Direct mental interactions with living systems (DMILS)

Full text

Direct Mental Interaction in Living Systems (DMILS) –
Assessing Healing in the Physiological Lab
A Systematic Literature Review
Stefan Schmidt
Institute for Environmental Medicine
and Hospital Epidemiology (IUK)
University Hospital Freiburg
Hugstetter Str. 55
D-79106 Freiburg
Germany
Tel. ++49-761-270-5494
Fax ++49-761-270-7224
e-mail sschmidt@ukl.uni-freiburg.de
Introduction
Direct Mental Interaction in Living Systems (DMILS) represents an experimental
paradigm in parapsychology that was developed in the mid seventies. These experiments
investigate whether there is an interaction between two spatially separated persons. This
interaction is operationalized as covariation between the intentions of one person and the
physiological reactions of the other person.
First, I will outline the procedure of a typical DMILS experiment and will then relate this
experiment to Healing Energy Research (HER), in order to find a set of appropriate questions
on DMILS research from an HER perspective. On the basis of these questions, I will describe
a systematic literature review of DMILS which can provide answers. Furthermore, this
systematic literature review will also describe the latest knowledge of DMILS research in
general.
DMILS – an Experimental Paradigm
DMILS is a laboratory experiment. Usually, two participants and an experimenter are
needed to conduct a DMILS session although the experimenter can also combine the two
roles of agent and experimenter. The two participants have different tasks and are often
referred to as “sender” or “agent” and “receiver”. This description is helpful to describe the
experiment although it is somewhat misleading as it entails certain implications of
information transfer which are not necessary for the experiment to be successful. Therefore I
will use them in a preliminary way. Possible misunderstandings will be addressed in the next
paragraph.
The two participants are physically separated from each other to preclude any conventional
means of communication. At least one of the participants, i.e. the receiver, has to be housed in
an acoustically and ideally also electromagnetically shielded chamber. Several physiological
measures are recorded from the receiver such as EDA, respiration, heart rate, blood pressure
or EKG. During the experimental session the receiver’s task is simply to maintain a wake but
relaxed and open state. At the same time the agent is placed in front of a monitor. The monitor
displays an extract of the physiological recording taken from the receiver to keep the agent
informed of the level of arousal or activation of his counterpart (feedback condition). The

agent’s task is to activate or to calm the receiver from the distance. Therefore the session time
is divided into several epochs usually lasting 30-60 sec. During each of theses epochs one of
two conditions (e.g. calm or activate) is displayed as a word to the agent requesting her or him
to perform in any way that might lead to the appropriate reaction in the receiver. The
sequence of the epochs is randomized and balanced. Often the epochs are interspersed with
rest intervals. The agent is not restricted in the way she or he is able to activate or calm the
receiver, various strategies including imagery or physical methods are allowed (see Figure 1).
For evaluation, the physiological data is summed up according to the two different
conditions, e. g. 10 1-minute calm epochs are compared with 10 1-minute activate epochs.
Under the null hypothesis the same amount of activity or arousal is expected in both
conditions. If the physiological activity differs significantly in the intended direction (more
activation during activate epochs) one can speak of a so called DMILS effect.
ACTIVATE
EDA
Receiver Agent
Figure 1: A typical DMILS situation. EDA data of the receiver are fed back to the agent. The agent tries to
change the physiological activity of the receiver by means of intention .
DMILS from a Healing Perspective
From a parapsychological perspective the purported DMILS effect can be interpreted in
many ways. This is due to the fact that almost nothing is known about the way in which
parapsychological or psi effects are mediated. There are several competing theories (e.g. see
Lucadou 1995, May et al 1995, Millar 1978 or Stokes 1997, for an overview) but they are all
more or less rough models that cannot provide detailed information about the functioning of
these effects. There is also no dominant theory or model that is accepted by a majority of the
researchers in this field.
The basic mechanism of these effects is as yet unknown; we also do not know which
constituents are needed and how this system, consisting at least of two persons and a certain
laboratory equipment, is organized.
Therefore many different interpretations are possible. The most common one is that the
sender as the active part is influencing the receiver’s physiology. But this is only the most

obvious interpretation, because this is how the experiment is described, and because the
receiver is called receiver and not, for instance, “remote viewer” or “observer”. It might also
be true that the receiver acts subconsciously as a remote viewer, who, perceiving the
conditions from the sender's monitor, influences his or her own physiology accordingly. With
that interpretation the sender is not needed. It might also be that the experimenter foresees the
best moment to start the experiment with the natural fluctuation of more and less activation in
the receiver just falling in the corresponding epochs. This is why the abbreviation DMILS
stands for Distant Interaction in Living Systems. In some way there might be an
unconventional interaction within the whole experimental system or between parts of it, that
can be seen in the recorded data.
From a healing perspective DMILS can be seen as a research paradigm of the very
mechanism underlying a distant healing process. The DMILS setup has some parallels to a
distant healing situation. There is a sender or a healer who tries to change the physical or
physiological state of a healee or receiver from a distance by intentional means. This is why
the DMILS paradigm is of major importance for any healing research that takes the possibility
of distant healing into account. The DMILS setup can provide us with an answer to the very
basic question as to whether it is in principle possible for an organism to react physiologically
in accordance with a remote intention. Or in other words, is there a mechanism independent
from any healing context that is responsible for effects of distant intention and that might also
be used in a distant healing situation. If one assumes that DMILS experiments and distant
healing rely on the same effect, then the DMILS laboratory is an excellent setting to study this
effect under controlled conditions. These investigations can address three areas, proof-,
process- and model-oriented research.
Taking this analogy into account, healing research can pose the following questions to the
laboratory:
• Is there empirical evidence for an effect of distant intention that might also apply to
distant healing (proof oriented)?
• Are there conditions that improve or diminish the size of the effect (process oriented)?
• Are there further empirical results that help us understand distant healing in a
theoretical sense or that help us generate an appropriate model of distant healing
(model oriented)?
Systematic Literature Review – Methods
This literature review is the first step of a meta-analysis in preparation. The meta-analysis
will include all DMILS studies and Remote Staring studies (a related paradigm, see below)
with EDA as dependent variable, as EDA is by far the most included variable in DMILS
research. Some studies report several different psychophysiological variables including EDA
as dependent variable; but I know of only one study on human DMILS that measured blood
pressure rather than EDA, but to my knowledge this study was never published .
For this review all studies using a non-human target systems such as gerbils (Braud 1979),
fish (Braud et al 1979), blood conserves (Braud 1990), or enzymes (Bunnell 1999) have been
excluded. Furthermore, all studies having behavioral measurements as dependent variables
such as a concentration task (Brady & Morris 1997, Braud et al 1995), control of
muscletremor (Braud et al 1989) or ideomotoric reactions (Braud & Jackson 1982) have also
been excluded. Moreover, the data set does not include studies with prerecorded targets. In
these studies the psychophysiological data are already recorded when a participant tries to
interact with them (e.g. Radin et al 1998). These studies can be considered as an own category
that has been recently addressed by William Braud (2000).

