Content uploaded by Dean Radin
Author content
All content in this area was uploaded by Dean Radin on Feb 23, 2015
Content may be subject to copyright.
PHYSICS ESSAYS 25, 2 (2012)
Consciousness and the double-slit interference pattern: Six experiments
Dean Radin,
1,a)
Leena Michel,
1
Karla Galdamez,
1
Paul Wendland,
2
Robert Rickenbach,
3
and Arnaud Delorme
4
1
Institute of Noetic Sciences, 625 Second St., Petaluma, California 94952, USA
2
4558 La Brea St., Oxnard, California 93035, USA
3
Micronor Inc., 750 Mitchell Rd., Newbury Park, California 91320, USA
4
Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California
San Diego, 9500 Gilman Dr., La Jolla, California 92093, USA
(Received 9 April 2011; accepted 25 January 2012; published online 16 May 2012)
Abstract: A double-slit optical system was used to test the possible role of consciousness in the
collapse of the quantum wavefunction. The ratio of the interference pattern’s double-slit spectral
power to its single-slit spectral power was predicted to decrease when attention was focused toward
the double slit as compared to away from it. Each test session consisted of 40 counterbalanced
attention-toward and attention-away epochs, where each epoch lasted between 15 and 30 s. Data
contributed by 137 people in six experiments, involving a total of 250 test sessions, indicate that on
average the spectral ratio decreased as predicted (z=-4:36, p=6·10-6). Another 250 control
sessions conducted without observers present tested hardware, software, and analytical procedures
for potential artifacts; none were identified (z=0:43, p=0:67). Variables including temperature,
vibration, and signal drift were also tested, and no spurious influences were identified. By contrast,
factors associated with consciousness, such as meditation experience, electrocortical markers of
focused attention, and psychological factors including openness and absorption, significantly
correlated in predicted ways with perturbations in the double-slit interference pattern. The results
appear to be consistent with a consciousness-related interpretation of the quantum measurement
problem. Ó2012 Physics Essays Publication. [DOI: 10.4006/0836-1398-25.2.157]
R´
esum ´
e: Un syst `
eme optique de double fente est utilis ´
epourtesterlerˆolepossibledelaconscience
dans la r ´
eduction du paquet d’onde en m ´
ecanique quantique. On fait l’hypoth `
ese que l’intensit ´
e
relative de la figure d’interf ´
erence produite par une double fente devrait diminuer quand le sujet
focalise son attention sur la double fente. Cette intensit ´
e est estim ´
ee en calculant une mesure
normalis ´
ee de la puissance spectrale `
alafr
´
equence spatiale o `uleph
´
enom `
ene d’interf ´
erence est
observ ´
e. Chaque session est compos ´
ee de 40 essais contrebalanc ´
es d’une dur ´
ee comprise entre 15 et
30 secondes, o `u le sujet porte son attention soit vers soit au loin de la double fente. Les donn ´
ees
fournies par 137 sujets, pour un total de 250 s ´
eances d’essais recueillis lors de 6 exp ´
eriences,
indiquent qu’en moyenne, l’intensit ´
e de la figure d’interf ´
erence diminue dans la direction escompt ´
ee
(z =-4,36, p =6·10
-6
). En outre, 250 sessions de contr ˆole effectu ´
ees en l’absence de tout
observateur ont permis de tester la pr ´
esence d’art ´
efacts potentiels en provenance du mat ´
eriel, des
logiciels et des m ´
ethodes d’analyse; aucun effet n’est observ ´
e(z=0,43, p =0,67). D’autres variables
telles que la temp ´
erature, les vibrations et la d ´
erive du signal ont ´
egalement ´
et ´
etest´
ees, et aucune
influence parasite n’a pu ˆetre identifi ´
ee. En revanche, les facteurs associ ´
es `
a la conscience, comme
l’exp ´
erience de la m ´
editation, les mesures ´
electrophysiologiques de l’attention focalis ´
ee, et les
facteurs psychologiques comme les mesures d’absorption et d’ouverture d’esprit sont significative-
ment corr ´
el ´
es dans une direction pr ´
edictible avec les perturbations d’intensit ´
edelafigure
d’interf ´
erence. Ces r ´
esultats sont en accord avec les interpr ´
etations qui lient la conscience au
probl `
eme de la mesure quantique.
Key words: Quantum Measurement Problem; Consciousness; Double-Slit Experiment; Mind–Matter Interaction.
I. INTRODUCTION
[The double-slit experiment] has in it the heart of
quantum mechanics. In reality, it contains the only
mystery.
Richard Feynman
1
In this opening quotation, Feynman is referring to
the quantum measurement problem (QMP), and in
particular to the curious effect whereby quantum objects
appear to behave differently when observed than when
unobserved.
2
The QMP is a problem because it violates
the common-sense doctrine of realism, which assumes
that the world at large is independent of observation. The
conflict between naive realism and what the QMP implies
a)
dradin@noetic.org
0836-1398/2012/25(2)/157/15/$25.00 Q2012 Physics Essays Publication157
forced many of the early developers of quantum theory to
ponder the meaning of observation and measurement.
3,4
Some, like Pauli, Jordan, and Wigner, believed that some
aspect of consciousness—referring to mindlike capacities
such as awareness, attention, and intention—was funda-
mental in understanding the QMP.
5,6
Jordan wrote,
Observations not only disturb what has to be
measured, they produce it. . . . We compel [the
electron] to assume a definite position. . . . We
ourselves produce the results of measurement.
7
This strong view of the role of consciousness in the
QMP has been endorsed by physicists ranging from
d’Espagnat to von Neumann, from Stapp to Squires.
8–11
The significance of the proposition and the prominence of
those who have proposed it have made the idea difficult to
blithely ignore, but to many it challenges a deeply held
intuition that the physical world was here, more or less in
its present form, long before human consciousness
evolved to observe it. As a result, many continue to resist
the idea that consciousness has anything to do with the
formation of physical reality.
12,13
One approach to eliminating the observer from the
QMP has been to reframe the problem by proposing that
all that observation does is increase our knowledge about
a measured system. Thus, according to Zeilinger,
From that position, the so-called measurement
problem . . . is not a problem but a consequence of
the more fundamental role information plays in
quantum physics as compared to classical physics.
14
Another approach is to argue that decoherence theory
obviates the QMP, but this proposal is not without
problems.
15,16
Others have attempted to finesse the QMP
by denying that there ever was a problem. According to
Goldstein,
Many physicists pay lip service to . . . the notion that
quantum mechanics is about observation or results
of measurement. But hardly anybody truly believes
this anymore—and it is hard for me to believe
anyone really ever did.
12
Still others have proposed that the only unambiguous way
to avoid the role of the observer in physics is to deny the
belief that we have free will.
17
While free will as a brain-
generated illusion is the prevailing assumption in the
neurosciences today,
18
that idea remains at odds with the
only direct form of contact we have with reality—
subjective experience—which paradoxically allows for
the experience of deciding to believe that free will does
not exist.
Philosophical and theoretical arguments aside, the
double-slit experiment suggests a way to explore the
meaning of observation in the QMP, and in particular the
possible role of consciousness. It is based on two
assumptions: (a) If information is gained—by any
means—about a photon’s path as it travels through two
slits, then the interference pattern will collapse in
proportion to the certainty of the knowledge gained;
and (b) if some aspect of consciousness is a primordial,
self-aware feature of the fabric of reality, and that
property is modulated by us through capacities we know
as attention and intention, then focusing attention on a
double-slit system may in turn affect the interference
pattern. The first assumption is well established.
