SAIC EXPERIMENT DATABASE

• Document Type: CREST
• Collection: STARGATE
• Document Number (FOIA) /ESDN (CREST): CIA-RDP96-00791R000200190047-6
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 7
• Document Creation Date: November 4, 2016
• Document Release Date: May 19, 2000
• Sequence Number: 47
• Publication Date: July 25, 1995
• Content Type: MEMO

SAIC - [phone] -Created: Thursday, July 27,19951 x:20 - Page 1 of 6
---'r---------
---- apl3~`bvistl-Fzsr~t~as~26(~3169~96~-C-I,~-RBP96rAA~9~A(~0~-0Aa9AA4Z-6------------------
The Cognitive Sciences Laboratory
330 Cowper Street, Suite 200, Palo Alto, CA 94301
~, Voice: 415.327.2007 -Fax: [phone]
An Employee-Owned Company a-mail: [e-mail]
Memorandum
SG11
SG11
Date: 25 July 1995
To:
From: Edwin C. May, Ph.D.
Re: SAIC Experiment Database
I have been asked to identify a complete list of experiments that were conducted at Science
Applications International Corporation (SAIC) from 4 February 1991 through 30
September 1994. This memorandum lists the experiments, describes their justification, and
tabulates their overall results. In addition, complete references to the original reports will
be provided.
Background
It is instructive to understand the context in which SAIC was contracted to conduct research
into anomalous mental phenomena. Following the example of the previous Cognitive
Sciences Program at SRI International, we were asked to create a Scientific Oversight
Committee (SOC) that would be actively involved in monitoring Che research. The SOC's
charter was to:
? Review and approve a detailed written protocol, including all statistical analyses, for each
proposed experiment.
? Exercise unannounced "drop-in"privileges to observe experiments in progress.
? Review, in written form, the final reports and the conclusion of each study.
? Provide guidelines for future research.
To comply with aCongressional-directed requirement, the first four months of the contract
were used to prepare a S-year integrated research plan. Following that, technical and
human-use protocols had to be officially approved before any experimentation could begin.
Thus, data collection began in January 1992. Experiments continued, albeit with gaps to
prepare and seek approval for more protocols, through March, 1994.
All approved technical and human-use (i. e., Institutional Review Board) protocol
documents, final reports, and SOC comments are on file with the sponsor and are available
for review.
Approved For Release 2003/09/16 :CIA-RDP96-007918000200190047-6 Page 1
 SAIC - {415) 322-7960 -Created: Thursday, July 27,199514:20 - Page 2 of 6
- -----------------
_------------------~4pp~=ev~ecl-Fey-Release 2063tf?9118--et~t~fJP91;-o~0~7~9'IRa~o>)zOQ1-9Ifo~7-s
Experiment Database Summary
All experiments may be found in one of two documents or additionally in separate
publications:
(1) May, E. C., Luke, W. L. W., and Lantz, N. D. (3 February 1993).
Phenomenological Research and Analysis. Final Report: 6.2 and 6.3.
(2) May, E. C., Luke, W. L. W., and James, C. L. (29 September 1994).
Phenomenological Research and Analysis. Final Report.
Ten experiments were conducted by SAIC under government contract. These experiments
include all pilot and formal series, and they contain all the trials that were conducted (i.e.,
there is no file-drawer problem, by definition). Table 1 shows the summary information
for the 10 SAIC experiments.
Number
Ex riment
Reference
Trials
Effect Size
P-Value
1
~arpctTktende*+cies
1 (11-30)
200
0.1241 0.071
0.040
2
AC of Binary Targets
1 (31-3$)
300
0.12310.058
0.017
3
MEG Replication
1 (39-48)
12,OOOs
MCE
MCE
4
AC With Binary Coding
1 (49-56)
40
-0.0670.158
0.664
5
AC in Lucid Dreams (B-Line)
1 (58-60)
24
0.08810.204
0.333
6
AC in Lucid Dreams (Pilot)
1 (58-60)
21
0.36810.218
0.046
7
Rpmc~tP Observation
2 (2-3)
48
0.36110.144
0.006
S
ERD EEG Investigation
2 (3-11)
7,OOOs
MCE
MCE
9
ERD AC Behavior
2 (3-11)
70
0.30310.120
0.006
10
FntrolZy II
2 (11-12)
90
0.55010.105
9.1 x l 0~s
References are shown as Document Number (Beginning-Final Page). Those experirne nts
that are underlined may also be found elsewhere. Experiments 1 and 10 have been
published in the peer-reviewed journal, Journal of Parapsychology, Vol. 58. 285-302, and
experiment number 7 may be found in an Institute of Noetic Sciences publication. (These
additional documents have been made available to the review team.)
Experiments 3 and 8 were physiologically oriented and the number of trials shown is the
approximate number of stimuli.
