual cells. The laser Doppler apparatus allows for velocity
measurements to be averaged over any number of cells ranging from one to an arbitrarily
large number. A typical velocity for a single cell is of the order of 20 micrometers/sec.
The data from the laser Doppler device are accumulated, displayed, and retained on an
IBM PC. All participant and monitor interactions with the experimental apparatus are
performed through the IBM PC.
B.
Statistics
The velocity data, like those for most living systems, have a large degree of variance, for
which there are a number of contributing factors, including:
· Living systems inherently exhibit a low signal to noise ratio.
· Dunaliellae are sensitive to environmental factors such as light and low
£
frequency electromagnetic radiation.
· Dunaliellae appear to exhibit a 24-hour circadian rhythm.
The initial attempt to analyze this kind of data included a double difference technique. Each
set of data involved (1) a putative RA period, during which the participant attempted to
modify the swimming velocities, (2) a control period immediately thereafter, and (3) two
matching periods collected when the participant was absent, i.e., pseudo-RA and
pseudo-control, respectively. The total run score by this technique is given by
Run Score = Σ(χcontrol - ΒΑ) - Σ(x]5
n
מ
j=1
, pseudo control
pseudo-RA
)
j=1
Xj
,
where the X's are the velocity averages for each data point, [j].
4
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Although many data were collected and analyzed by this technique, the statistical
method proved to be unstable; reliable probability assessments of any effects, therefore, were
difficult to obtain. The current method in use at the University of Delaware is a modification
of the above procedure. Run scores are accumulated for a particular participant, and the
resultant distribution is fit with a Gaussian curve. Global controls are collected under
conditions that emulate the experimental conditions as closely as possible (e.g., using the same
phase angle in the circadian cycle, the same RA/pseudo session spacing, etc.). These data
are also fit with a Gaussian curve. The statistical significance of the RA session is estimated
by the ratio of the variances between the periods of effort and the periods of global control.
..
5
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

₱
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
III RESULTS AND RECOMMENDATIONS
The subcontract with Delaware entails a two-year effort; therefore, there are no new
experimental results to report as of this writing. We recommend that a number of points be
considered before initiating any formal trials.
A.
Protocol
Before any tests for mechanisms are warranted, it must first be demonstrated that there
is some valid psychoenergetic effect. This obvious criterion affects the selection of
participants, the details of the participant-monitor interaction protocol, and the statistics. We
suggest that one or more pilot series be conducted, using the final protocol and statistics to aid
in the selection of "good" participants.
As described in the introduction, the tests for mechanisms require a series of trials using
a known causal component (e.g., a light source). We suggest that trials be conducted when a
participant is attempting to modify the algal swimming velocities; these trials should be
counterbalanced by identical trials during which the participant is replaced with a light source
adjacent to the cells.
B.
Statistics
As part of our recommendations for methods of analysis, we include a detailed
description of a technique that is commonly used in geophysics and that may be applicable to
the Delaware experiment.
1.
Background
Radar technology has provided a number of powerful techniques to perform time
series analysis upon "noisy" data similar to that described above. We will focus our attention
upon a technique that is in common use in geophysics.
In 1913, C. Chree introduced a method to correlate sunspot activity with terrestrial
magnetism.2 His approach was a particularly powerful one and exists today under a number
6
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
Approved For, Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
of different names (e.g., epoch analysis, and signal averaging). Signal averaging is a widely
used technique in such diverse fields as brain research and radar; therefore, quantities such as
signal-to-noise ratios, system sensitivity, and sampling limitations are available in the literature.
When applied to research with the Dunaliella cells, the technique contains one important
underlying assumption -- namely, that there is a time-stable (however small) component to the
cell velocities that may correlate with the participant's effort. To clarify the technique, we will
use an example from central nervous system (CNS) research. To measure a visual evoked
response, the following steps are performed:
1.
A light is flashed in a subject's eyes (the event).
2.
Occipital EEG is measured every millisecond for 250 ms (the time
series data).
3.
Steps 1 and 2 are repeated for approximately 100 flashes -- the data
from each step are averaged into the preceding data.
If there is no time-stable component of the occipital EEG to the light flashes, the
data will average to zero. What is found, of course, is that in some fraction of the
population, a persistent signal survives the averaging. That signal is called the visual evoked
response.
To apply this technique to the Dunaliella experiment, let us assume that we have
measured the cellular velocities, V, average over [q] cells (RA condition). Further, assume
that we have [n] number of events (similar to the light flash above) consisting of single button
presses by the participant to initiate a trial consisting of [m] velocity measurements.
Figure 1 shows a matrix of data that can be constructed from such a situation.
Vjk is the velocity measurement at time [k] for each button press [j], and tx (k =
1,m) represents the kth data point (after all the data have been detrended). As in the CNS
research example, we average the velocities across all button presses for each data point to
produce column means (the V-bars in this example). The row means, V-bars, will be
discussed below. A similar data matrix can be constructed when the apparatus is completely
unattended (the MCE condition). We must now determine if the RA matrix is significantly
different from the MCE matrix. An Analysis of Variance (ANOVA) technique is the
appropriate statistic to use to understand this situation. An ANOVA can determine:
1.
If the RA-condition column means differ significantly from the
MCE-condition column means (an ANOVA interaction term).
7
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
2.
If the variation observed in the RA-condition column means is
highly significant.
Button
Press
t1
12
·
t
m-2
t
m-1
t
m
1
V11
.
2
.
Vik
.
.
.
.
.
n
Vn1
Vnm
150
15
V2
V.
m
FIGURE 1 SIGNAL AVERAGING DATA MATRIX
The second case above is particularly interesting. If the number of button presses
is large, and if the RA-condition means are highly significant (i.e., - p < 10-15), then there is
strong evidence in favor of a causal interaction rather than an IDS process. IDS is not a
competing process if a system possesses a nearly infinite signal-to-noise ratio -- the case that is
required if item (2) above is true.
.
Under the condition that the ANOVA shows a significant interaction between the
RA and MCE conditions, then a second analysis is imperative when working with Dunaliella.
The problem is that the Viks are not necessarily statistically independent and cannot be
considered random variables in the usual sense. By ignoring possible persistent temporal
components to Vik (i.e., the 24- hour circadian rhythm), Forbush et al.3 demonstrated that a
gross underestimate of the residual variance may occur leading to a highly inflated F-ratio
from the ANOVA. In their test example in geomagnetic data, the inclusion of a correction
for a known temporal persistency reduced the F(26,3874) ratio from 4.96 to 1.16 (a p-value
increase from 10-15 to 0.3)! Forbush et al. suggest that a variance correction must be made.
Although there are a number of different techniques to provide a variance correction, the
technique that Forbush et al. suggest provides a quantitative estimate of a "coherence length"
beyond which a variance correction is not needed. Thus, it is possible to determine
8
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
. . .