19
The
second, based on the idea of panpsychism, is a
controversial but respectable concept within the philoso-
phy of mind.
20
A. Experimental background
Three earlier experiments have employed optical
interferometers to investigate the possibility of what we
will call the ‘‘consciousness collapse hypothesis.’’ Two of
those experiments employed a double-slit system
21
and
one used a Michelson interferometer.
22
In the first study,
a team at York University used a HeNe laser and double-
slit apparatus to test unselected volunteers who were
asked to
observe, by extra-sensory means . . . monochromatic
light passing through a double-slit optical appara-
tus, prior to its registration as an interference
pattern by an optical detector.
21
That experiment was followed up at Princeton University
with participants who were experienced at attention-
focusing tasks, and with a refined version of the York
optical system.
21
The goal in both experiments was to
shift the mean of a variable that measured the wavelike
versus particlelike nature of the interference pattern. The
York team reported a nonsignificant mean shift opposite
to the predicted direction (although curiously, the data
showed a significantly larger variance than would be
expected by chance); the Princeton team reported a
modestly significant mean shift in the predicted direction
(p=0:05).
The third experiment involved a Michelson interfer-
ometer located inside a light-tight, double-steel-walled,
electromagnetically shielded chamber.
22
Participants one
at a time sat quietly outside the chamber and were
instructed to direct their attention toward or away from
one arm of the interferometer. Interference patterns were
recorded once per second and the average intensity levels
of those patterns were compared in 30 s counterbalanced
attention-toward and attention-away epochs. At the
completion of the experiment, the results were in
accordance with the prediction (p=0:002), i.e., interfer-
ence was reduced during the observation periods. This
outcome was primarily due to nine sessions involving
experienced meditators (p=9:4·10-6). The remaining
nine sessions with nonmeditators did not produce effects
differing from those of chance (p=0:61). Control runs
using the same setup but with no observers present also
produced chance results.
In sum, of three experiments relevant to the issue at
hand, two support the consciousness collapse hypothesis
to a statistically significant degree, and one does not.
Given the importance of the QMP and the potential of
158 Phys. Essays 25, 2 (2012)
this type of experiment to inform it empirically, we
conducted a new series of studies to reexamine the
hypothesis using a double-slit optical system. To avoid
potential biases associated with selective data reporting,
all completed test sessions in the experiments described
here, both preplanned and exploratory, were considered
part of the formal experimental database and are
reported. A half dozen incomplete sessions conducted as
brief demonstrations are not included, nor are a few
sessions that were interrupted by power failures or data-
acquisition glitches.
II. EXPERIMENT 1
A. Method
1. Apparatus
A 5 mW linearly polarized HeNe laser beam (632.8
nm; model 25 LHP 151–249, CVI Melles-Griot, Albu-
querque, NM, USA) was passed through a neutral density
filter (i.e., a gray filter that attenuates all wavelengths
equally, in this case by 90%; Rolyn Optics, Covina, CA,
USA), and then through two slits etched through a metal
foil slide with widths of 10 lm and a separation of 200 lm
(Lenox Laser, Glen Arm, MD, USA). The resulting
interference pattern was recorded by a 3000-pixel charge-
coupled device line camera, which had a pixel size of 7.0
by 0.2 lm and 12-bit analog-to-digital resolution (Thor-
labs Model LC1-USB, Newton, NJ, USA). The camera
was located 10.4 cm from the slits. This laser was selected
because it has a coherence length of more than a meter,
which helped to produce sharp interference fringes. Prior
to its use in this study it had been in operation for several
thousand hours; this provided improved power-output
stability as compared to a new laser. A duplicate optical
system was also constructed with similar components, for
tests conducted outside the laboratory.
The apparatus was housed inside a custom-machined
aluminum housing and painted matte black inside and out
(see Fig. 1). The laser and camera were allowed to warm
up for a minimum of 45 min prior to test sessions. The
experiment was controlled by a Windows Vista computer
running a program written in Microsoft Visual Basic 2008
and augmented by software libraries from Thorlabs and
National Instruments Measurement Studio 8.1.
To measure perturbations in the wavefunction, the
interference pattern recorded by the line camera was
analyzed with a fast Fourier transform to quantify the
power associated with the two dominant spatial wave-
lengths: a shorter wavelength associated with the double-
slit interference pattern (call this power PD) and a longer
wavelength associated with the diffraction pattern pro-
duced by each slit (PS) (see Fig. 2). The fraction of (log)
spectral power associated with the interference pattern
was D=½PD=ðPD+PSÞ, and that with the diffraction
pattern was S=½PS=ðPD+PSÞ. The ratio of these frac-
tions, R=D=S, was the preplanned variable of interest.
2. Procedure
During a test session, participants were instructed by
the computer to direct their attention toward the double-
slit apparatus or to withdraw their attention and relax. To
announce the attention-toward task, a computer-synthe-
sized voice said, ‘‘Please influence the beam now’’; for
attention away, it said, ‘‘You may now relax.’’
Participants were asked to direct their attention
toward two tiny slits located inside a sealed black box
(the double-slit optical system). It was explained that this
task was purely in the ‘‘mind’s eye,’’ i.e., an act of
imagination. To many this instruction proved to be
somewhat abstract, so to assist their imagination they
were shown a 5 min animation of the double-slit
experiment, where a particle detector was portrayed as
FIG. 1. From the left, the regulated power supply for the laser, the HeNe
laser tube extending out of the optical apparatus, and the camera
attached to the right side of the apparatus. The housing is a precision-
machined aluminum box, painted matte black and optically sealed.
FIG. 2. (Color online) (A) Interference-pattern intensity recorded by the
3000-pixel line camera, averaged over 10,000 camera frames. (B) Log of
spatial spectral power with double-slit power peaking around wave-
number 45 (equivalent to a wavelength of about 69 pixels) and single-slit
power at 1. The peak around wavenumber 90 is a harmonic of the
double-slit frequency.
Phys. Essays 25, 2 (2012) 159
analogous to a human eye. If the task was still unclear, it
was suggested that they could try to mentally block one of
the slits, or to ‘‘become one with’’ the optical system in a
contemplative way, or to mentally push the laser beam to
cause it to go through one of the two slits rather than both.
Once a test session was under way, the computer-
synthesized voice instructions directed the participants’
attention toward or away from the optical system in 15 s
epochs. A single test session consisted of 40 such epochs
presented in a counterbalanced order. The counterbal-
ancing scheme consisted of five randomly assigned
groups, where a group followed either the assignment
order ABBA BAAB or the order BAAB ABBA, where A
and Brefer to attention toward and attention away. Test
sessions began by collecting between 15 and 20 s of
baseline data, followed by the 40 instructed epochs.