The effect size and its associated p-value is computed across all conditions and across all
subjects and is provided as a conservative guideline.`
~` For this casual meta-analysis, I combined, post hoc, the results across receivers and conditions. With
some exceptions, it is not my policy to combine data across individuals. I believe this is incorrect if one
is studying the parameters of exceptional performance.
Table 1.
SAIC Experiment Database (1991-1994)
Page 2
Approved For Release 2003/09/16 :CIA-RDP96-007918000200190047-6
 SAIC - [phone] -Created: Thursday, July 27, 199514:20 - Page 3 of 6
--------=--------~ra~vedl=aTt~teas~~flfl3f@9146-~~R-R{~P96-89~9.1+~(IOa2~8A~9A(147-~------------------
Experiment Details
This section includes the raison d' titre and conclusions for each experiment listed in Table
1. The protocol details may be found in the technical proposals to the SOC and in the
formal descriptions either in the primary two reports or their associated publications.
y. Target and Sender Dependencies
In anomalous cognition (AC)'" experiments, it is important to know if the quality of the data
depend upon whether a second individual (i.e., a sender) is focusing attention on the target
material. Secondly, it is claimed in the Ganzfeld literature that dynamic targets (i.e., video
clips) are "better" targets than are static ones (i.e., photographs).
We found that a sender is not a necessary condition far goad AC and, contrary to the
Ganzfeld results, we found that static targets produced stronger AC than did the dynamic
ones. In addition, we found a significant correlation with the quality of AC and the
gradient of Shannon entropy for the static targets. The effect size quoted in Table 1 is
computed across sender and target conditions.
These results led to the publication of Managing the Target-Pool Bandwidth: Possible
Noise Reduction for Anomalous Cognition Experiments, Journal of Parapsychology , Val.
58, 303-313 and a revised version of Shannon Entropy as an Intrinsic Target Property:
Toward a Reductionist Model of Anomalous Cognition, in press, Journal of
Parapsychology and Appendix D in document number 2.
2. Enhancing Detection of AC of Binary Targets
There are a few examples in the literature that suggest that information theory and error
correction may increase the de tection of AC. -This experiment was designed to enhance
binary hitting rate by applying sequential analysis to a sequence of single AC decisions at
the binary bit level. Two novices receivers and one experienced one contributed 100
overall binary trials.
The experienced receiver (i.e., 531) who had produced significant results in SRI 's random
number generator experiment in 1979, produced a significant result in this experiment.
Although this individual 's result was significant at the 10 "~ level, the effi ciency was low.
That is, 213 individual binary calls were required to eventually arrive at a single sequential
analysis decision. The effect size and p-value shown in Table 1 are computed for all three
receivers; however, the significant combined result is exclusively because of receiver 531.
3. Magnetoencephalograph Replication
At the close of the SRI International Cognitive Sciences Program, May et al. reported
significant neurornagnetic responses to sensorially and physically isolated image stimuli.
The stimuli lasted for 100 rns and consisted of spatially low frequency sinusoidal grating.
This experiment was an extensive replication attempt with substantially more data.
Although the preliminary results were promising , there was no statistical evidence of an
effect. That is the effort and control conditions were statistically inseparable and care was
A definition of terms may be found in the Glossary at the end of this memorandum.
Approved For Release 2003/09/16 :CIA-RDP96-007918000200190047-6 Page 3
 SAIC - [phone] -Created: Thursday, July 27,199514:21 -Page 4 of 6
----------------
___________________.AppKgvecl-Fey=Release 2f}6-3f09t1.9-:-et~fZD'P'SC~=007J'I~b~~00190047-6
exercised so that the experiment possessed sufficient statistical power that an effect of the
magnitude of the SRI study would have been easily seen.
After the fact, a major fatal flaw was discovered in this and in the earlier experiment. The
SOC, the neuroscience staff at Los Alamos National Laboratory, and the staff of the
Cognitive Sciences Laboratory all failed to recognize the error. The dependent variable was
the instantaneous phase shift of the dominant alpha rhythm concomitant with the stimulus.
The Crarnrner-Rao condition, which sets lower limits on the variance in such measurements
prevented us from seeing an effect, even if one existed.
This experiment, provided significant insight in the design of the next attempt to identify a
central nervous system correlate to AC.
4. Enhancing Detection of AC with Binary Coding
The reason for this experiment was to develop a message sending algorithm using AC.
Sets of five dichotomies were identified in our standard target pool (e. g., mountains versus
flat) to act as "binary" bits in a standard two-by-five, error correcting block code.
This experiment produced no statistical evidence of AC, although the statistical power was
quite low. Each of five receivers contributed only eight trials. Receivers 'individual effect
sizes ranged from -0.559 to 0.224.