1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
experimentally if the circadian rhythm (or any other periodicity) is confounding the
interpretation of the data.
C.
IDS Considerations
It is possible that the above technique will simplify the IDS aspect of the experiment.
Suppose a participant produces [n] button presses each resulting in [m] data points (average
velocity over [q] cells each). Suppose further that associated with such a data matrix is an
identical MCE matrix (possibly collected just before to the participant's first button press).
For the purpose of an IDS analysis, we consider this situation as [n] data points in which the
independent variables are all equal to [m] and the dependent variable is given by
- vị(m) ,
where Vý is the grand mean for the MCE matrix where the number of columns is [m],
and v, is the row mean for the jth of the [n] data points described above.
If we collect a number of such data sets by allowing [m] to vary, then using the IDS
formalism, the IDS and RA hypotheses will produce the curves that are shown in Figure 2.
In (m)
-
RA
In - (m)|
IDS
- MCE
FIGURE 2
PREDICTIVE CURVES FOR THE RA AND IDS HYPOTHESES
If our initial assumption is incorrect (i.e., that there are no psychoenergetic effects),
then the data will lie along the MCE curve shown in Figure 2. If, however, there is a
9
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

.
1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
psychoenergetic effect that can be described by an informational process, then the data will lie
along a straight line with the same slope (-0.5) as the MCE line, but with a significantly
different intercept. Should the effect prove to be causal, then the data are expected to lie
along a curve similar to the RA curve.
One advantage of working with a living system as an RA target, is that, in this case, a
known cause-and-effect relationship can act as an RA-like control. If the shift of the
velocity distribution because of the phototropic behavior of the cells can be determined, then
the phototropic data as a function of [m] can be calculated in advance.
Thus, if the phototropic data lie along the predicted RA-like curve and the data
produced by a participant lie along the IDS curve, then we should have compelling evidence
in favor of the IDS hypothesis.
10
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
IV CONCLUSIONS
By using the most powerful statistical techniques available, psychoenergetic effects if
present, can be observed in this exceptionally noisy system. If psychoenergetic effects can be
verified, we believe that the experiment conducted by Dr. Pleass constitutes one of the most
important tests of the IDS hypothesis.
11
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

4
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
REFERENCES
1.
May, E. C., Humphrey, B. S., and Hubbard, G. S., "Electronic System Perturbation
Techniques," Final Report, SRI International, Menlo Park, California (September 1980).
2.
Chree, C., "Some Phenomena of Sunspots and of Terrestrial Magnetism-Part II," Phil.
Trans. Roy. Soc. London, Vol. 213A, pp. 245-277, (1913).
3.
Forbush, S. E., Pomerantz, M. A., Duggal, S. P., and Tsao, C. H., "Statistical
Considerations in the Analysis of Solar Oscillations Data by the Superposed Epoch
Method," Solar Physics, Vol. 84, pp. 113-122 (1983).
12.
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

.
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Appendix
BIO LASER DOPPLER CONSCIOUSNESS RESEARCH
C. M. Pleass
College of Marine Studies
University of Delaware
Newark, Delaware
A Progress Report for 1985-1986
to
SRI International
Contract No. C-11498
. ..
.
13
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
BASELINE DATA
A study of the structure of 1985 baseline velocity data derived from cultures of the
motile marine alga Dunaliella has shown that any given data string may or may not exhibit
trend, abrupt mean shifts, periodicity, or any combination of these. Examples are given in
the Annex.
Only the diurnal (or circadian) rhythm is consistently manifest. Because it is long
compared to the duration of a typical set of psi runs, it normally appears as trend in the data
strings that report the velocities. Figures 1 and 2 show 24 hours of velocity and vector data
from Dunaliella. The correlation between circadian phase angle, the expected magnitude of
the observed velocity, and the direction of motion is evident.
Our 1986 baselines are "quieter" and rarely show abrupt mean shifts or periodicity,
other than the trend that derives from the circadian rhythm. This results from changes in
laboratory practice described in the section on microbiology.
PROTOCOL FOR AN IDS EXPERIMENT
A protocol has been developed for 1986-87 studies of the IDS hypothesis. The protocol
allows the participant (we choose to use the word participant, instead of the more usual term
operator, to try to encourage a synoptic view of the experiment) to determine the length of
the psi run, which is marked on the evolving data string by using preprogrammed keys on the
computer keyboard. The participant touches the key Fl when they are ready to begin the psi
run, and the key F10 at the end. The psi task has been to visualize the algae at the laser
beam crossover, and the variable is the resolved value of the algal velocity. We ask
participants using this procedure to end their runs as abruptly as possible, the moment that an
intrusive thought diverts their attention. They then read a dictionary while control data
evolves. Figure 3 illustrates a hypothetical set of three runs. Note that while the participants
are given a cue on the monitor at the end of the minimum control period B, they may be
engrossed in Webster's definition of fescennine and, therefore, choose to delay their next run.
14
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

-
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Absolute Velocity- 24 Hour Record
100
80
Velocity
msec-1x10-6
60
9 8
40
20
0
-
0200
0600
1000
1400
1800
Time in hours
FIGURE 1
Circadian Variation in Dunaliella in Swimming Velocity
During the First 24 Hours in Continuous 6328Å Light
15
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Direction of Movement- 24 Hour Record
43
0200
5600
1400
1800
38
% moving
down
33
28
23
18
13
Time in hours
FIGURE 2
Circadian Variation in Dunaliella in Direction of Motion
in the First 24 Hours in Continuous 6328A Light
16
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
1
17
Format for PK86
FIGURE 3
3
1
A
B
A
B
1
1
1
1
!
!
!
1
.
F1
F10
F1
F10
Fi
C
A = psi period
B = minimum control period. B = 2A
C = variable extension of the control period
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
-
-
-
0
0
,
1
1