Participants one at a time sat quietly about 2 m from
the sealed optical apparatus (see Fig. 3). They were
instructed not to touch or approach the device at any
time. Test sessions were conducted inside a solid steel,
double-walled, electromagnetically shielded chamber at
the Institute of Noetic Sciences (Series 81 Solid Cell
chamber, ETS-Lindgren, Cedar Park, TX, USA). Elec-
trical-line power in the chamber was conditioned through
a high-performance electromagnetic interference filter
(ETS-Lindgren filter LRW-1050-S1), and to further
reduce potential electromagnetic interference, the optical
system and computer were powered by a battery-based
uninterruptable power supply. This testing chamber in its
unadorned state is a rather imposing steel cube without
windows, so to make it more welcoming the walls and
ceiling were covered with a tan-colored muslin fabric,
antistatic carpeting was installed on the floor, and
comfortable furniture was placed inside the chamber.
The computer presented a strip-chart display updated
once a second with the spectral-ratio Rvalues. Participants
were invited to look at the graph to gain near real-time
feedback about their performance, or to view an alternative
meter-type display showing the same information.
3. Analysis
The computer calculated Rfive times a second and
then stored and displayed the average of the last eight
measurements once per second. To combine Racross
different test sessions, these values were normalized into
standard normal deviates as Rz=ðR-lRÞ=rR, where lwas
the mean of all Rvalues in a given session and rwas the
standard deviation. A second data array of the same
length specified the attention-toward and attention-away
conditions for each sample in Rz(call this array C)From
these two arrays, the differential measure DR=
RzA-
RzB
was formed, where
RzA was the mean of Rzvalues
collected during in the attention-toward condition and
RzB the mean in the attention-away condition.
To assess whether DRcould have occurred by chance,
a nonparametric randomized permutation procedure was
employed. This avoided distributional assumptions about
Rzand took into account possible autocorrelated
dependencies between successive Rzsamples. To perform
this analysis, (1) the original array of Rzsamples was
circularly shifted Nsteps, where Nwas randomly selected
between 1 and the total number of samples in the array;
and (2) using the new, randomly time-shifted array with
the original condition array C, the differential measure DR
was determined (as described in the previous paragraph)
and stored. Steps 1 and 2 were then repeated 5000 times.
This procedure generated a distribution of possible DR
outcomes, differing from the original only by when each
test session effectively began. The statistic used to assess
DRo was z=ðDRo-lDRÞ=rDR, where the subscript oin DRo
refers to the experimentally observed value for DR,the
term lDRindicates the mean of the distribution of
randomly time-shifted DRvalues, and the term rDR
indicates the standard deviation of that same distribution.
The value zwas thus a standard normal deviate, with
expected mean 0 and standard deviation 1.
4. Hypothesis
The consciousness collapse hypothesis predicted that
the act of focusing attention toward the double-slit would
cause Rrecorded during attention-toward epochs to
decrease as compared to during attention-away epochs.
Because being able to concentrate one’s attention was an
important aspect of the assigned instructions, it was
further predicted that participants with attention training,
such as meditators, would perform better than those
without such training.
B. Results
A minimum of 30 sessions were planned. Unselected
participants were recruited by convenience, and 15
participants ended up contributing 35 sessions (the five
FIG. 3. Experiments were conducted inside a double-steel-walled,
electromagnetically shielded chamber. The computer (PC) controlled
all aspects of the experiment, including announcement of the attention-
toward and attention-away instructions and acquisition of interference-
pattern images from the double-slit device. In experiment 3, thermo-
couples were placed on the laser tube (T1), on the double-slit housing
(T2), near the housing (T3), and about 1.5 m in front of the participant
(T4).
160 Phys. Essays 25, 2 (2012)
extra sessions are included in this analysis to avoid
selective reporting biases). At 15s=epoch·1R=s·
40 epochs=session, this produced a total of 21 000 R
measurements, half in the attention-toward condition and
half in the attention-away. Twenty-four sessions were
contributed by people who reported some meditation
experience and 11 by nonmeditators. No attempt was
made to distinguish among different styles of meditation
or to formally assess meditation expertise. An additional
34 calibration sessions were run using the same hardware
and software and with the apparatus in the same location
as during experimental runs, but with the computer’s
speakers muted and no one present in the shielded
chamber.
Across the 35 sessions the recorded Rsignal was
fairly stable. The variation across all sessions was an
average of v=0:5%, where vper session was calculated as
v=½max ðRÞ-min ðRÞ=
R.
Figure 4 illustrates the normalized Rvalue and the
counterbalanced condition assignments for one session.
As with any interferometer, some drifts and oscillations
are to be expected, thus the signal variations in Fig. 4 are
not as large as they may appear to be; also, the graph
exaggerates the apparent variance due to the signal-
normalization process. The figure also indicates the means
of Rin the attention-away and attention-toward condi-
tions, along with one standard error of the mean error
bars (determined using the randomized-permutation
technique).
Table I indicates that for experiment 1, the spectral
ratio Rdeclined modestly in accordance with the con-
sciousness collapse hypothesis (z=-1:56, where zis the
normalized statistic associated with DR,as described in
Subsection II.A.3), and a somewhat stronger effect was
observed for meditators (z=-1:58) than for nonmeditators
(z=-0:43). Control tests using the same equipment in the
same location, but without participants present, showed a
slightly positive result (i.e., opposite to the hypothesis,
z=0:15). Table I also displays the effect size for each
condition, where es=z=ffiffiffiffi
N
pand Nis the number of
sessions. Effect size is a convenient way to compare the
magnitude of effects across different studies and subsets of
studies, because it is independent of the number of sessions.
From this perspective, meditators produced effects 2.5
times as large as those produced by nonmeditators
(es=-0:32 and -0.13, respectively). The effect size for all
data combined (es=-0:26) is comparable in absolute
magnitude to experimental effects commonly observed in
the behavioral and social sciences. For example, a 2003
meta-analysis of a century of social-psychology experi-
ments, involving some 25 000 studies and 8·106people,
23
found that the grand mean effect size was es=0:21. Effects
of this magnitude are generally considered ‘‘small.’’ Small
effect sizes in the behavioral sciences are considered real
phenomena,
24
but they require repeated testing to provide
enough statistical power for detection.
The last two columns in Table I refer to a lag/lead
analysis. This was conducted because it took the computer
a few seconds to speak aloud the condition assignment and
for participants to shift their mental attention; thus if the
change in Rwas indeed related to shifts of attention, then
we might expect a short lag in the response time of R.The
result is illustrated in Fig. 5. The x-axis shows the effects of
FIG. 4. (Color online) Example of the data for the normalized spectral
ratio Rrecorded in one test session. Attention assignments are plotted
for illustrative purposes as the values +1 for pre- and postsession
baseline periods, 0 for attention away, and -1 for attention toward. On
the right side of the graph the means for Rby condition are shown, along
with one standard error of the mean error bars. Normalizing the signal
exaggerates its apparent variance; this signal varied from the grand mean
measured over all 35 sessions by an average of 0.5%.
TABLE I. Summary of results from experiment 1. Effect size es expresses
the magnitude of the effect per session, zis the outcome of comparing R
between the two attention conditions, pis the one-tailed probability
associated with z, ‘‘0 lag’’ indicates the analysis with Rin exact time
synchrony with the change of the attention condition, and ‘‘-2 lag’’
indicates the same analyses with Rlagged 2 s after the change of the
attention condition.