The primary purpose was to use the error correction to "send a message ." The effect size
shown in Table 1 is computed across all five receivers.
5. AC in Lucid Dreams (Baseline)
In a pilot series to determine if AC can be enhance din the altered state known as a lucid
dream (i. e., a dream in which the dreamer becomes aware that she or he is dreaming an can
signal the waking world of this awareness). Three of the receivers were Lucidity Institute
personnel had not been previously involved in AC experiments.
Each of these receivers contributed eight AC trials. One produced solid evidence for AC
(i. e., ES = 0.265). The effect sizes from the remain ing two receivers cannot be obtained
without consulting the raw analysis data; thus, the value in Table 1 assumes an effect size
of 0.0 for each of the these receivers.
6. AC in Lucid Dreams (Pilot)
Twenty one trials were conduced in this pilot study. The conditions were not ideal in that
each dreamer was allowed to take the intended target home for the night. Each target was
doubly sealed in opaque envelopes and covert threads were attached to indicate any
potential tampering. This breach of supervision was tolerated because of the additional
benefit of the receivers being cornfartable in their own beds. The effect size shown in
Table 1 was computed from the rank-order analysis across 21 trials.
7. Autonomic Detection of Remote Observation
Many of the experiment conducted in the Former Soviet Union (FSU) suggest that
biological systems make excellent targets for anomalous perturbation (AP). This particular
experiment involves an isolated sender a ttempting to physiologically arouse a sensorially
and isolated receiver. The state of arousal is measured by electrodermal activity.
Page 4
Approved For Release 2003/09/16 :CIA-RDP96-007918000200190047-6
 SAIC - [phone] -Created: Thursday, July 27, 199514:21- Page 5 of 6
---?--------------~4~~rov~tFarR~le~~e~003ttJ91`tfi^et~ - t~o~~t~o?rsva~~~------------------
This experiment successfully replicated the early work and lends support for the claims in
the FSU.
8. Central Nervous System Response to AC Signals
Generally, there is no known human activity that does not interrupt the production of alpha
rhythm. This includes responding to external stimuli, internal mental activity, or
intentionally moving a body part. We assumed, therefore, that AC would also interrupt
alpha. Improving on earlier experiments, we collect EEG data and standard behavioral AC
data that could be analyzed in the usual way. In addition, event related desynchronizations
(ERD) resulting from direct stimulation by the AC target during the feedback phase of a
trial were used as templates for match filters. These filters are the most sensitive way to
search for the assumed similar ERD in the EEG record that was collected during the AC
portion of the trial.
We have only analyzed one EEG lead and found no evidence of an AC-induced ERD.
When time and resources become available, we will finish the analysis for the remaining
leads.
Regardless of the outcome of this analysis, it will be a useful result; it is almost as
interesting if there is no ERD while there is significant evidence for AC.
9. ERD AC Behavior
Seventy trials were blind judged by the rank-order technique. Two of the receiver
produced independently significant results, but the effect size shown in Table 1 is the
combined result for all three receivers.
10. The Gradient of Shannon's Entropy
This experiment is a replication of the entropy and target-type portion of Experiment 1.
The protocol was substantially improved by nar rowing the target-pool bandwidth and by
monitoring the sessions. Conceptually, the entropy result appeared to replicate; that is, the
larger the Shannon entropic gradient in the target, the better the AC. The significantly
stronger effect size in the dynamic targets reflect the improved protocol. Because the
significant correlation primarily came from the target difference, it is premature to assure
that the gradient of the entropy is the deciding factor. Added to the manuscript for
publication, were the successful Monte Carlo results that favorably support this hypothesis,
however. The effect size shown in Table 1 was computed across target type and receiver.
Comments
By definition, this data set does not contain a so-called file drawer problem. That is, all the
data from the SAIC database were included for examination . For the six experiments that
used a more traditional AC protocol, regardless of conditions, the weighted (by number of
trials) effect size is 0.23010.047 in 455 trials (p = 5.7x10-).`
All these experiments had been approved by the SOC and IRB.
Page 5
Approved For Release 2003/09/16 :CIA-RDP96-007918000200190047-6
 SAIC - [phone]- Created: Thursday, July 27,199514:21 -Page 6 of 6
-----------------
________.~__--------~4pp~=eyed-Fe~Refeasg 2f}D-3t09t't1~~Ct~~D~P5~Q07S11~a~~00190047-6
GLOSSARY
Not all the terms defined below are germane to this report, but they are included here for
completeness. In a typical anomalous mental phenomena (AMP) task, we define:
? Anomalous Cogni ion (AC)-A form of information transfer in which all know sensorial
stimuli are absent. That is some individuals are able to gain access to information by an
as yet unknown process. This is also known as RP,_mote Yi~~ing