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Feedback is provided at the end of each session, in the form of a plot of V psi minus V
control for each run within the set.
This procedure is likely to result in run times varying from 30 seconds to 10 minutes,
with clustering around 1 to 4 minutes. However, we want to start in this mode. In the
1986-87 proposal there is a flow chart of the proposed research and notes about future trials
with sequence lengths at equally spaced intervals of log (sequence length), and experiments in
which psi alternates with light.
STATISTICAL PROCESSING
Because trend in the data is the norm, statistical processing starts with detrending. The
method used through most of 1984-85 involved double differencing and is described in the
Annex. In early 1986, a more straightforward technique was developed; all data are regressed
onto a line, and the slope of this line is then used to detrend each datum. Somewhat to our
surprise, detrending in this way does not blur the structure of plots of run scores (Figures 4
and 5).
After detrending, the data are processed in the simplest possible manner by using the
difference in the mean velocity during the psi and control periods as the "score" for the trial.
Because there will normally be a "score," even if there is no psi mediation, individual values
of [psi-control] do not relate exclusively to the null hypothesis that there is no psi effect; they
are qualitative indicators. Histograms of these scores are also inconclusive until they are
compared to equivalent histograms prepared from data strings that have no psi mediation at
all. In 1986 we have chosen to take this "global control" data on the following day at exactly
the same circadian phase angle (time). The technique used is to transfer the time markers
from the real psi and control data string to the global control string. This global control data
accumulates while the experimenter is doing other work in adjacent laboratories, blind to the
process.
One way of comparing accumulated experimental and global control data to see if there
is evidence of a psi effect is to describe Gaussians that best fit the histograms, and then
compare the experimental to the global using a F test. We are presently moving away from
this approach because the distributions of experimental data are so non-normal, but it is
informative to use this approach to check that detrending by regression does not change
results originally obtained by detrending using double differencing.
18
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
-52
-57
-62
-67
-72
-77
E
-82
0 1 2 3
4 5 6 7 8 9 10 11
12
13 14 15
TIME
FIGURE 4
Standard Error Bars For Test 706 - Raw Velocity
17
12
7
.
2
.
t
-3
-8
-13
0
1 2
3 4 5 6 7 8 9 10 11 12 13 14 15
-
TIME
FIGURE 5
Standard Error Bars For Test 706 - Detrended Velocity
19
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

.
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Figure 6 records F tests on the output from these two different analyses. Because the
data are not normally distributed the p values which result from the F test are only indicators,
but they are both low, and remarkably similar.
The PK86 data comes from much quieter baselines (see microbiology) and the global
data are quite a good approximation to a normal distribution, with X2 = 0.44 (Figure 7). The
experimental data (Figure 8) have the characteristic non-normal distribution that we are
beginning to recognize as evidence of response to a stimulus. Once again, the F test is only
an indicator, but the numbers are quite encouraging.
F =
EXPERIMENTAL
GLOBAL
= 2.64, DF = 176/158
p < 10-6
(Statgraphics limits!)
Note that like the old WAVE data given in the Annex, the PK86 data have larger
variance in the experiment. PK85 data still seem to reflect one participant's tendency to calm
the algae (Annex, page 16). As the other individual data bases grow, we will be able to sort
this out.
Spontaneous variation in the duration of our 1985-86 psi runs has provided an
opportunity for a preliminary examination of the variation of the psi score with sequence
length. Figures 9 through 12 show plots of PK85 and PK86 scores against sequence length
with the corresponding globals. (Watch the varying scale on the ordinate caused by the
software used). As yet, we do not feel we have enough data or enough spread in sequence
length to form even tentative conclusions: that will be a task for 1986-87.
20
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
GLOBAL CONTROL
PK85 F =
EXPERIMENTAL
CASE 1 - PK85 data detrended and scored by double differencing:
(Vcontrol - Ýpsí) - (V]
pseudocontrol
- vp
pseudopsi
-
control
F = 1.64, 204 DF
p = 2×10-4
CASE 2 - PK85 data detrended by regression, then scored as:
Vosť
psi
· V
-
control
F = 1.707, 204 DF
p = 7x10-5
FIGURE 6
F Tests On Best Fit Gaussians From Histograms of Scores
21
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