The studies have been retrieved as follows. We have included all the publications
mentioned in a meta-analysis on EDA-DMILS studies published by Marilyn Schlitz and
William Braud in 1997. We have conducted an additional literature search for EDA-DMILS
studies from 1996 up to 2000 in the "Journal of Parapsychology", "The Journal of the
American Society for Psychical Research", "The Journal of the Society for Psychical
Research", "The European Journal of Parapsychology", "The Journal for Scientific
Exploration" and the "Proceedings of the Annual Convention of the Parapsychological
Association". Furthermore, we have searched for so far unpublished material for the meta-
analysis, but the 7 studies we found will not be included in this literature review. Two EDA-
DMILS studies by Braud & Schlitz have never been published but will be included as their
overall results are presented in various overviews (e.g. Braud & Schlitz 1989a, Braud &
Schlitz 1991, Schlitz & Braud 1997).
Results
Data set
In summary, 22 studies published as single papers or in overviews from 1978-1999 were
found (see Table 1). They form a data set of 24 single experiments with an overall number of
636 sessions, conducted in 4 different laboratories (San Antonio, Edinburgh, Las Vegas and
Freiburg). As the various sessions in different experiments were of different length one might
also count the number of epoch pairs (e.g. calm-control, activate-control or calm-activate) to
assess the size of the entire data set. There were 6842 such pairs. The number of sessions
within these studies varied from 10 to 80, the length of an epoch pair from 40-120 sec.
(without interspersed rest periods) and the number of epoch pairs per session from 5 to 16.
# authors year
target-
system N1 additional hypotheses signi-
ficance
effect
size (r)
1 Braud 1978 EDA 10 p=0.02 2t 0.72
2 Braud 1978 EDA 10 p=0.01 1t 0.70
3 Braud, Davis &
Wood
1979 EDA (SRR) 10 gifted sender p=0.04 1t 0.56
4 Braud 1979 EDA (SRR) 10 n.s. 1t - 0.25
5a Braud & Schlitz 1983 EDA (SRR) 16 participants with labile EDA p=0.014 1t 0.53
5b 16 participants with stabile EDA n.s. 1t - 0.02
6 Schlitz & Braud 1985 EDA (SRR) 15 senders were healing
practitioners
n.s. 2t 0.16
7 Braud, Schlitz,
Collins & Klitch
1985 EDA (SRR) 24 feedback vs. no feedback p=0.04 2t 0.35
8 Braud, Schlitz,
Collins & Klitch
1985 EDA (SRR) 32 Blocking vs. no blocking n.s. 0.20
9 Braud, Schlitz,
Collins & Klitch
1985 EDA (SRR), HR,
skin temperature,
muscle tension,
respiration
30 Specificity vs. generality n.s. 0.08
10 Braud & Schlitz 1989 EDA (SRR) 40 Model of Intuitive Data Sorting
(IDS) pilot study
n.s. 0.03
11a Braud & Schlitz 1989 EDA (SRR) 32 IDS disfavoring condition p=0.019 1t 0.40
11b 32 IDS favoring condition n.s. 1t - 0.09
12 Braud & Schlitz n.p.2 EDA (SRR) 30 direction of influence n.s. 0.08
1 Number of sessions
2 not published

13 Braud & Schlitz n.p. EDA (SRR) 16 modulation of magnitude n.s. 0.32
14 Delanoy & Sah 1994 EDA (SRR) 32 p=0.043 1t 0.31
15 Radin, Taylor &
Braud
1995 EDA(SRR) 16 n.s. 1t 0.27
16 Wezelman,
Radin, Rebman &
Stevens
1996 EDA(SCL), HR,
BVP
11 healing rituals n.s. 0.083
17 Rebman, Radin,
Hapke &
Gaughan
1996 EDA(SCL), HR,
BVP
16 Healing rituals p=0.0004 1t 0.373
18 Delanoy & Morris 1998 EDA (SCL) 36 Experimenter trainees n.s.1t 0.16
19 Schneider, Binder
& Walach
2000 EDA (SCL) 40 Personal vs. Neutral
experimenter interaction
n.s. 2t 0.17
20 Watt, Ravenscroft
& Mc Dermott
1999 EDA(SCL) 32 Blocking vs. co-operating n.s. 1t -0.01
21 Watt, Ravenscroft
& Mc Dermott
1999 EDA(SCL) 50 Blocking vs. co-operating p=0.038 0.25
22 Delanoy, Morris,
Brady & Roe
1999 EDA(SCL) 80 Biological related pairs vs. close
related pairs
n.s. 1t -0.04
table 1: overview of all DMILS experiments by authors year of publication, numbers of sessions performed,
additional hypotheses tested, level and direction of significance testing and according effect size (r).
Abbreviation of physiological variables are: EDA = elctrodermal activity, SRR = skin resistance
response, SCL =skin conductance level, HR = heart rate, BVP = blood volume pulse.
Significance
Nine out of 24 experiments yielded significant results according to their hypotheses. This
means that 37.5 % of the experiments had significant results compared to 5 % expected by
chance.
Effect Sizes for EDA
Similar to the procedure used by Braud and Schlitz (1991, 1997) we calculated an effect
size for the results of the EDA data of each experiment. This effect size is Rosenthal's r
(Rosenthal 1991). It is free of any distribution assumptions and can be used with various test-
statistics, including non-parametric ones. In this case two different test scores were employed,
t-scores stemming mostly from single-mean t-tests and z-scores stemming from the non-
parametric Wilcoxon test. The formulas were r=z/√N and r=√t2/(t2+df). For N we took the
number of sessions per experiment. For the studies performed by Wezelman et al (1996) and
Rebman et al. (1996) we took the number of epoch pairs as N, as this was their unit for
statistical evaluation. One problem with calculating Rosenthal's r in that way is that one
cannot calculate a confidence interval for the effect size. Although Rosenthal provided a
formula for the calculation of r's variance (Rosenthal 1994), this formula presupposes normal
distribution and is only recommended for data stemming from parametric test-statistics.
The effect size of the 24 experiments ranged from r= –0.25 to r=0.72, with negative results
indicating an outcome opposite to the expected direction. The mean is at r = 0.22 with SD =
0.25. The 95% CI of the mean ranges from 0.11 to 0.32 and is significantly different from
zero. Figure 2 shows the effect sizes for the single experiments.
3 Instead of the number of sessions the number of epoch pairs was taken as N to calculate r as the original test
statistic was based on that number