Sessions es z 0 lag p0 lag z-2 lag p-2 lag
All sessions 35 -0:26 -1:56 0.06 -1:84 0.03
Meditators 24 -0:32 -1:58 0.06 -1:64 0.05
Nonmeditators 11 -0:13 -0:43 0.34 -0:87 0.19
Controls 34 -0.15 -0.85 0.80 -0.85 0.80
FIG. 5. (Color online) The zscores for lag/lead analysis. Zero lag refers
to Rsynchronized with the beginning of the announcement of the
attention conditions. Negative lags refer to Rafter changes in attention
assignments, and positive lags to Rbefore such changes.
Phys. Essays 25, 2 (2012) 161
time-lagging Rwith respect to the onset of the attention-
condition assignments. Thus a lag of 0 refers to Rrecorded
in synchrony with the onset of the attention assignment, a
lag of -1 to the same analysis with Rlagged 1 s after the
condition assignments, a lag of +1toRleading 1 s before
the condition announcement, and so on. This analysis
indicates that for all data combined, zdeclined when
lagged a few seconds. By contrast, the control test data
remained positive and relatively flat through 10 s of lag/
lead analysis. The figure shows that the optimal lag length
was about 2 s, so Table I and subsequent tables summarize
this analysis for a lag of -2.
From the example session shown in Fig. 4,thequestion
may arise as to whether the primary effect in this experiment
might have been due to fortuitous matches between the
condition assignments and slow oscillations in R.Inthe
present case,randomly assigned counterbalanced conditions
were employed to reduce the possibility of chance depen-
dencies on such oscillations, but this was further examined
by smoothing Rwith a 15-sample moving linear regression (1
sample/s) and then determining the difference between the
original signal and the smoothed signal to create residuals
(see Fig. 6). Whenresiduals from all sessions werecombined
and analyzed using the permutation technique, the result
remained significant at zðlag 0Þ=-2:8, as shown in Fig. 7.
Thus drifts and oscillations do not appear to be responsible
for the observed results. To help maintain as transparently
simple an analytical approach as possible, the normalized
unfiltered signal was used in all subsequent analyses.
III. EXPERIMENT 2
Participants’ comments about the first experiment
suggested that the attention-focusing task might be easier
to perform with audio feedback to allow the experiment
to be performed with eyes closed; in addition, a number of
participants suggested that 15 s periods were too short to
fully reorient their attention. Thus in this experiment an
audio-feedback technique was developed and the condi-
tion-epoch lengths were increased to 30 s. Along with the
original double-slit device, a duplicate system was used
for some remote test sessions conducted at a Zen
Buddhist temple, which was a convenient location to
recruit meditators for the experiment.
A. Method
For audio feedback, during attention-away periods
the computer played a soft, continuous drone tone, and
during attention-toward periods it played a musical note
that changed pitch to reflect the real-time value of R.
Participants were instructed to direct their attention
toward the double-slit device as in the initial study. If
they were successful, then the double-slit spectral power
was predicted to decline, and in turn the pitch of the
musical note would also decline.
B. Results
A minimum of 30 sessions were planned. The test
concluded after 19 participants contributed 31 sessions. In
the Institute of Noetic Sciences (IONS) laboratory, three
meditators contributed 11 sessions and four nonmedita-
tors contributed 7 sessions. In the Zen Buddhist temple,
12 meditators contributed a total of 13 sessions. The latter
tests were supervised by one of the present authors
(Wendland), using a duplicate double-slit apparatus.
Three control sessions, each consisting of 31 sessions,
were later conducted in the IONS lab using the original
double-slit apparatus.
Table II summarizes the results. To perform the
control tests, the condition sequences assigned in the
FIG. 6. (A) Original Rsignal (in black) and 15-sample smoothed fit (in
white). (B) Difference between curves in part (A); means and error bars
on the right side indicate that the attention-toward mean residual
remains significantly below the attention-away mean residual.
FIG. 7. (Color online) Lag/lead analysis for residuals.
162 Phys. Essays 25, 2 (2012)
actual experiment were compared against new Rdata
generated without human observers present. This proce-
dure was performed three times, each time using a new set
of control data.
This study provided modest evidence in favor of the
hypothesis (z=-1:39), and in the IONS lab meditators
again showed superior performance (z=-2:04) as com-
pared to nonmeditators (z=-0:49). In terms of effect size,
the overall results were nearly identical to those observed
in the first experiment (-0:25 in experiment 2 vs. -0:26 in
experiment 1). There was no evidence that Rlagged the
real-time condition assignments as it did in experiment 1.
Control tests identified no artifacts that might have biased
the results recorded in the IONS lab.
The Zen meditators using the second double-slit device
failed to support the hypothesis. Two differences in that
study are noteworthy because they may have introduced
unforeseen variance: (a) The test sessions took place outside
a controlled laboratory environment, and (b) the Zen
temple was in a rural location with erratic line power. In an
attempt to accommodate unpredictable ‘‘brownouts’’ in line
power, a voltage converter was used to provide uniform
power to the laser below its usual 110 V operating
conditions. Both of these factors may have altered the
experiment outcome in unanticipated ways, but we never-
theless include those data here for the sake of completeness.
IV. EXPERIMENT 3
Participants in sessions in the first two studies that
were conducted in the laboratory were seated about 2 m
from the optical system. In some cases they may have
leaned toward the device or faced it more directly when
focusing their attention, and leaned back or turned away
while relaxing. If this occurred systematically, it might
have introduced changes in radiant heat impinging on the
optical system, and that in turn might have influenced the
interference pattern. For example, the distance between
the slits or the length of the HeNe laser tube might have
expanded or contracted slightly due to temperature
fluctuations. To test this possibility, a third experiment
was designed to explore the effects of human body heat in
proximity to the double-slit apparatus.
A. Method
Sessions were conducted with thermocouples (model
5TC-IT-T-30-72, accuracy rated at 0.5 8C; Omega Engi-
neering, Stamford, CT, USA; monitored by a Measurement
Computing Corporation analog-to-digital converter, model
USB-TC-AI, Norton, MA, USA placed in four locations, as
shown in Fig. 3. Each session consisted of 40 counterbal-
anced 30 s epochs. Individual sessions were identified that
showed a marked differential decline in R, and then the
thermocouple measurements in those sessions were com-
pared between the two conditions to see if temperature also
showed a differential effect. If it did, then what was
previously observed might have been due to subtle changes
in ambient temperature rather than to shifts in attention.
B. Results
A minimum of 30 sessions were planned; a total of 33
were contributed. Six meditators contributed 22 sessions
and seven nonmeditators contributed 11 sessions. The 22
meditator sessions resulted in a significant decline in z, with
an effect size comparable to that observed in the first
experiment (-0:39 vs. -0:32, respectively; see Table III). To
test for possible temperature-mediated effects, all meditator
sessions resulting in negative zscores were selected (N=16)
to form a subset with an especially strong statistical effect.
TABLE II. Summary of results from experiment 2. (See Table I caption for explanations.)
Number es z 0 lag p0 lag z-2 lag p-2 lag
All sessions 31 -0:25 -1:39 0.08 -1:24 0.11
Meditators 11 -0:62 -2:04 0.02 -2:01 0.02
Nonmeditators 7 -0:19 -0:49 0.31 -0:34 0.37
Zen Buddhist test 13 -0:01 -0:05 0.48 -0.03 0.51
Control 1 31 -0.08 -0.44 0.67 -0.37 0.64
Control 2 31 -0.15 -0.83 0.80 -0.77 0.78
Control 3 31 -0.16 -0.91 0.82 -0.77 0.78
TABLE III. Summary of results from experiment 3. (See Table I caption for explanations.)