-
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
ENTER THE NAME OF THE VARIABLE CONTAINING YOUR DATA: DELTA
NUMBER OF OBSERVATIONS = 158
(2 MISSING VALUES EXCLUDED)
SAMPLE AVERAGE = 0.066076
SAMPLE VARIANCE = 26.755
SAMPLE STANDARD DEVIATION = 5.1725
MINIMUM VALUE = - 12.353
MAXIMUM = 15.167
RANGE = 27.52
LOWER AND UPPER QUARTILES = - 3. 1363 2.8868
INTERQUARTILE RANGE = 6.0231
MEDIAN = 0.029543
COEFF. OF SKEWNESS = 0. 11154
STANDARDIZED VALUE = 0.57237
COEFF. OF KURTOSIS = 3.0251
STANDARDIZED VALUE = 0.064358
Press ENTER to continue.
Frequency Histogram
PK86 - Global Single Differences
16
12
8
.
4
C
-13
-7
-1
5
11
17
Difference - (PSI - Control)
ESTIMATED PARAMETERS: 0.066076 5.1725
CHI*2 GOODNESS-OF-FIT STATISTIC = 15.147 WITH 15 DEGREES OF FREEDOM
PROBABILITY OF A LARGER VALUE = 0.4409
A= V"psi"
1
V"control"
FIGURE 7
Global PK86 - Detrended By Regression
22
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
ENTER THE NAME OF THE VARIABLE CONTAINING YOUR DATA: DELTA
NUMBER OF OBSERVATIONS = 176
(O MISSING VALUES EXCLUDED)
SAMPLE AVERAGE = - 0.031364
SAMPLE VARIANCE = 70.551
SAMPLE STANDARD DEVIATION = 8.3995
MINIMUM VALUE = - 50. 165
MAXIMUM = 49.546
RANGE = 99.71
LOWER AND UPPER QUARTILES = - 3.5909 3.0301
INTERQUARTILE RANGE = 6.6209
MEDIAN = - 0.083973
COEFF. OF SKEWNESS = - 0.48565
STANDARDIZED VALUE = - 2.6303
COEFF. OF KURTOSIS = 18.356
STANDARDIZED VALUE = 41.585
Press ENTER to continue.
.
Frequency Histogram
PK86 - Experiment Single Differences
52
39
26
n
13
-
-
0
-51
-27
-3
21
45
69
Difference - ( PSI - Control)
ESTIMATED PARAMETERS: - 0.031364 8.3995
CHI*2 GOODNESS-OF-FIT STATISTIC = 50.005 WITH 6 DEGREES OF FREEDOM
PROBABILITY OF A LARGER VALUE = 4.6902E-9
A = Ýpsi - V
psi
-
entro?
control
FIGURE 8
Experimental PK86 - Detrended by Regression
23
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Simple Regression of DELTA on NUM
Parameter
Estimate
Standard
Error
T
Value
Prob.
Level
Intercept
-4.6399
1.7187
-2.6996
7.5254E-3
Slope
0.38384
0.10726
3.5785
4.3169E-4
Analysis of Variance
Source
Sum of Squares '
Df
Model
1308.7197
1
Mean Square
1308.7197
F-Ratio
12.8059
Error
20848.099
204
102.197
Total (Corr.)
22156.818
205
Correlation Coefficient = 0.24304
Stnd. Error of Est. = 10.109
Regression of DELTA on NUM
PK85 - Single Difference
53
33
.
13
.
-7
.
-27
0
10
20
30
40
50
NUM
(
A = Score; NUM = Sequence length
FIGURE 9
Experimental PK85 - Detrended By Regression
24
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Simple Regression of DELTA on NUM
Parameter
Estimate
Standard
Error
T
Value
Prob.
Level
Intercept
6.6804
2.2524
2.966
3.3755E-3
Slope
-0.45958
0.14074
-3.2654
1.2812E-3
Analysis of Variance
Source
Sum of Squares
Df
Mean Square
F-Ratio
Model
1878.9294
1
1878.9294
10.6627
Error
36123.957
205
176.214
Total (Corr.)
38002.886
206
Correlation Coefficient = - 0.22236
Stnd. Error of Est. = 13.275
Regression of DELTA on NUM
Global PX85 - Single Difference
70
40
10
-20
-50
-80
0
10
20
30
40
50
NUM
(
4 = Score; NUM = Sequence length
FIGURE 10
Global PK85 - Detrended By Regression
25
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Simple Regression of DELTA on NUM
Parameter
Estimate
Standard
Error
T
Value
Prob.
Level
Intercept
-0.48485
1.1132
-0.43553
0.66402
Slope
4.5057E-3
9.1509E-3
0.49238
0.62342
Analysis of Variance
Source
Sum of Squares
Df
Mean Square
16.988159
.242439
Error
7777.9715
111
70.0718
Total (Corr.)
7794.9596
112
Correlation Coefficient = 0.046684
Stnd. Error of Est. = 8.3709
Regression of DELTA on NUM
PK86 - Single Difference
69
49
29
AHHH₫
9
-11
-31
-51
0
100
200
300
400
500
NUM
(
A = Score; NUM = Sequence length
FIGURE 11
Experimental PK86 - Regression
26
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
F-Ratio
Model
16.988159
1

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Simple Regression of DELTA on NUM
Parameter
Estimate
Standard
Error
T
Value
Prob.
Level
Intercept
3.8801E-3
0.75413
5.1452E-3
Slope
2.0449E-3
6.1989E-3
0.32987
0.9959
0.74212
Analysis of Variance
Source
Sum of Squares
3.4990537
Df
Mean Square
3.4990537
F-Ratio
Model
1
. 1088168
Error
3569.2564
111
32.1555
Total (Corr.)
3572.7555
112
Correlation Coefficient = 0.031295
Stnd. Error of Est. = 5.6706
Regression of DELTA on NUM
Global PK86 - Single Difference
17
12
.
7
2
-3
-8
-13
0
100
200
300
400
500
NUM
(
A = Score; NUM = Sequence length
FIGURE 12
Global PK86 - Regression
27
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
PARTICIPANT POOL CHARACTERISTICS
Our present pool of participants (33) is drawn from professionals (both scientists and
not), young adults, and non-professional adults.
There are about eight principal participants who will probably do most of the SRI work.
None profess psychic ability. All have come to us by "networking" with others closer to the
research. We have held an informal lunch-hour seminar to allow them to meet and comment
on the program. Psychological evaluation of the participants will be carried out by Dr. David
Saunders of MARS Measurement Associates. Formal psychological evaluations will not begin
until a participant has completed an experimental series (10 sets) or shown unmistakable
evidence of long-term commitment.
COMMUNICATION WITH SRI
Our PK85 and WAVE data bases and the corresponding global control data have been
shared with SRI International and Dr. Jessica Utts. Dr. Edwin May, SRI, suggested an
interesting approach to data processing based on a technique of epoch analysis first reported
by Chree, * and now widely used in identifying rhythms in geomagnetic data. Because two of
the authors of a recent substantial paper on Chree's analysis are from the University of
Delaware, they were contacted, and a seminar was held. While the conclusion was that
sequential events in our present data files were independent, and that Chree's analysis was
unlikely to be immediately useful, it may well come in handy in future work with low-level
conventional stimuli (like light), applied rhythmically. The seminar served a second very
useful purpose: it exposed the work to three more senior faculty in the Physics Department.
The response was positive and encouraging. The climate seems to be changing, perhaps
because the data are so robust and the physics of the method so tightly defined.
The computer network seems to be functional. Dr. Dean Dey uses it to leave messages
at SRI, and we will try to use it increasingly. I am hoping that all modes of communication
can be enhanced in 1986-87; we want to keep in close contact to promote the free exchange
of ideas.
*Chree, C., "Some Phenomena of Sunspots and of Terrestrial Magnetism - Part II,"
Phil. Trans. Roy. Soc.
London, Volume 213A, pp. 245-277 (1913).
28
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
MICROBIOLOGY
During 1985, we made some substantial improvements in our technique for culturing
Dunaliella and transferring aliquots to the laser room. By using laminar flow hoods, sterile
techniques, and extensive rinsing of vessels that have been acid-cleaned, we have reached a
point at which cultures can be kept in the laser room for more than 24 hours without
appreciable deterioration. This technique gives us the option of inserting new samples just
before 4:30 p.m. each day, so that we are ready for data collection the following day.
Baselines obtained in this way are much quieter. This effect can be observed in Figure 8, the
global control for the 1986 data. We are pleased to find that the differences that represent
the "scores" are almost normally distributed.
CONCLUSION
A protocol has been developed that allows the compilation of data relating sequence
length and psi "score." Pilot runs contain strong evidence of a psi effect, and the results
have indicated some tentative, but interesting, suggestions of variation of score with sequence
length. Details of work to be carried out in 1986-87 will be discussed before data collection.
29
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
ANNEX 1
30
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