-1
-0,5
0
0,5
1
1,5
1
2
3
4
5a
5b
6
7
8
9
10
11a
11b
12
13
14
15
16
17
18
19
20
21
22
mean
effect size (r)
Figure 2 Graphical display of effect size (r) of all 24 single experiments including overall unweighted mean
effect size. Error bars represent the 95% CI.
As can be seen from the graph the largest effect sizes were obtained within the first five
experiments. They are all relatively small with a N of 10 (# 1,2,3 and 4) or 16 (#5a, 5b)
sessions. Therefore, it might be interesting to calculate an effect size weighted by the number
of sessions. As expected, the effect size drops down to r = 0.16, a similar result of r = 0.17 is
obtained if the studies are weighted by the number of epoch pairs.
The large range of effect sizes might indicate that not all the data are estimates of the same
effect. This can be assessed by applying a statistical model to this sample and looking for
possible moderators of the effect size. We are planing to do this in the meta-analysis
mentioned above. But for a preliminary investigation of the characteristics of that sample we
have applied a test of homogeneity proposed by Rosenthal (1991, p. 73-74). It yielded a
χ2=29.76 with df=23. The corresponding p-value is p=0.15 proving the sample to be
homogenous on this test.
Dependent Variables other than EDA
There are only three studies in this data set that record physiological measures other than
EDA. These are heart rate, peripheral skin temperature, frontalis muscle tension and breathing
rate in experiment 9 and heart rate and blood volume pulse in experiments 16 and 17. In
experiment 9 the authors found some differential results for the heart rate, but all variables
showed no overall effect. In experiment 16 the results for heart rate and blood volume pulse
were not significant. The same was true for the blood volume pulse in experiment 17, while
the heart rate was significant lower (p=0.01) in the experimental condition.
In a recent pilot study (Schmidt et al 2000) we explored the relationship between a DMILS
effect in EDA and respiration. The data showed that irregularities in respiration (sudden
changes in the rhythm) accounted for substantial parts of the EDA-DMILS effect found in the
study.
Different Laboratories
The DMILS studies were all conducted in four laboratories. The Mind Science Foundation
in San Antonio accounts for 50.8 % of all sessions, Edinburgh for 33 %, Freiburg for 11.9 %

and Las Vegas for 4.2 %. While research activities have stopped in Texas and Nevada the
European laboratories still conduct DMILS research. Approximately 400 unpublisched
sessions have recently been conducted in those two laboratories. This unpublished data set has
more than 60 % of the size of the data set presented in this review.
Comparing effect sizes by laboratory shows that San Antonio has a mean of r=0.25 while
the other three laboratories are within the range of r=0.15 to r=0.17.
Methodological Quality of the studies
EDA
As EDA is the only dependent variable in most of the DMILS experiments its
measurement by state-of-the-art techniques is crucial. To assess whether the published
DMILS studies meet the proposed standard procedures published by a committee of the
Society of Psychophysiological Research in 1981 (Fowles et al 1981), we evaluated all
publications according to these guidelines (Schmidt & Walach 2000). This analysis revealed
that none of the DMILS studies met theses standards. DMILS researchers have mostly
neglected the knowledge provided in psychophysiological literature. For instance, none of the
measurements in the experiments reviewed used an appropriate electrode gel that is needed
for the recording of EDA signals. Most of the studies do not report their methods in detail -
which made it more difficult to judge whether inappropriate measurements had invalidated
the data. For the discussion of possible artifacts due to these methodological shortcomings the
next topic is of special importance.
Intraindividual Design
For all discussions of methodological shortcomings the DMILS design holds one big
advantage. DMILS always deals with intraindividual differences. Every test statistic
calculated on DMILS data compares experimental and control epochs of the same participant.
As these epochs are presented in a changing sequence, this means that any shortcoming holds
for both the experimental epoch as well as for the according control or comparison epoch. By
taking the differences between these epochs all eventual artifacts will be filtered out. This
means methodological inappropriateness can always result in not finding an effect that might
be there, but it is hard to imagine that a weak methodology will result in finding a DMILS
effect that would not be found with a better one.
This argument of the advantages of the intraindividual comparison relies strongly on one
fact, the balanced and randomized sequence of the epochs.
Balanced and randomized sequence
The sequence of the usually 10 epoch pairs have to fulfill three criteria to guarantee an
artifact free comparison of the epochs. (i) There must be the same number of epochs of each
kind and, of course, all epochs have to have the same length. (ii) The sequence of the
experimental and control epochs must be randomized. (iii) The sequence of the epochs must
be balanced in a way that does not lead to any effects due to the natural variation of the
recorded variable. While (i) is trivial and (ii) is obvious, (iii) is crucial. Consider the following
example of a receiver becoming more and more tired and therefore less activated throughout a
20 minute recording period. If the 10 activation and 10 calm epochs are then presented in a
sequence as e.g. AACAAACAACACCACCCACC one will find an DMILS effect even if
nothing extraordinary happened. This is because the sequence is unbalanced and contains
more activation periods at the beginning of the session than at the end. The same is true of any
repeating pattern within the data that correlates by change with a rhythmic pattern of the