Number es z 0 lag p0 lag z-2 lag p-2 lag
All sessions 33 -0:13 -0:72 0.24 -0:72 0.24
Meditators 22 -0:39 -1:84 0.03 -1:78 0.04
Nonmeditators 11 -0.50 -1.66 0.95 -1.56 0.94
Selected subset 16 -0:91 -3:65 0.0001 -3:65 0.0001
Laser temperature 16 -0.17 -0.67 0.50 -0.67 0.75
Apparatus temperature 16 -0.16 -0.65 0.52 -0.65 0.74
Ambient temp. near apparatus 16 -0.15 -0.61 0.54 -0.65 0.74
Ambient temp. near participant 16 -0.19 -0.76 0.45 -0.79 0.79
Phys. Essays 25, 2 (2012) 163
This selected subgroup produced a combined z=-3:65. If
that decline was attributable to systematic fluctuations in
ambient temperature, then this should have been detectable
in the thermocouple measurements.
Results of the analysis showed no significant temper-
ature differences measured on the laser tube (z=0:67), on
the housing of the double-slit apparatus (z=0:65), in front
of the apparatus (z=0:61), or within a meter in front of
the participant (z=0:76; see Table III). It is interesting
that all of these differential measurements were positive,
indicating that on average, temperature did increase
slightly on and around the double-slit system during the
attention-toward condition. But the lack of statistical
significance in the temperature measurements suggests
that the robust decline in Robserved in the 16 selected
sessions was probably not driven by systematic variations
in radiant heat. Other physical influences, such as human
bioelectromagnetic fields and vibrations, were not explic-
itly tested, although the optical system’s metal shielding
would have attenuated higher-frequency electromagnetic
influences, and vibrations would have been minimal
because all participants (especially the meditators) were
sitting quietly during the experiment. Nevertheless, the
possibility that undetected factors associated with prox-
imity of the human body influenced the results prompted
the design of the next experiment.
V. EXPERIMENT 4
If the consciousness collapse interpretation of the
QMP is valid, then this implies that the collapse occurs
when observation takes place, and not when the event is
generated.
25
To test this idea, a retrocausal version of the
experiment was designed. This test also provided a more
rigorous way to test the effect of participants’ proximity
to the optical system, because the data in this study were
generated and recorded with the apparatus located by
itself inside the electromagnetically shielded chamber, and
with no one else in the laboratory.
A. Method
Fifty sessions with 30 s counterbalanced epochs were
recorded in the IONS laboratory in April 2009. No one
was present during the process of data generation and
recording, and the data remained unobserved. In June
2009, participants were asked to view a strip-chart
display, which unbeknownst to them played back
prerecorded but previously unobserved data. As in the
other experiments, they were invited to cause the value R
to go as low as possible when the computer gave them the
instruction ‘‘Interfere with the beam now,’’ and to relax
when instructed ‘‘Now, please relax.’’ The design feature
that made this a retrocausal experiment was that the
attention-condition assignments were generated and
assigned during the observation phase, which took place
3 months after the data were generated and recorded.
Twenty-two people attending a conference in Tucson,
Arizona, USA, were recruited by convenience and run in
this experiment in an office at the conference hotel. The
participants signed an informed consent, filled out a brief
questionnaire asking about their meditation experience
and belief in phenomena of mind–matter interaction, and
then ran the experiment. After participants completed
their sessions, 22 of the remaining unobserved data files
were subjected to the same analysis as a control.
Individual samples in the control data sets were not
observed at any time.
B. Results
Of the 22 participants, 10 indicated that they had a
regular meditation practice; the remaining 12 were
classified as nonmeditators. As shown in Table IV, the
meditator subgroup supported the hypothesis, with an
effect size of es=-0:80.
VI. FIRST FOUR EXPERIMENTS COMBINED
Data pooled across the 121 sessions collected in the
first four experiments comprised a total of just over
135 000 Rsamples, one per second. Of those sessions, 67
were contributed by meditators, 41 by nonmeditators,
and 13 by meditators using a second double-slit apparatus
in a remote location. A total of 149 control sessions were
also conducted; combined, the control database consisted
of approximately 175 000 Rsamples.
The principal hypothesis predicted that overall z
would be negative, indicating a differential drop in the
spectral ratio Rduring focused-attention periods relative
to no-attention periods. A secondary hypothesis was that
the differential drop would be stronger for meditators
than for nonmeditators, and a tertiary hypothesis was
that the effect would become more negative when lagged a
few seconds. All of the hypotheses were confirmed, with
the principal and secondary predictions confirmed to a
significant degree (see Fig. 8 and Table V). Figure 9
compares effect sizes obtained in the four experiments
and for all data combined.
VII. EXPERIMENT 5
The first four experiments appeared to confirm the
consciousness collapse hypothesis, so a fifth study was
designed both as a formal replication and as a means of
exploring two new factors: (a) An electrocortical marker
of attention was recorded to see whether an objective
correlate of attention would be associated with declines in
R, and (b) the interference-pattern peaks and troughs
TABLE IV. Summary of results from experiment 4. (See Table I caption
for explanations.)
Number es z 0 lag p0 lag z-2 lag p-2 lag
All sessions 22 -0:13 -0:59 0.28 -0:82 0.21
Meditators 10 -0:80 -2:53 0.006 -2:61 0.005
Nonmeditators 12 -0.50 -1.74 0.96 -1.51 0.93
Control 22 -0:13 -0:62 0.27 -0:61 0.27
164 Phys. Essays 25, 2 (2012)
were measured to provide secondary measures of the
predicted effect. In addition, to eliminate the possibility
that results observed in the initial experiments might have
been due to differential vibrations associated with the
computer’s spoken instructions, the computer’s automat-
ed condition assignments were presented over head-
phones. A total of 50 sessions were preplanned.
A. Method
1. Apparatus
The software used to acquire data and control the
experiment was rewritten in Matlab (version 2009b,
MathWorks, Natick, MA, USA). This improved the
efficiency of the data-acquisition process, which in turn
enabled all of the 3000-pixel interference patterns
generated during a test session to be captured and stored
at 20 camera frames/s. In addition, an electrocortical
marker of attention was collected via one electroenceph-
alograph (EEG) channel, which collected data at 512
samples/s. The latter was accomplished using a wireless
EEG system which also provided stereo audio channels
via headphones (Mindset, NeuroSky, San Jose, CA,
USA). This EEG was adapted for our purposes by
enhancing its electrode system to provide improved
contact with the skin and by positioning its sensor to a
left occipital site to avoid frontal-muscle artifacts. The
measurement of interest was a neural correlate of focused
attention: aband event-related desynchronization (ERD;
arefers to the brainwave rhythm 8–12 Hz), which can be
detected most readily over the brain’s occipital lobe.
26
The idea was to test whether this objective measure of
shifts in attention would correlate with changes in the
interference pattern.