,
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
BEHAVIORAL RESPONSE OF MICROORGANISMS TO psi STIMULUS
PART II: STATISTICAL ANALYSES OF DATA FROM DUNALIELLA
C. M. Pleass
University of Delaware
College of Marine Studies
Newark, DE 19716
N. Dean Dey
University of Delaware
College of Marine Studies
Lewes, DE 19958
ABSTRACT
A consciousness experiment in which the Doppler shift of He/Ne laser light was
used to describe changes in the velocity and vector of a marine alga, Dunaliella, was
reported by Pleass and Dey in 1985. Because the subject of the consciousness
experiment is living, we expect strings of baseline velocity and vector data which are,
at some level, inexplicably time-variant. This complexes the statistical procedures
which must be used to analyze the data.
This paper examines the variation in baseline data strings, and describes two
alternative statistical procedures which have been used to determine the probability of
consciousness effects. Two levels of control are applied, allowing global comparison of
frequency distributions of experimental scores with similar distributions derived
artificially from baseline data. In both cases the null hypothesis is that there is no psi
effect. The data quite strongly suggest the rejection of the null hypothesis.
31
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

-
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
BEHAVIORAL RESPONSE OF MICROORGANISMS TO psi STIMULUS
PART II: STATISTICAL ANALYSES OF DATA FROM DUNALIELLA
INTRODUCTION
The College of Marine Studies of the University of Delaware have over a decade
of experience with the culture of marine microalgae. Initially, these cultures were
prepared as experimental diets for oysters in controlled environment mariculture. In
1982, Pleass and Dey began a series of studies of the response of a hardy green alga,
Dunaliella to psi stimulus, using the Doppler shift of scattered laser light as a rapid,
accurate measure of the change in swimming velocity and direction of individual cells.
The apparatus and techniques used were described at the 28th Parapsychological
Association Convention and published in the Proceedings (Pleass and Dey 1985). This
reference should be taken as Part I in an ongoing series. It contains secondary
references relevant to the experiment per se.
In summary, cells of Dunaliella are approximately 10-5 m in length. Using
whiplike flagella in a motion similar to breast stroke, they swim at velocities up to ca.
2x10-4 msec-1 (Figure 1). The laser Doppler apparatus uses a He/Ne laser in the
mode illustrated in Figure 2. The measuring volume is the ellipsoid where the split
laser beams cross. It has a volume of ca. 1x10-6 cc. Cultures of Dunaliella can
easily sustain 106 cells per cc, with intercellular spacing of the order of 10-4 m.
Under these conditions, bursts of scattered light from the laser crossover will normally
correspond to the passage of one individual cell. On the rare occasions when the
measuring volume contains more than one, the observed velocity will be the mean.
Data rates are quite high, ca. 70 sec-1, depending on the cell concentration and the
average motility. The apparatus is single-component: it will record velocity and
direction along any one chosen axis. Since algae frequently migrate vertically in
response to circadian rhythms, the vertical axis is normally used, and it may be assumed
unless an alternate is specified. The apparatus is interfaced with an IBM PC and
baseline data are normally stored as the average of 100 or 200 velocity readings.
Laser Doppler studies of the circadian rhythms displayed by motile marine algae
have been described (Pleass and Dey 1985). Baseline data describing algal velocities
must be assumed to contain trend, rhythm and nonrandom noise, and this complexes the
preparation of statistics describing the results of experiments whose objective is to
record behavioral response to an exogenous stimulus. This paper will focus on this
problem using results obtained from consciousness experiments with Dunaliella.
The most fundamental point, which is rarely found explicitly stated in the
literature, is that any experiment with a living system is by definition, irreproducible.
In fact, the hard physical sciences which derive from observation of inanimate
structures, and the application of logic, encourage an invalid extrapolation: that precise
replicability is a sine qua non of experimental science.
Theoretically, every observation must ultimately reduce to a probabilistic
conclusion, in part because of the Heisenberg uncertainty principle, and in part because
no system can be conceived which is perfectly stable through time at any temperature
above absolute zero. This point has been widely recognized in the writings of many
distinguished physicists and philosophers. Jahn and Dunne have brought these together
in an excellent treatment of the "quantum mechanics of consciousness." (Jahn and
Dunne 1984) which may provide a starting point for a formal theory of consciousness.
32
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

1
2
Approved For Release 2000/08/09: CIA-RDP96-00789R002200030001-4
1
1
Flagella at the front end, approximating a breast stroke.
Jerky forward motion. Examples are Dunaliella and Chlamydomonas.
1
1
Flagella at rear end, rotating like a propeller on a
natural universal joint. The body counter-rotates slowly.
Figure 1
Two common propulsive mechanisms used by microorganisms
33
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
3
Detector Fringe Mode
Bran Cott
0
T -
Ortector Reference Mode
·
1 .
-
Lover
Probe Volume
Transmitting Optic
x -
d./2 .
-02/2
Figure 2
Single-component, fringe mode LDA probe volume
34
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
-