receiver's physiology. Therefore a balanced and randomized sequence of the epochs is crucial.
Wiseman & Smith (1994, p. 475) found out that even a sequence that looks pretty much
balanced at first glance may lead to spurious findings. Therefore the data of this review were
checked to see if the description of the procedures guaranteed a balanced design.
All experiments took care to ensure a basically balanced design except for the first four.
We excluded them from the analysis and calculated an unweighted and a weighted (by
session) mean effect size on the remaining 20 studies. The unweighted effect size dropped to
r=0.17 with a smaller standard deviation of SD=0.18 and 95% CI ranging from r=0.09 to
r=0.26. The weighted effect size is r=0.15, more or less the same as before.
Selection of participants
Most study subjects were unselected participants. Due to the intraindividual design there is
no need to randomize participants into different groups or to match them. Moreover, as the
focus of interest in the DMILS experiment is to explore whether there is in principle a
hitherto unexplained interaction between two separated persons, DMILS experiments do not
necessarily rely on a representative sample. It might even be possible that only some gifted
persons are able to perform above chance in DMILS experiments. Under this premise a
representative sample may not find an existing effect.
On the other hand representative samples are needed for any research asking process-
oriented questions on interindividual differences that could be generalized to the population.
This might be an important point for future research, but from the perspective of today's
situation there is no need for representative samples.
Statistics
In a recent analysis we (Schmidt et al 2000) compared different statistical evaluation
strategies in DMILS experiments. Because of the complicated structure of the data this is not
a trivial problem. Any strategy must deal with three problems: (i) parameterization and
accumulation of the epochs, (ii) intraindividual comparison of experimental and control
epochs, and (iii) accumulation of all sessions' data to a score that allows for statistical testing
of the whole experiment.
Our analysis revealed that the predominant statistical method used in DMILS research so
far, the so called percent influence score (PIS, Braud & Schlitz 1991, p. 5) has a severe lack
of statistical power for some kinds of data. This presumably means that some of the DMILS
effects in the data have not been found due to inappropriate methods.
Process-oriented results
A variety of different process-oriented hypotheses were tested in different DMILS
experiments. Table 1 presents also an overview of all process oriented hypotheses.
There is no agreed upon process-oriented knowledge for the DMILS setup in the sense that
any process oriented hypothesis was replicated at least once successfully. There are several
reasons for this. The problem is that for some so far unknown reason some DMILS
experiments yield significant results and others do not. The logical consequence of that fact is
the search for so far unknown moderators which will result in testing different process-
oriented hypotheses. Thus, quite a lot of different hypotheses have been tested, but only very
few replications of positive findings have been done. Furthermore, the absence of a
hypothesized DMILS effect in general makes the situation even more difficult. For in this
case one cannot draw any conclusion on the process-oriented hypothesis to be tested.

A complete mapping of all process-oriented findings would be rather extensive. A good
overview can be found in Schlitz & Braud (1997). So I will focus only on the results that are
of relevance from a healing perspective.
In experiment 3, a so called gifted person, Matthew Manning, acted as a sender in ten
sessions. The study yielded a significant result (p=0.04, 2-tailed) with a large effect size of
r=0.56. The study had more of an exploratory character. There was no comparison between
"gifted" and "non-gifted" senders, that means the effect does not depend necessarily on the
special sender. Furthermore as stated in the section on "Methodological Quality of the
Studies" the randomization in that experiment was not balanced; thus, the effect could also be
due to inappropriate randomization.
In experiment 6, Schlitz and Braud employed three healing practitioners with experience in
Reiki as senders. Each of them acted as sender in 5 sessions. The study did not reveal any
significant result.
Experiments 16 and 17 addressed explicitly the analogy between DMILS and some aspects
of traditional healing. The authors state that "...the DMILS experiment itself is actually a
modern version of the same principles that underlie certain traditional magical healing rituals
that fall into the general category of magic." (Wezelman et al 1996, p. 2). Accordingly they
prepared their setup similar to a traditional (voodoo) healing situation. The senders were
called healers, their chamber was a darkened room illuminated by a golden candle. Ritual
objects such as a doll with personal belongings of the receiver were placed on black cloth
within the cabin. The results of these experiments are hard to interpret as the authors did not
define precisely what should be regarded as relevant outcome of their studies. They present
various p-values by different statistical evaluation methods. For this review we chose to take
the cumulative difference between the control and the experimental condition at the end of the
60 sec. epochs. This was not significant for EDA in the 54 pairs of epochs compared in
experiment 16. In experiment 17, the same statistic for 78 epoch pairs yielded a highly
significant z-score of z=3.3 (p=0.0004, 1-tailed). Nevertheless, the data allow only for limited
conclusion regarding the healing context as there was no comparison condition.
Results on Model Testing
Another line of DMILS research might be to develop a theoretical model that explains the
supposed DMILS effect. Then single hypotheses can be derived from the model and tested
empirically. One problem in parapsychological research is, as stated above, that there is no
predominant or agreed-upon model that could explain findings such as DMILS effects.
There is only one study that tests a theoretical assumption known by the name of "Intuitive
Data Sorting" (IDS). According to this model, a predecessor of the DAT model (see May et al
1995, May et al 1996), "the influencer or experimenter psychically, yet unconsciously, scans
the future electrodermal activity stream of the subjects and begins an experimental session at
a time that maximizes the degree of fit between the ongoing electrodermal activity and the
prescribed schedule of influence and control epochs." (Braud & Schlitz 1989b, p. 290). Braud
& Schlitz therefore varied the degrees of freedom for the sender to start the epochs in the
experiments 11a (low) and 11b (high). They expected a higher DMILS effect in experiment
11b under the assumption of the IDS model. The opposite was true, experiment 11a scored
significantly, 11b didn't.
Remote Staring – a DMILS Related Paradigm
In parapsychological research there is an experimental paradigm that is very similar to the
DMILS setup. It was also promoted by William Braud and colleagues (Braud & Schlitz 1991,