2. Procedure
All test sessions were conducted in the IONS
laboratory. The counterbalancing scheme alternated
between attention-toward and attention-away epochs,
each lasting for 20 s and repeated 20 times, for a total of
TABLE V. Summary of results combined across sessions in experiments
1–4. (See Table I caption for explanations.)
Number es z 0 lag p0 lag z-2 lag p-2 lag
All sessions 121 -0:20 -2:17 0.015 -2:34 0.0097
Meditators 67 -0:46 -3:80 0.00007 -3:82 0.00007
Nonmeditators 41 -0.19 -1.25 0.89 -1.01 0.84
Control 149 -0:03 -0:39 0.35 -0:38 0.35
FIG. 8. (Color online) Combined results (weighted Stouffer z) from 121
experimental sessions and 149 control sessions pooled across the first
four experiments. ‘‘Med’’ refers to meditators, ‘‘non med’’ to non-
meditators.
FIG. 9. (Color online) Comparison of effect sizes for the first four experiments and controls. The meditator effect size shown for experiment 2 does not
include the Zen Buddhist meditators, as that portion of the study was conducted in a rural location with uncontrolled environmental conditions and
erratic electrical-line power.
Phys. Essays 25, 2 (2012) 165
40 alternating epochs. Each session began after the
collection of between 20 and 100 s of baseline data. To
ensure that participants understood the nature of the
task, prior to each session with a new participant an
animated video of the double-slit experiment was shown,
followed by a discussion of the task using a drawing of
the apparatus to illustrate how the double-slit device
worked and where the slits were located in the actual
device. As much time as necessary was spent with the
participants to ensure that they understood the nature of
the task.
Audio performance feedback was provided by
associating the playback volume of a Sanskrit chant to
variations in the spectral-ratio value R. To do this, the
average Rvalue measured during the last 5 s of the
previous attention-away epoch was used as a baseline
measure, and then the value of Rrecorded during the
attention-toward epoch was compared in real time to that
value. To avoid sudden jumps in volume, the volume-
adjustment scheme was based on variations in Rover a 3 s
sliding window. Through this method, during attention-
toward periods, the chant played more loudly as R
declined. During the attention-away periods, the chant
played at a continuous, low-level volume and was not
coupled to variations in R. The attention assignments
were announced by the computer via prerecorded audio
files. Differences in Rby attention condition were
analyzed as in the earlier experiments; the hypothesis
predicted a negative zscore.
3. Exploratory measure: Peaks and troughs
The average heights of the interference-pattern
troughs and peaks were determined for each camera
frame (20 frames/s). To accomplish this, troughs and
peaks were determined between pixels 600 and 2400 (see
Fig. 10), and then average differences in these values
between the two attention conditions were evaluated
using a permutation technique similar to that used to
assess changes in R.
4. Exploratory measure: Event-related
desynchronization (ERD)
To determine the correlation between the electrocor-
tical marker of attention and changes in the interference
pattern, for each second of EEG data the spectral power
spectrum at 11 Hz (roughly the middle of the aband) was
determined, and then the average Rvalue over the same
second was calculated. Data for one session thus
consisted of 800 pairs of aand R, one pair per second
(40 epochs ·20 s=epoch). A Pearson linear correlation
between these pairs was determined and then evaluated
using a randomized permutation technique, resulting in a
zscore. These scores were then combined over all
sessions.
5. Hypotheses
As in earlier experiments, the principal hypothesis
predicted a decline in R. A corollary hypothesis was that
the average trough height would rise and the average peak
height would drop. A secondary hypothesis predicted that
the correlation between aand zwould be positive,
because as attention increased, both apower and R
should decrease, and vice versa.
B. Results
Thirty-one people contributed 51 sessions. The first
50 preplanned sessions, consisting of 858 000 frames of
interference-pattern data, are considered here. Fifty
control sessions were also conducted. Figure 11 and
Table VI indicate that the principal hypothesis was
supported with a significant drop in R(z=-4:48) and
the average peak height (z=-2:87) , and a significant rise
in the average trough height (z=3:03). Effect-size magni-
tudes were in alignment with the meditator effect sizes
observed in the first four experiments.
The experimental effect size obtained in this study
was about 3 times as large as that observed in the first
four experiments (-0:63 vs. -0:20). This raises the
question of whether the counterbalancing scheme of
alternating toward and away used in this study might
have produced an artifact. That is, if Rincreased on
average over the course of a session, then the comparison
½mean Rtowards-½mean Raway would invariably produce
negative values and thus spurious support for the
hypothesis.
To test this possibility, the Rvalues recorded in each
sessionwereconvertedintodetrendedresidualsby
determining the best-fit linear trend to the data and
subtracting that line from the original data. Then the
residuals were analyzed and the zscores associated with
the original data were compared to the same scores from
FIG. 10. (Color online) Interference pattern with peaks and troughs
identified. Statistical differences in the average heights, compared
between the two attention conditions, were evaluated using a nonpara-
metric permutation technique similar to that used for the spectral ratio R.
166 Phys. Essays 25, 2 (2012)
the detrended data. The same process was performed for
the control runs. If the original experimental results were
due to systematic positive drifts in R, then there should be
no relationship between the two sets of zscores.
The result was a strong positive correlation for both
the experimental data (linear correlation r=0:83,
p=7·10-14; see Fig. 12) and the control data (r=0:81,
p=6·10-13), indicating that the original results were not
due to positive linear drifts. This outcome confirms the
analysis of the nonlinear residuals in experiment 1,
suggesting that the result of this experiment was not due
to mundane drifts in the data.
In preparation for conducting the EEG event-related
desynchronization analysis, the first step was to examine
the EEG spectrum from 1 to 50 Hz, second by second, in
the attention-toward versus attention-away periods.
Differences in spectral power were determined using a
randomized permutation technique, which resulted in a z
score indicating the differential change in apower. The
result, shown in Fig. 13, is the expected ERD effect—a
decline in spectral power, especially within the aband.
This confirms that ERD can be used as a simple
electrocortical marker for shifts in attention.
The next step was to calculate the ERD and Rvalues
in each session and then the Pearson linear correlation
across the 50 pairs of data. The magnitude of the resulting
correlation was modestly positive, as predicted (mean
r=0:014, combined zðrÞ=1:4, p=0:09, one-tailed). Of the
50 test sessions, nine were independently associated, with
zðrÞ>1:65 (p=0:05, one-tailed), which by the exact
binominal test is associated with p=0:0008. This finding
prompted a post hoc examination of the EEG signals in
each session, and that in turn identified 19 of the 50
sessions with exceptionally noisy signals, possibly due to
intermittent electrode contact with the scalp. Analysis of
the apower versus Rcorrelation in the remaining 31
sessions resulted in a combined zðrÞ=2:7(p=0:004, mean
r=0:027). This finding supports the prediction that an
objective measure of shifts in attention would be
positively correlated with changes in R, but due to the
post hoc nature of this analysis, prudence is warranted.
VIII. EXPERIMENT 6
This study investigated relationships between partic-
ipants’ personalities and beliefs and their performance on
the double-slit task. A series of 50 sessions was
preplanned with 50 different participants selected to
represent a broad range of personality traits, meditation
experience, and beliefs. All participants listened to the
condition-assignment announcements over headphones.
A. Method
Participants filled out a questionnaire asking about
their belief in psychic phenomena, years of meditation, or
other attention-training experience, and the Tellegen
Absorption Scale, a 34-item questionnaire that measures
the degree to which one becomes immersed in a task while
focusing.