4
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
Nonliving systems change their visible form through time in response to physical and
chemical laws. Since the kinetics of many of the reactions involved in these changes
are slow, relative to human lifetimes, we may conveniently speak of replicable changes
such as the expansion of a bar of pure metal which is subjected to a standard rise in
temperature.
This allows us to distinguish living systems as those which change measurably in
response to endogenous stimuli within the time span of human observation. Challenging
these definitions is interesting. A bacterium falls in the category of living, and a
macromolecule does not. The key to the separation is that we can, at least in theory,
explain the movements of the macromolecule as the result of environmental
(exogenous) stimuli such as molecular collisions and changes in van der Waals forces.
When we observe a bacterium we cannot escape the conclusion that endogenous stimuli
are involved in determining its behavior. It may absorb chemicals from solution to
acquire information suggesting a preferred direction of motion, but that by itself
cannot explain its time-variant behavior such as its "run-and-twiddle" pathway.
This commentary is relevant to the analysis of velocity and vector data from
Dunaliella. Because it is living, we must expect data from our BLD experiments which
is inexplicably time variant. Only in the most general case such as 24 hour circadian
rhythm will a consistent pattern be visible. If data from successive psi and control
periods are to be accumulated, statistical protocols which "see through" trend and
rhythm become most important: without them meaningful results cannot be expected
unless the data extends many times the length of the longest cycle, and runs are
randomly distributed in time.
The general question of rhythmic changes which could possibly be anticipated by a
participant in a psi experiment must be. thoughtfully addressed. In biological
experiments which have paralleled our psi research we have sought for, but failed to
find replicable rhythms other than the circadian or "diurnal". This was an important
prerequisite to the experiments to be discussed. If, for example, time series analysis
had revealed a subtle replicable rhythm with a ten minute period, this information
would allow an informed participant who wished to create an artificially high score to
execute psi runs at ten minute intervals. Fortunately, the only relicable rhythm has
such a long period (24 hours) that it cannot be used to advantage. It is manifest in the
data strings as a trend. Note that most participants choose to work in the mornings at
approximately the same circadian phase angle. The data is detrended by our statistical
protocol.
The most relevant reference to other current work seems to be to May et al.,
(1985), and references therein. These authors offer evidence for an informational
model called Intuitive Data Sorting (IDS) in which individuals "sort" locally deviant
sequences from a random sequence into significant and non-significant bins. Although
their data were derived from a meta-analysis of binomial data (Radin et al., 1985) the
concept of IDS provides an intellectual "foothold" for someone interested in examining
our data. Secondary references within these two papers are also useful.
RESULTS
Since the statistical results developed in this section are quite robust, the reader
may wish to consult Appendix I which describes the precautions taken to ensure target
system and data security. This information complements the description of the
laboratory and the method given in Pleass and Dey (1985).
35
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
5
Baseline Data Properties
The understanding of the statistical properties of our baseline data stream is an
essential and ongoing endeavor. Baseline data, meaning a stream of velocity readings
taken by the computer when neither participant nor experimenter are present in the
laboratory, have been collected and subjected to numerous visual inspections (via time
plots) and more quantitative analyses. Note that the examples given in this section are
chosen to represent extreme cases. With that caveat, the following properties have
been found to exist in our "normal" baseline:
A. Noise - the variation of velocity readings around an average value. This
variation is neither a constant nor a constant percentage of the average. Standard
deviations (o) of baseline velocity readings range from 5 to 35x10-6 msec-1 with
coefficients of variation 100g/x from 10 to 60 percent. A particularly disconcerting
example of baseline noise is shown in Figure 3, where noise appears to consistently
increase for approximately two hours.
B. Trends - where velocity is consistently increasing or decreasing for periods of
several hundred to several thousand readings. Figure 4 shows two examples of long
term (decreasing) trends in baseline velocity readings.
C. Periodicity - repeating cycles of varying frequency. Periodicity includes
strings of velocity readings with significant autocorrelation. Figure 5 is an interesting
example. In the raw velocity readings, Figure 5A, periodicity is nearly obscured by
noise. However, Figure 5B shows significant autocorrelation(s) in the same series even
after 20 lags. Finally, Figure 5C, a plot of cumulative Z score versus time, shows a
definite cycle in the data with a period of approximately 450-500 readings. Note that
no replicable frequencies characteristic of the organism or its environment (other than
the diurnal or circadian) have been found.
D. Trends and periodicity - combinations of repeating cycles with consistently
increasing or decreasing average velocity readings. Figure 6A shows both trend
(increasing) and cycles in a sample of over 1300 baseline readings. Also, Figure 6B
shows significant autocorrelations through 50 lags for the same series, with some slight
periodicity visually apparent in the correlogram itself.
E. Mean Shift - an obvious and relatively abrupt shift in the average velocity.
This characteristic may be a "discontinuity" in an apparent long range trend and may or
may not include periodicity. An example is shown in Figure 7.
General Observations of Baseline Data
. All of above, items A thru E, exist, or can exist in the data stream at any
given time.
. Existence of one or more of these phenomena can only be determined after the
fact, i.e., after the data have been collected and analyzed.
. No a priori adjustment of data to "compensate" for these phenomena is
conceivable. They must all be treated as inherent properties of the "baseline".
. Statistical treatment of the data, e.g., to test the null hypothesis of no psi,
must be "robust" enough to cope with these issues.
36
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
. . . .. .... ...

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
6
GURE 3
Baseline Data - (Averages of 10)
240
200
160
JUNOUMA
120
8 8 9
80
40
0
o
200
400
600
800
Datum Number
37
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
7
TIGURE 4A
Baseline Data
10
.
30
20
10
0
3
6
9
12
15
(X 100)
Datum Number
FIGURE 4B
Baseline Data - ( Averages of 10)
100
8 8 9 &
80
60
40
20
200
400
600
800
Datum Number
38
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

,
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
FIGURE SA
Baseline Data
240
200
160
& 8 2
120
80
0
0
4
8
12
16
20
24
Datum Number
(X 100)
FIGURE SB - Baseline Data
Estimated Autocorrelations of Velocity
1
0.5
0
-0.5
-1
0
10
20
30
40
50
Lag Number
FIGURE SC - Baseline Data
Cumulative Z Score from Velocity
120
Le
UDI N
2 8 88
70
20
30
80
-130
no
4
8
12
16
20
24
Datum Number
(X 100)
39
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