Braud et al 1993a, Braud et al 1993b) and is normally referred to as Remote Staring. It was
designed to investigate whether one person can detect another person staring at him or her.
In some experiments a person stares at participants through a one way mirror. Participants
are then asked whether they perceived a gaze or not (Baker 2000, Colwell et al 2000,
Schwartz & Russek 1999). But the setup by Braud and colleagues used a different way to
investigate whether a participant knows if somebody is staring at him or her. Instead of
assessing their conscious knowledge by asking them verbally, they assessed the EDA and
compared autonomic arousal for staring and non-staring periods.
The procedures in these experiments are more or less the same as those used in the DMILS
experiments described above. The sender is now referred to as starer, while the receiver is
called staree. Instead of seeing the EDA feedback on a monitor the starer has the possibility to
see the staree by means of a closed circuit television system. The observation effect is
measured by intraindividual comparison of epochs in which the starer is gazing at the monitor
(experimental condition) with epochs in which he or she is looking elsewhere.
The similarity to the DMILS setup is obvious. Instead of having an intention to activate or
calm his or her counterpart the starer is observing the person in the other chamber. The
intentional condition is operationalized in a slightly different way, but nevertheless this
experiment also investigates the relationship between an intentional state in one person and
resulting physiological changes in another person. Therefore, the Remote Staring data set may
provide additional knowledge of the questions of interest from a healing perspective.
Study retrieval
We included all experiments that studied the detection of a remote observation with
physiological measurements as dependent variables. Retrieval strategies were the same as for
DMILS studies. We included all Remote Staring studies from the meta-analysis by Schlitz &
Braud (1997) and performed an additional search of the parapsychological literature from
1996 to 2000.
Results
We found 7 publications (including conference proceedings) describing 13 experiments
with N=300 sessions (see Table 2). Studies were conducted in 4 different laboratories, San
Antonio, Palo Alto, Sausalito and Hertfordshire.
# authors year
target
system N specialty Signi-
ficance
effect
size (r)
1a Braud, Shafer &
Andrews
1993a EDA (SRR) 16 p=0.018 2t 0.57
1b Braud, Shafer &
Andrews
1993a EDA (SRR) 16 connectedness training p=0.048 2t 0.49
2 Braud, Shafer &
Andrews
1993b EDA (SRR) 30 p=0.06 2t 0.34
3 Braud, Shafer &
Andrews
1993b EDA (SRR) 16 p=0.05 2t 0.47
4 Schlitz & LaBerge 1994 EDA(SCR) 24 p<0.036 1t 0.36
5 Schlitz & LaBerge 1994 EDA(SCR) 24 p<0.014 1t 0.44
6 Wiseman & Smith 1994 EDA 30 several observers for each
receiver
n.s. 1t 0.26
7 Wiseman, Smith,
Freedman, Wasserman
& Hurst
1995 EDA 22 gender pairing, distance n.s. 1t 0.14

8 Wiseman, Smith,
Freedman, Wasserman
& Hurst
1995 EDA 20 gender pairing n.s. 1t 0.20
9a Wiseman & Schlitz 1997 EDA (SRL) 16 experimener Schlitz p=0.04 2t 0.51
9b 1997 EDA (SRL) 16 experimenter Wiseman n.s. 2t 0.11
10a Wiseman & Schlitz 1999 EDA (SRL) 35 experimenter Schlitz p=0.05 2t 0.33
10b 1999 EDA (SRL) 35 experimenter Wiseman n.s. 2t 0.07
Table 2: Overview on all Remote Staring experiments, by authors, year of publication, numbers of sessions
performed, special hypotheses tested, level and direction of significance testing and according effect
size (r). Abbreviation of physiological variables are: EDA = electrodermal activity, SCR = skin
conductance response, SRR = skin resistance response, SRL = skin resistance level, SCL =skin
conductance level.
Of these 13 studies, seven were significant (p<0.05) with an eighth experiment being
marginally significant at p=0.06.
The effect sizes range from r=0.07 to r=0.57. The mean unweighted effect size is r=0.33
(SD=0.16), the 95% CI ranges from 0.255 to 0.405 and is significantly different from zero.
The weighted (by the number of session) effect size is r=0.31. Figure 3 shows the effect sizes
of the experiments.
-0,6
-0,4
-0,2
0
0,2
0,4
0,6
0,8
1
1a 1b 2 3 4 5 6 7 8 9a 9b 10a 10b mean
effect size (r)
Figure 3: Graphical display of effect size (r) of all 13 single experiments including overall unweighted mean
effect size. Error bars represent the 95% CI. Filled bars represent calming effects, empty bars arousal
effects, experiment 9b could not be classified.
An interesting question is whether the starees were aroused or calmed by the remote
observation. Of the eight experiments with significant results four show an effect of arousal
and four an effect of calming. They are all displayed as positive effect sizes in Figure 2, as
only effect sizes opposite to the hypothesized direction are displayed as negative. Most of the
significant tests applied in Remote Staring are two-tailed. To map this difference in activation
the bars in Figure 2 are either filled (calming) or empty (arousal), with experiment 9b being

unclear of the direction of the effect. But it is important to mention that this classification is
somewhat preliminary as the often incomplete description of the EDA methodology does not
allow a clear interpretation in terms of calming or activation.
It might be interesting to note that experiment 1a and 1b are two halves of one study with
the same setup. While in the first experiment participants (starees) were untrained,
participants in experiment 1b had taken part in a twenty hour connectedness training
"..engaging in intellectual and experiential exercises involving feelings of interconnectedness
with other people and dealing with their own psychological resistance to merging with others"
(Braud, Shafer & Andrews 1990, p. 20). The experiment yielded a significant arousal when
stared at for the untrained participants while the trained participants calmed down when
observed.
Experimenter Effects
In contrast to the DMILS data set the Remote Staring data set yielded a process-oriented
result that was successfully replicated. This is related to often heard rumors within
parapsychology that some researchers are highly successful while others never find Psi. After
Braud and colleagues had presented their first findings on the Remote Staring effect several
others tried to replicate them. While Marilyn Schlitz (together with Steven LaBerge) was
successful in two replications (# 4 & 5), Richard Wiseman and colleagues found only non-
significant results in three experiments (# 6 – 8).
To find out whether these results are related to the experimenter or to other differences in
the setup, they decided to conduct an experiment together with both using the same equipment
and procedures in the laboratory of the University of Hertfordshire where R. Wiseman is
employed (experiments # 9a & 9b). While the 16 participants introduced and observed by M.
Schlitz showed a significant overall result, the 16 participants cared for by R. Wiseman
performed at chance. The same was true for a larger replication (35 sessions each) that was
conducted in M. Schlitz's laboratory in Sausalito, Ca. In both cases the difference between the
experimenters was not large enough to reach a level of significance but both studies stated
clearly that with M. Schlitz as experimenter participants performed better than chance.
Discussion
Let us turn back to the questions posed from a healing perspective to the data of the
DMILS data set. The first question was whether there is any empirical evidence for an effect
of remote intention measured in terms of physiological changes. The answer is yes. The
published DMILS dat aset containing more than 600 sessions shows a small to medium mean
effect size that differs significantly from zero. This situation remains more or less the same
when the effect sizes are weighted by the size of the study. A preliminary test proved the data
set to be homogenous.
If there is such strong evidence for Psi the logical consequence is to investigate whether
the effect can be explained in any other way that does not rely on a parapsychological and
previous unexplained concepts. The assessment for the methodological quality of the studies
in this review showed that there are indeed some methodological insufficiencies, especially
for the applied EDA methods and statistics. But it is difficult to imagine how these
shortcomings might have led to artificial DMILS effects. From my perspective most of the
insufficiencies reported would result in not finding effects even if they were present, rather
than the other way round. The reason for this is the intraindividual comparison that is
calculated in every single DMILS experiment. It was pointed out above that here
randomization plays a vital role. By excluding all studies with inadequate randomized