27
The belief question was employed because
experiments conducted since the 1940s have shown that
openness to the possibility of extrasensory perception is a
reliable predictor of performance in this type of task.
28
The first half of the sessions in this experiment were
conducted in a hotel room at a conference where it was
convenient to recruit participants with meditation expe-
rience; the remaining sessions were conducted in the
IONS laboratory. Both the original and the secondary
FIG. 11. (Color online) Effect sizes and one standard error of the mean
error bars for the spectral ratio Rand average trough and peak heights,
for experimental and control data in experiment 5.
TABLE VI. Summary of results from experiment 5. (See Table I caption
for explanations.)
Number es z 0 lag
Ratio R
Experiment 50 -0:63 -4:48
Control 50 -0.02 -0.13
Trough
Experiment 50 -0.43 -3.03
Control 50 -0.20 -1.42
Peak
Experiment 50 -0:41 -2:87
Control 50 -0:06 -0:41
FIG. 12. The solid line shows a linear correlation between the original
and detrended experimental data (normalized Rvalues, black dots); the
dashed line shows the same for control data (white circles).
Phys. Essays 25, 2 (2012) 167
double-slit system were used, to allow two individuals to
conduct the study at the same time. In addition, the task
instructions were prescripted and read aloud to each
participant prior to each session in an attempt to create a
more uniform introduction to the task.
There were three other procedural changes from the
earlier experiments: (1) To reduce the time required to
conduct a session, we did not show the video animation of
a double-slit experiment; (2) we used varying-length
condition epochs to prevent participants from anticipat-
ing when each attention assignment would begin and end
(each epoch was assigned a random length from 20 to 30
s); and (3) each session began and ended with a 4 min
silent period, during which data continued to be collected.
The principal hypothesis predicted that Rwould drop
in attention-toward epochs as compared to attention-
away epochs; a corollary was that the average trough
height would rise and the average peak height would
drop. A secondary hypothesis predicted that correlations
between zper session and participants’ belief in psychic
phenomena, capacity for absorption, and years of
meditative experience would all be negative.
B. Results
Fifty participants contributed a total of 1:5·106
frames of interference data. The effect size associated with
Rwas in alignment with the hypothesis, with es=-0:17,
z=-1:21; for the 50 control runs the results were in the
opposite direction, with es=0:20, z=1:43. Effect sizes for
the trough and peak measurements were somewhat
stronger, with trough es=0:25, z=1:75, an d peak
es=-0:27, z=-1:93. The secondary hypothesis was
significantly supported for the correlations between R
and belief in psychic phenomena (r=-0:27, p=0:03, one-
tailed) and between Rand capacity for absorption
(r=-0:21, p=0:07, one-tailed), but not for the correlation
between Rand years of meditation experience, which
instead resulted in a positive correlation (r=0:10, p=0:75,
one-tailed). As a post hoc examination, given that the
average peak measurement produced a stronger result
than R, the correlation between years of meditation and
the peak measurement was examined. This resulted in
r=-0:20, p=0:08, more in alignment with what was
observed in previous experiments. Again, since this was a
post hoc measure, caution in interpreting this outcome is
warranted.
IX. DISCUSSION
Six experiments testing a consciousness collapse
hypothesis led to a combined 4.4-sigma effect in the
predicted direction (p=6·10-6). Control sessions provid-
ed no evidence of procedural or analytical artifacts that
might have been responsible for these effects (z=0:43,
p=0:67). Additional investigations examining the possi-
bility that results were due to heat generated by proximity
of the body, or sound vibrations associated with
announcements of the condition assignments or perfor-
mance feedback, or systematic drifts or oscillations, also
failed to identify viable artifacts.
Figure 14 summarizes the DReffect sizes in the
experiments, Fig. 15 shows a cumulative zscore for the
same data, and Fig. 16 shows a similar analysis for
meditator versus nonmeditator sessions. Twenty-nine
sessions collected in the last two experiments were not
part of the preplanned design (labeled ‘‘extra’’ in Fig. 15),
but are included in the final analysis for completeness.
The average effect size per session over the first four
experiments was es=-0:20. A power analysis based on
that effect size would predict a 63% chance to achieve an
outcome at p=0:05 (one-tailed) or better across 100
sessions. The results observed in the 100 sessions of
experiments 5 and 6 surpassed this prediction, mainly due
to the stronger effect size observed in experiment 5. That
outcome may have reflected our intention to optimize
results by (a) conducting all sessions in the controlled
laboratory environment, (b) implicitly reminding partic-
ipants through use of an EEG monitoring system that
paying attention during the session was important, (c)
ensuring that those participants understood what was
expected of them by showing an animation of a double-
slit experiment and then spending as much time as was
necessary to discuss the nature of the task, and (d)
preferring participants with meditation experience and
openness to the nature of the task. By contrast, in
experiment 6 we recruited people with a broad range of a
priori beliefs and meditation experience, and we read
instructions from a fixed script to decrease the time
required to conduct each session.
A. Potential environmental factor
A factor that was possibly responsible for some of the
performance variance observed across test sessions was
nanotesla-scale fluctuations in the Earth’s geomagnetic
field (GMF). This variable has been shown to be a
significant factor in many areas of human performance,
FIG. 13. (Color online) Drop in EEG spectral power when comparing
attention-toward versus attention-away conditions across all sessions, in
terms of standard normal deviates. Notice the particularly strong drop
around the aband (8–12 Hz).
168 Phys. Essays 25, 2 (2012)
including stock-market behavior,
29
airplane crashes,
30
suicides,
31
cardiac health,
32
and—of special relevance to
the present studies—a greater frequency of reported
spontaneous psychic experiences as well as enhanced
performance in controlled extrasensory-perception
tasks.
33
To explore whether this factor might have
influenced performance in the present experiments, the
GMF ap index (a measure of global geomagnetic activity)
was retrieved for the day of each test session.
34
Based on
previous GMF correlation studies, the prediction was that
during days of quiet GMF, performance on the double-
slit task would increase (indicated by a decrease in R)as
compared to days with noisier GMF.
Figure 17 shows the results in terms of combined z
score for sessions conducted on the Nquietest and N
stormiest days, with Nrepresenting the GMF contrast of
interest. For example, a contrast of N=7 refers to all
sessions conducted during days with the seven largest and
seven smallest ap values and to combine the zscores
associated with changes in Robtained in those sessions.
The figure indicates that for high GMF contrasts there
was, as predicted, a strong difference in experimental
performance, with better results on lower-ap days than on
higher-ap days. For N=7, the mean natural log of GMF
on the quietest days was lnðapÞ=0:49, and the session z
score combined across those days was z=-2:3. By
contrast, the mean for the seven noisiest-GMF days was
ln ðapÞ=3:30, and the combined result for sessions run on
those days was z=1:97. The statistical difference between
sessions conducted on high- and low-GMF days was
z=-3:0. Figure 17 indicates that the direction of this
difference was maintained regardless of the magnitude of
the contrast, suggesting that this performance difference
was indeed associated with GMF flux. Of greater
importance, it indicates that a variable known to correlate
with performance in similar ‘‘extrasensory’’ tasks was also
observed in the present experiments.