r-
.
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
9
FIGURE 6A
Baseline Data
200
18
160
140
120
0
3
6
9
12
15
(X 100)
Datum Number
TICURE 6B - Baseline Data
Estimated Autocorrelations of Velocity
1
00044000000
0.5
0
....
-0.5
.
-1
0
10
20
30
40
50
60
Lag Number
40
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
10
FIGURE 7
Baseline Data
120
8 8 8 8 9 8
10
&
50
40
Lu
SHOULD
0
wo
0
3
6
9
12
15
(X 100)
Datum Number
41
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
11
. All of the above points argue for the use of "global" comparisons and relatively
large sample sizes to provide meaningful statistical tests of the null hypothesis.
Statistical Analyses
Two experiments known as PK85 and Wave will be discussed. Both involve
marking the ongoing data string in similar formats, illustrated in Figure 8, using
preprogrammed keys on the computer keyboard. The psi interval is followed
immediately by a control interval of the same length. After a pause of random
duration, pseudo psi and pseudo control intervals which have identical dimensions to the
real run are automatically marked in the data file. In each analysis the null hypothesis
was that there was no psi effect extant, and the data were detrended by using the
double difference
Σ (Χ
control
ntrol -Xpsí) - (X)
pseudocontrol
trol - Špseud
pseudopsi'
-
as the score for the run.
PK85
Additional details of the PK85 protocol are given in Pleass and Dey (1985). The
participants were asked to try to visualize the algae in the vicinity of the laser
crossover during the psi period using informal, light-hearted imagery, while excluding
all unrelated conscious thoughts. The initial procedure required ten runs to complete a
set. The random interval (Figure 8A) was limited to a maximum of up to 5000 raw data
points. After one run cycle was completed, a symbol would appear in the bottom left
corner of the monitor, indicating that another run could be initiated whenever the
participant elected.
The time required to complete a set of ten runs was 1.5 - 2 hours. The
participants soon learned that selecting the shortest psi period possible would shorten
the length of the set. Typical psi intervals ranged from 30 seconds to 2 minutes. After
discussions with participants the number of runs in a set was reduced to five, which
could be completed in approximately one hour. This helped to eliminate the "mid-set
blues": the feeling that there were still many more runs to do.
The statistics routinely calculated were the differences in the means and standard
deviations. The results were presented to the participant at the conclusion of each set,
and saved to file with the raw data. To accumulate from set to set scores were
calculated for each run from the double difference (mean control - mean psi) - (mean
pseudocontrol - mean pseudopsi) and these scores were developed into histograms with
the best fitting Gaussian overlaid. Figure 9 reports the results of 22 sets, treated as
205 runs.
A significant step forward occurred when the results of these experiments were
compared to results artificially generated from data collected when no psi experiments
were in process. In this additional level of control, which will be referred to as "global
control", the data file dimensions of the sequence psi/control/spacer/pseudo-psi/pseudo-
control were taken from the markers identifying the real psi experiment data string and
transferred to data taken without psi intent, either before or after the set. Note that
although overnight data were available, we chose not to use these for global control
42
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 ; CIA-RDP96-00789R002200030001-4
Figure 8
PK85 and Wave Procedures
Null hypothesis: no significant difference between
= (s control - Sosi) - SIS
pseudocontrol - S
pseudopsi
where S is any statistic such as the mean
A. PK85
RUNX
RUNX+1
Ongoing Data
String
F1
F2
F1 F2
B. WAVE
RUNX
PSI Period
-
-
Random Space
Control Period
Pseudo PSI Period
Pseudo Control Period
Ongoing Data
String
RUNX+1
Same data dimensions as RUNX
F.
43
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
13
TICURI 9: PX85 Experiment Results
205 Runs
SCORE=( MEAN CONT. MEAN PSI)-
( MEAN : PCONT. - MEAN ?PSI)
LACHODHZUY
33
22
11
0
-80
-50
-20
10
40
70
SCORE ( micrometers/sec)
FIGURE 10: Global PK85 Experiment Results
205 Runs
60
SCORI={ MEAN: CONT. - MEAN PSI)-
( MEAN: PCONT. - MEAN PPS!)
-
45
AAHOOHZUY
30
15
-100
-60
-20
20
60
100
SCORI ( micrometers/sec)
.
44
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

* 1
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
14
since the circadian phase angle differs so much from that extant during the sets, which
were carried out during the day. Figure 10 illustrates the global control "scores".
To develop a probability value we compared the best fit Gaussians representing
the scores from the real experiment with the global control, using an F test.
o "global
2
apsi
= 1.64
With 204 degrees of freedom this gave a probability of a chance difference less
than 2x10-4, suggesting that the null hypothesis should be rejected.
WAVE
This modification to the PK85 protocol was a consequence of discussions with our
participants during the informal debriefing period which normally followed each
session. It is illustrated in Figure 8B. In it the participant was asked to begin his psi
effort at his own pace, at any time after the computer signaled its readiness to accept
another run. This warm-up and psi period had no set length; indeed participants often
took several minutes warming up to their psi effort. We asked them to touch the Fl key
on the computer keyboard when they felt their psi effort was at a strong peak, and to
immediately turn away to the trivial control period task. Reading a dictionary was
found to be an effective way of quickly transferring attention. In this way, we hoped to
observe the release of psi pressure. It was suggested that the movement to press the F1
key be a response to a subconscious cue, and none found this difficult. Once the F1 key
was pressed, the monitor screen was turned blue during collection of the control, the
spacer segment, the pseudo-psi and the pseudo-control data. At the conclusion of this
period, the monitor screen turned grey, signaling readiness for another run. The peak
psi period was taken to be the 25 data records (2500 raw data points) preceding the F1
keypress; thus all Wave runs have the same data dimensions. As usual, the control,
pseudo-psi and pseudo-control all had the same dimension as the psi. We continued to
use the same null hypothesis, algorithm, and format for feedback.
Initially, participants were asked to complete ten runs for each set during one
visit to the laboratory. This took nearly two hours, due to the manner in which most
participants chose to undertake their psi effort. As a consequence, the required effort
for a set was shortened to five runs per set. Preliminary analysis of this data was
conducted at the conclusion of each set, and graphical displays of differences in means
were provided as feedback, just as in PK85.
Quite striking results were obtained from indiviudal sets (Pleass and Dey 1985).
Run scores were then compared with global controls. Figures 11 and 12 show the data,
with best fit Gaussians. In this case the experimental results have a much larger range
than the global results:
F =
· psi
a global
2.
= 5.99
45
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