sequences it could be demonstrated that the remaining studies still have a significant mean
effect size of r=0.17.
There have been various alternative hypotheses to explain the data such as experimenter
fraud, sensory cueing or cheating participants discussed in the literature (see e.g. Schlitz &
Braud 1997, Wiseman & Schlitz 1997). Although this seems very implausible it could only be
checked by a set of independent replications. As the group of parapsychologists doing DMILS
research is very small there is a lack of independent publications. There are only four
laboratories which have done DMILS research, and all their experience, (tacit) knowledge and
staff can be traced back to the beginnings in San Antonio where William Braud started the
DMILS setup. Braud visited Edinburgh in the early nineties and taught amongst others
Deborah Delanoy, Robert Morris and Dean Radin how to perform a DMILS experiment.
Delanoy and Morris then started the Freiburg laboratory, Radin the one in Las Vegas. So,
although there is a lack of independent replication, the fact that all four laboratories found
more or less the same effect sizes is a preliminary argument against alternative explanations
such as sensory cueing or cheating participants.
The review of the Remote Staring studies can be regarded as a kind of validation data set
for the DMILS data set. The design is almost the same only the way the intention is
operationalized is slightly different in a way one could describe as less overt. The mean effect
size for Remote Staring is noticeably higher than the one for DMILS but their confidence
intervals share an interval from r=0.255 to r=0.30. One cannot judge from these results
whether Remote Staring effects are different from DMILS effects, but the Remote Staring
results clearly confirm the findings reported for DMILS.
Regarding the second question on process-oriented research that might help to understand
a healing situation, the situation is disappointing. While the overall DMILS effect is quite
stable in various experiments, very little is known about the conditions necessary to increase
or decrease the effect.
The only process-oriented result that survived a first replication concerns the experimenter
effect and stems from Remote Staring studies. It looks as if the finding of a Remote Staring
effect is related to a certain person. One problem with this result is that Schlitz and Wiseman
had served several functions in their experiments. They both were principal investigators
(organizing the whole study), experimenters (interacting with the participants) and starers.
Therefore one cannot draw a firm conclusion from this finding, but it is one of the most
promising research lines within the actual DMILS and Remote Staring activities.
Another line of process-oriented research is just about to start. This will ask questions on
the relationship between effects in EDA and other psychophysiological systems. So far
nobody has addressed the question how EDA is more activated or calmed in a DMILS
session. I would call this line of research differential EDA research. Preliminary results such
as exploring the variance of the EDA (Watt et al 1999), comparing different EDA parameters,
or assessing the relationship between EDA and respiration (Schmidt et al in print) indicate
that this will help a lot to understand what happens within the physiology of the receiver. The
Freiburg laboratory plans additional EEG mapping during DMILS experiments (Plihal 2000)
and interesting results can be expected from that work.
As far as the third question on theoretical understanding and modeling of the DMILS
situation is concerned, not much can be said in terms of empirical evidence confirming certain
models. The only test applied so far failed. As stated before there are several good candidates
for a theory explaining psi effects. But beside the efforts made by Braud & Schlitz (1989) so
far nobody has applied them to the DMILS laboratory in deriving and testing hypotheses.

Conclusions
DMILS studies provide an interesting data set showing that effects of remote intention can
be found within electrodermal activity in laboratory situations. However, the field faces
limited resources and interest from mainstream science is also limited. Therefore, so far, only
four different laboratories have investigated DMILS while the number of researchers working
with this setup is no greater than 20. Currently only two laboratories are actively conducting
research. Therefore, there is a severe lack of independent replications. The strong proof-
oriented findings call for an extension of that research. By extension I mean (i) to extend from
EDA to other physiological and behavioral measures (ii) to extend the research into the
mainstream psychophysiological research by promoting the DMILS findings to researchers in
that field (iii) to extend theoretical understanding of the situation by applying and testing
different models in experimental situations and (iv) to extend the process oriented
understanding of the supposed effect by testing promising moderator variables and by
conducting differential research on EDA.
Acknowledgement
My work is generously supported by the Institut für Grenzgebiete der Psychologie und
Psychohygiene e.V. The DMILS meta-analysis is supported by a grant from the Institute of
Noetic Sciences. I am grateful to Harald Walach, Rainer Schneider and Deborah Lawrie for
helpful comments on the paper and for correcting my faulty English. I am also grateful to
Ulrike Biedermann for her help with the literature.
Reference List
Baker RA 2000 Can We Tell When Someone Is Staring at Us. Skeptical Inquirer 24: 34-40
Brady C, Morris RL 1997 Attention Focusing Facilitated Through Remote Mental Interaction: A Replication and
Exploration of Parameters. The Parapsychological Association 40th Annual Convention.
Proceedings of Presented Papers (pp. 73-91). The Parapsychological Association: Durham, NC
Braud WG 1979 Conformance Behavior Involving Living Systems. In: Roll, WG (ed) Research in
Parapsychology 1978. Abstracts and Papers from the Twenty-first Annual Convention of the
Parapsychological Associaton, 1978 (pp. 111-115). The Scrarecrow Press: Metuchen, N.J.
Braud WG 1990 Distant Mental Influence of Rate of Hemolysis of Human Red Blood Cells. Journal of the
American Society for Psychical Research 84: 1-24
Braud WG 2000 Wellness Implications of Retroactive Intentional Influence: Exploring an Outrageous
Hypothesis. Alternative Therapies 6: 37-48
Braud WG, Davis G, Wood R 1979 Experiments with Matthew Manning. Journal of the Society for Psychical
Research 50: 199-223
Braud WG, Jackson J 1982 The use of Ideomotor Reactions as Psi Indicators. Parapsychology Review 13(2):
10-11
Braud WG, Schlitz MJ 1989a A Methodology for Objective Study of Transpersonal Imagery. Journal of
Scientific Exploration 3: 43-63
Braud WG, Schlitz MJ 1989b Possible Role of Intuitive Data Sorting in Electrodermal Biological Psychokinesis
(Bio-PK). Journal of the American Society for Psychical Research 83: 289-302
Braud WG, Schlitz MJ 1991 Conscious interactions with remote biological systems: Anomalous intentionality
effects. Subtle Energies 2: 1-46
Braud WG, Schlitz MJ, Schmidt H 1989 Remote Mental Influence of Animate and Inanimate Target Systems.
The Parapsychological Association 32nd Annual Convention. Proceedings of Presented Papers
(pp. 12-25). The Parapsychological Association: Durham, NC

Braud WG, Shafer D, Andrews S 1990 Electrodermal Correlates of Remote Attention: Autonomic Reaction to
an Unseen Gaze. The Parapsychological Association 33rd Annual Convention. Proceedings of
Presented Papers (pp. 14-28). The Parapsychological Association: Durham, NC
Braud WG, Shafer D, Andrews S 1993a Further Studies of Autonomic Detection of Remote Staring, New
Control Procedures, and Personality Correlates. Journal of Parapsychology 57: 391-409
Braud WG, Shafer D, Andrews S 1993b Reactions to an Unseen Gaze (Remote Attention): A Review, with New
Data on Autonomic Staring Detection. Journal of Parapsychology 57: 373-390
Braud WG, Shafer D, McNeill K, Guerra V 1995 Attention Focusing Facilitated Through Remote Mental
Interaction. Journal of the American Society for Psychical Research 89: 103-115
Bunnell T 1999 The Effect of "Healing with Intent" on Pepsin Enzyme Activity. Journal of Scientific
Exploration 13: 139-148
Colwell J, Schröder S, Sladen D 2000 The ability to detect unseen staring: A literature review and empirical
tests. British Journal of Psychology 91: 71-85
Fowles DC, Christie MJ, Edelberg R, Grings WW, Lykken DT, Venables PH 1981 Publication
Recommendations for Electrodermal Measurements. Psychophysiology 18: 232-239
May E, Utts JM, Spottiswoode SJP 1995 Decision Augmentation Theory: Toward a Model of Anomalous
Mental Phenomena. Journal of Parapsychology 59: 195-220
May E, Utts JM, Spottiswoode SJP 1996 Decision augmentation theory: Applications to the random number
generator. Journal of Scientific Exploration 9: 453-488
Millar B 1978 The Observational Theories: A Primer. European Journal of Parapsychology 2: 304-332
Plihal W 2000 Arbeitsprogramm. Abteilung Experimentelle Studien am Institut für Grenzgebiete der
Psychologie und Psychohygiene e.V. (Unpublished)
Radin DI, Machado FR, Zangari W 1998 Effects of Distant Healing Intention Through Time and Space: Two
Exploratory Studies. The Parapsychological Association 41st Annual Convention. Proceedings of
Presented Papers (pp. 143-161). The Parapsychological Association: Durham, NC
Rebman JM, Radin DI, Hapke RA, Gaughan KZ 1996 Remote Influence of the autonomic nervous system by a
ritual healing technique. The Parapsychological Association 39th Annual Convention. Proceedings
of Presented Papers (pp. 133-147). The Parapsychological Association: Durham, NC
Rosenthal R 1991 Meta-Analytic Procedures for Social Research, revised edn. Newbury Park: Sage
Rosenthal R 1994 Parametric Measures of Effect Size. In: Cooper H, Hedges LV (eds) The Handbook of
Research Synthesis (pp. 231-244). New York: Russell Sage Foundation
Schlitz MJ, Braud WG 1997 Distant Intentionality and Healing: Assesing the Evidence. Alternative Therapies in
Health and Medicine 3: 62-73
Schmidt S, Schneider R, Binder M, Bürkle D, Walach H (2001). Investigating Methodological Issues in EDA-
DMILS: Results from a Pilot Study. Journal of Parapsychology 65: 59-82
Schmidt S, Walach H 2000 Electrodermal Activity (EDA) - state of the art measurement and techniques for
parapsychological purposes. Journal of Parapsychology 64: 139-162
Schwartz GE, Russek LGS 1999 Registration of Actual and Intended Eye Gaze: Correlation with Spiritual
Beliefs and Experiences. Journal of Scientific Exploration 13: 213-229
Stokes DM 1997 The Natur of Mind. Parapsychology and the Role of Consciousness in the Physical World.
Jefferson: McFarland
von Lucadou W 1995 The model of pragmatic information (MPI). European Journal of Parapsychology 11: 58-
75
Watt CA, Ravenscroft J, McDermott Z 1999 Exploring the Limits of Direct Mental Influence: Two Studies
Comparing 'Blocking' and 'Co-Operating' Strategies. Journal of Scientific Exploration 13: 515-
535
Wezelman R, Radin DI, Rebman JM, Stevens PR 1996 An experimental test of magical healing rituals in mental
influence of remote human physiology. The Parapsychological Association 39th Annual
Convention. Proceedings of Presented Papers. (pp. 1-12). The Parapsychological Association:
Durham, NC