4
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
15
FIGURE 11: WAVE Experiment Results
293 Runs
124
SCORE= ( MEAN CONT. - MEAN :PSI ) -
( MEAN PCONT. - MEAN PPSI )
93
62
31
0
-320
-200
-80
40
160
280
SCORE ( micrometers/sec)
FIGURE 12: Global WAVE Experiment Results
293 RUNS
100
SCORE=( MEAN: CONT. - MEAN PSI ) -
( MEAN PCONT. - MEAN PPSI)
75
50
25
OL
-110
-70
-30
10
50
90
SCORE ( micrometers/sec)
46
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
16
The probability of obtaining the result by chance is very low, p = 10-6, again
suggesting that the null hypothesis should be rejected.
It is appropriate to ask if the two "global control" distributions are similar, within
experimental error, since the sample numbers are quite high, and it would be difficult
to explain any difference. Applying an F test to them we find
a Wave global control
F =
o 2PK85 global control
= 1.11
p = 0.2
This gives us substantial confidence in the chosen procedures.
DISCUSSION
Since PK85 and Wave differ substantially in "feel" and in the method of defining
the psi and control periods, the inversion in the F ratios is not unreasonable. There is
also evidence that some participants have a general tendency to reduce mean velocities
and biological "noise", while others enhance these data properties. Only three
participants contributed to the PK85 data shown, with a run ratio of 20:1:1, and it may
be that the data reflects one persons "signature". In contrast, fourteen participants
contributed more evenly to the Wave data. As our data base expands, we hope to obtain
a better understanding of participant-related variables.
It is appropriate to use the most cautious interpretation of the results: the physics
of the method used in the research is straightforward, but it is conceivable that there
are alternate explanations for the differences between experimental and global data,
since the global data cannot, by definition, be taken at the same time. The only method
which can be used to obtain global data at exactly the same circadian phase angle is to
take it from a different diurnal period. This possibility will be examined.
The insignificant difference between the distributions of global control data from
the two experiments (p = 0.2) helps to validate the statistical procedures. However, it
is reasonable to question the use of the F test, since the histograms of scores contain
several extreme values. This test is used in an exploratory mode, for want of a better
method of describing the differences between the two data sets. Suggestions for
alternate methods of statistical processing would be most welcome.
CONCLUSIONS
Two statistical analyses of psi runs carried out with the green alga Dunaliella
have been reported. They examine the significance of changes in psi and control period
statistics, derived from strings of velocity data. Taken at face value, the results
suggest the rejection of the null hypothesis that there is no psi effect. However, this is
the first experiment of this type to be attempted, and the results should be treated with
caution until they are substantiated by further research and various alternate statistical
treatments.
47
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
17
ACKNOWLEDGEMENTS
The consistent support of the McDonnell Foundation has enabled this work. We
would also like to thank our colleagues at the Princeton School of Engineering and
Applied Science for continued inspiration, and many excellent ideas, and Ms. Anju
Joglekar, Ms. Lisa Knight, and Mr. Gary Defibaugh for their assistance with the
analytical work.
48
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
.

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
18
REFERENCES
1.
Pleass, C. M. and N. Dean Dey. 1985. "Using the Doppler effect to study
behavioral responses of motile algae to psi stimulus." Proceedings of the 28th
Parapsychological Association Convention, August 1985, p. 373-406.
2.
Jahn, R. G. and B. J. Dunne. 1984. "On the quantum mechanisms of consciousness
with application to anomalous phenomena." Technical Note PEAR 83005.1,
Princeton Engineering Anomalies Research, Princeton University, School of
Engineering and Applied Science.
3.
May, E. C., D. I. Radin, G. S. Hubbard, and B. S. Humphrey. "Psi experiments with
random number generators: an informational model." Idem, p. 235-266.
4.
Radin, D. I., E. C. May, and M. J. Thompson. "Psi experiments with random
number generators: Meta-analysis Part 1." Idem, p. 199-234.
49
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
19
Appendix 1
TARGET SYSTEM AND DATA SECURITY
The lab building
Chosen to be free from extraneous vibration and electrical noise, the lab is
located on a sand dune known as Beach Plum Island, adjacent to the CMS campus at
Lewes, Delaware. To access the lab one must cross the Broadkill River in a small boat.
There are no other projects ongoing in the lab.
The laser room
. The culture of Dunaliella, the laser, and the transmitting and receiving optics are
all mounted on a vibration isolation table in a very heavily insulated room.
Quantitative data describing the residual vibrations on the table under the
conditions of the experiment were obtained using an integrating FFT
oscilloscope. When compared to similar FFT's of the Doppler shifts from a culture
of Dunaliella on the same table (Figure A-1) it is evident that the vibration
isolation is effective.
.
There are no motors or lights other than the laser in the room with the subject
culture. The laser power supply, the tracking electronics and the PC are in the
adjacent room.
. The thermal insulation makes the room almost soundproof. Wind noise outside the
building is totally imperceptible. The laser tube is a minor heat source, and to
ensure a constant temperature at the culture vessel the room is cooled by water
from dedicated wells, circulated through the room and returned to the aquifer
under the lab. This avoids the problems of sound, e.m. field, and air turbulence
associated with air conditioning. The temperature of the well water may change
by 5ºC over the year, but no change is detectable over the period of a run.
Room temperature is routinely recorded.
. The door to the laser room was barred by a thick sheet of urethane foam
insulation, which was pressed into place in the early morning immediately after
the culture had been changed. The participant could not remove this without
being observed by the experimenter.
The participant's room
. This adjoins the laser room. The participant may close the door, isolating
themselves visually from the experimenter. Until 1986 there were a number of
things which a dishonest participant might have tried to do in an attempt to
create an artifically high score:
1)
Knock (vigorously) on the wall between the two rooms during either the psi or
control period. However, this would be clearly audible to the experimenter.
50
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
20
2)
"Play" with the knobs on the electronics adjacent to the computer keyboard.
This would normally throw the equipment out of lock, causing the data string
to terminate. If by some remote chance it did not, the data might show a
sudden mean shift. However, the probability of a participant being able to
repeat this without losing control is virtually zero. Since the experiment
appears to have potential, we have now (1986) placed all the electronics in a
glass fronted locked cabinet.
3)
"Play" with the computer keyboard. Nothing the participant could do at the
keyboard could give them access to the data or cause any change in the
evolving data string. Feedback was after the conclusion of the set.
The participants
. No participants professed any psychic ability. Some were from the University
community, and some were interested Lewes residents. The male/female mix was
approximately even.
. Each participant was allowed to study algae swimming in a haemocytometer prior
to their set, to help with visualization. This was done in a separate room. None
could have predicted changes in algal velocities based on this preview. Only a few
alga are visible in the microscope field, and their behavior there will be
dominated by the effect of the high intensity white light of the microscope.
51
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4

Approved For Release 2000/08/08 : CfA-RDP96-00789R002200030001-4
Figure A-1
25000 HZ
IBVARNS
00.00 WZ
LN
12MAY 09:18
85 :5001
VI
AS
SU
64
100
CURSOR
A:
Live Dunaliella
£225000 HZ
35.2E 3 VOLTI R.
LN
AZMAY 14119
3002
AS
EX
11
CURSORD F **
Vibration on pedestal
52
Approved For Release 2000/08/08 : CIA-RDP96-00789R002200030001-4
..