Wiseman R, Schlitz MJ 1997 Experimenter Effects and the Remote Detection of Staring. Journal of
Parapsychology 61: 197-207
Wiseman R, Smith MD 1994 A Further Look at the Detection of Unseen Gaze. In: Bierman DJ (ed), The
Parapsychological Association 37th Annual Convention. Proceedings of Presented Papers (pp.
465-478). The Parapsychological Association: Durham, NC
Reference List DMILS studies
Braud WG 1978 Allofeedback: Immediate Feedback for a Psychokinetic Influence upon another Person's
Physiology. In: Roll WG (ed) Research in Parapsychology 1977. Abstract and Papers from the
Twentieth Annual Convention of Parapsychological Association, 1977 (pp. 123-134). The
Scarecrow Press: Metuchen, N.J.
Braud WG 1979 Conformance Behavior Involving Living Systems. In: Roll WG (ed) Research in
Parapsychology 1978. Abstracts and Papers from the Twenty-first Annual Convention of the
Parapsychological Associaton, 1978 (pp. 111-115). The Scrarecrow Press: Metuchen, N.J.
Braud WG 1985 Enhancing and Suppressing Psi Functioning. Blocking/Shielding Psychic Functioning through
Psychological and Psychic Techniques: A Report of Three Preliminary Studies. In: White RA,
Solfvin J (eds), Research in Parapsychology 1984. Abstracts and Papers from the Twenty-seventh
Annual Convention of the Parapsychological Association, 1984 (pp. 42-44). The Scarecrow Press:
Metuchen, N.J.
Braud WG 1993 On the Use of Living Target Systems in Distant Mental Influence Research. In: Coly L,
McMahon JDS (eds) PSI-Research Methodology: A Re-Examination. Proceedings of an
International Conference Held in Chapel Hill, North Carolina October 29-30, 1988 (pp. 149-181).
Parapsycholocical Foundation: New York
Braud WG, Davis G, Wood R 1979 Experiments with Matthew Manning. Journal of the Society for Psychical
Research 50: 199-223
Braud WG, Schlitz MJ 1983 Psychokinetic Influence on Electrodermal Activity. Journal of Parapsychology 47:
95-119
Braud WG, Schlitz MJ 1989a A Methodology for Objective Study of Transpersonal Imagery. Journal of
Scientific Exploration 3: 43-63
Braud WG, Schlitz MJ 1989b Possible Role of Intuitive Data Sorting in Electrodermal Biological Psychokinesis
(Bio-PK). Journal of the American Society for Psychical Research 83: 289-302
Braud WG, Schlitz MJ 1991 Conscious interactions with remote biological systems: Anomalous intentionality
effects. Subtle Energies 2: 1-46
Braud WG, Schlitz MJ, Collins J, Klitch H 1985 Further Studies of the Bio-PK-Effekt: Feedback, Blocking;
Specifictiy/Generality. In: White RA, Solfvin J (eds) Research in Parapsychology 1984. Abstracts
and Papers from the Twenty-seventh Annual Convention of the Parapsychological Association,
1984 (pp. 45-48). The Scarecrow Press: Metuchen, N.J.
Delanoy DL, Morris RL 1998a A DMILS Study with Experimenter Trainees. The Parapsychological Association
41st Annual Convention. Proceedings of Presented Papers (pp. 22-35). The Parapsychological
Association: Durham, NC
Delanoy DL, Morris RL 1998b A DMILS training study utilising two shielded environments. European Journal
of Parapsychology 14: 52-67
Delanoy DL, Morris RL, Brady C, Roe A 1999 An EDA DMILS Study Exploring Agent-Receiver Pairing. The
Parapsychological Association 42nd Annual Convention. Proceedings of Presented Papers (pp. 68-
82). The Parapsychological Association: Durham, NC
Delanoy DL, Sah S 1994 Cognitive and physiological PSI responses to remote positive and neutral emotional
states. The Parapsychological Association 37th Annual Convention. Proceedings of Presented
Papers (pp. 128-138). The Parapsychological Association: Durham, NC
Radin DI, Taylor RK, Braud WG 1995 Remote Mental Influence of Human Electrodermal Activity: A Pilot
Replication. European Journal of Parapsychology, 11, 19-34

Rebman JM, Radin DI, Hapke RA, Gaughan KZ 1996 Remote Influence of the autonomic nervous system by a
ritual healing technique. The Parapsychological Association 39th Annual Convention. Proceedings
of Presented Papers (pp. 133-147). The Parapsychological Association: Durham, NC
Schlitz MJ, Braud WG 1985 Reiki-plus natural healing: an ethnographic and experimental study. PSI-Research
4: 100-123
Schlitz MJ, Braud WG 1997 Distant Intentionality and Healing: Assesing the Evidence. Alternative Therapies in
Health and Medicine 3: 62-73
Schneider R, Binder M, Walach H. 2000 Examining the role of neutral and personal experimenter-participant
interaction: Results from an EDA-DMILS experiment. Journal of Parapsychology, 64, 181-194
Watt CA, Ravenscroft J, McDermott Z 1999 Exploring the Limits of Direct Mental Influence: Two Studies
Comparing 'Blocking' and 'Co-Operating' Strategies. Journal of Scientific Exploration 13: 515-535
Wezelman R, Radin DI, Rebman JM, Stevens PR 1996 An experimental test of magical healing rituals in mental
influence of remote human physiology. The Parapsychological Association 39th Annual
Convention. Proceedings of Presented Papers. (pp. 1-12). The Parapsychological Association:
Durham, NC
Reference List Remote Staring studies
Braud WG, Shafer D, Andrews S 1993a Further Studies of Autonomic Detection of Remote Staring, New
Control Procedures, and Personality Correlates. Journal of Parapsychology 57: 391-409
Braud WG, Shafer D, Andrews S 1993b Reactions to an Unseen Gaze (Remote Attention): A Review, with New
Data on Autonomic Staring Detection. Journal of Parapsychology 57: 373-390
Schlitz MJ, LaBerge S 1997 Cover Observation Increases Skin Conductance in Subjects Unaware of When They
are Being Observerd: A Replication. Journal of Parapsychology 61: 185-196
Wiseman R, Schlitz MJ 1997 Experimenter Effects and the Remote Detection of Staring. Journal of
Parapsychology 61: 197-207
Wiseman R, Schlitz MJ 1999 Experimenter Effects and the Remote Detection of Staring: An Attempted
Replication. The Parapsychological Association 42nd Annual Convention. Proceedings of
Presented Papers (pp. 471-479). The Parapsychological Association: Durham, NC
Wiseman R, Smith MD 1994 A Further Look at the Detection of Unseen Gaze. In: Bierman DJ (ed) The
Parapsychological Association 37th Annual Convention. Proceedings of Presented Papers (pp.
465-478). The Parapsychological Association: Durham, NC
Wiseman R, Smith MD, Freedman D, Wasserman T, Hurst C 1995 Examining the Remote Staring Effect: Two
Further Experiments. The Parapsychological Association 38th Annual Parapsychological
Convention. Proceedings of presented papers (pp. 480-490). The Parapsychological Association:
Durham, NC