B. These studies in context
Because it is central to interpretations of quantum
mechanics, the physics literature abounds with philo-
FIG. 14. (Color online) Effect sizes and one standard error of the mean
error bars in all experiments and controls, and across 29 extra sessions
that were not part of the preplanned designs of experiments 5 and 6
(labeled ‘‘Extra’’).
FIG. 15. Cumulative zscore for all 250 experimental and control
sessions.
FIG. 16. Cumulative zscore for meditators and nonmeditators across all
six experiments.
FIG. 17. (Color online) Examination of possible daily influence of the
geomagnetic field on session performance across the six experiments. A
contrast size of, say, 7 refers to sessions conducted on days with the
seven largest and seven smallest ln ðapÞvalues (natural log of daily ap
index); these values are indicated along the xaxis by the upper dashed
lines labeled ‘‘high ln ðapÞ’’ and ‘‘low ln ðapÞ.’’ The combined experi-
mental zscores for sessions run on those dates are shown in the bottom
dotted lines.
Phys. Essays 25, 2 (2012) 169
sophical and theoretical discussions about the QMP,
including speculations about the role of consciousness.
One might expect to find a correspondingly robust
experimental literature testing these ideas, but it is not
so, and the reason is not surprising: The notion that
consciousness may be related to the formation of physical
reality has come to be associated more with medieval
magic and so-called New Age ideas than it is with sober
science. As a result, it is safer for one’s scientific career to
avoid associating with such dubious topics and subse-
quently rare to find experiments examining these ideas in
the physics literature. Indeed, the taboo is so robust that
until recently it had extended to any test of the
foundations of quantum theory. For more than 50 years
such studies were considered unsuitable for serious
investigators.
35
As Freire noted in discussing the history
of tests of Bell’s theorem,
Some of the physicists who decided not to hire
Clauser were influenced by the prejudice that
experiments on hidden variables were not ‘‘real
physics.’’ His former adviser, P. Thaddeus, wrote
letters warning people not to hire Clauser to do
experiments on hidden variables in quantum me-
chanics as it was ‘‘junk science,’’ a view shared by
other potential employers.
35(p.284)
However, this is not to say that the scholarly
literature is mute on this matter. A century-long empirical
literature can be found in the controversial domain of
parapsychology, which focuses on the interface between
mind and matter. Here we find over a thousand peer-
reviewed studies reporting (a) experiments testing the
effects of intention on the statistical behavior of random
events derived from quantum fluctuations,
36,37
(b) studies
involving macroscopic random systems such as tossed
dice and human physiology as the targets of intentional
influence,
38
(c) experiments involving sequential observa-
tions to see whether a second observer could detect if a
quantum event had been observed by a first observer, or if
time-delayed observations would result in similar ef-
fects,
39–41
and (d) experiments investigating conscious
influence on nonliving systems ranging from molecular
bonds in water to the behavior of photons in interferom-
eters.
42
Much of this literature has appeared in discipline-
specific journals, but given the controversial nature of the
topic, it is worth noting that some of it has also appeared
in better-known outlets including the British Journal of
Psychology,
38
Science,
43
Nature,
44
Proceedings of the
IEEE,
45
Neuroscience Letters,
46
Psychological Bulle-
tin,
47,48
and others. Cumulatively, these experiments
suggest that mind–matter interactions occur in a broad
range of physical target systems. Observed effects tend to
be small in absolute magnitude and are not trivially easy
to repeat on demand, but high variance and concomitant
difficulties in replication are to be expected because all of
these studies necessarily involved focused human atten-
tion or intention. As with any form of human perfor-
mance, the ability to focus attention varies substantially
not just from one person to the next, but within each
individual from day to day and throughout the course of
a single day.
49,50
Variables influencing the ability to
perform mental tasks go beyond simple factors such as
nervous-system arousal and distractions; they include
when one last dined and what was consumed,
51
interac-
tions between personal beliefs and the nature of the
task,
52
the state of the geomagnetic field, and so on. Such
factors conspire to make the mind side of a postulated
mind–matter interaction far more difficult to control than
the matter side. As a result, if one is prepared to take
seriously the proposition that some properties of quan-
tum objects are not completely independent of human
consciousness, then such a study cannot be conducted as a
conventional physics experiment, nor can be it conducted
as a conventional psychology experiment. The former
tends to ignore subjectivity and the latter tends to ignore
objectivity.
In an attempt to accommodate both sides of the
proposed relationship, we designed a physical system with
interference fringes as stable as possible, and we also
cultivated a comfortable test setting, encouraged a sense
of openness to the idea of extended forms of conscious-
ness, selected participants with practice in focusing their
attention, and spent generous amounts of time discussing
the nature of the task with the participants. The superior
results observed with meditators suggest that in spite of
unavoidable performance variance, it may be possible in
future studies to identify those aspects of attention and
intention that are most important in producing the
hypothesized effect.
It should be noted that some meditation styles, such
as mantra repetition, tend to train for focused or
concentrated attention, while others, such as mindfulness
meditation, tend to expand one’s attentional capacities.
53
No attempt was made in the present studies to assess
differences among reported meditation styles, or to
independently measure participants’ capacity for sustain-
ing focused attention. However, it is not unreasonable to
expect that future studies might find that different
meditative styles lead to different outcomes. In addition,
measuring participants’ capacity for sustaining attention,
investigating other brain and behavioral correlates of
performance, using single-photon designs, and developing
more refined analytical procedures would all be useful
directions to pursue.
In sum, the results of the present experiments appear
to be consistent with a consciousness-related interpreta-
tion of the QMP. Given the ontological and epistemo-
logical challenges presented by such an interpretation,
more research will be required to confirm, systematically
replicate, and extend these findings.
ACKNOWLEDGMENTS
These experiments were supported by grants from the
Bial Foundation, the Fetzer-Franklin Fund of the John E.
Fetzer Memorial Trust, the Mental Insight Foundation,
the Federico and Elvia Faggin Foundation, Inc., Dick
170 Phys. Essays 25, 2 (2012)
and Connie Adams, and the HESA Institute. Valuable
comments on a draft of this paper were offered by James
Johnston, George Weissmann, Richard Shoup, York
Dobyns, Roger Nelson, Nick Herbert, and two anony-
mous referees. We thank Dr. Robert Stek and Tre Tresa
Surbeck for recruiting and running participants in
experiment 4.
1
R. Feynman, R. Leighton, and M. Sands, The Feynman Lectures on
Physics (Addison-Wesley, New York, 1965).
2
S. Gr ¨
oblacher, T. Paterek, R. Kaltenbaek, C. Brukner, M. Zukowski,
M. Aspelmeyer, and A. Zeilinger, Nature 446, 871 (2007).
3
R. Jahn and B. Dunne, Found. Phys. 16, 721 (1986).
4
K. Wilber and F. M. Stein, Am. J. Phys. 53, 601 (1985).
5
E. Wigner, Am. J. Phys. 31, 6 (1963).
6
E. Wigner, The Monist 48, 248 (1964).
7
M. Mermin, Boojums All the Way Through: Communicating Science in
a Prosaic Age (Cambridge University Press, Cambridge, UK, 1990).
8
B. d’Espagnat, Sci. Am. 241 (5), 158 (1979).
9
J. von Neumann, Mathematical Foundations of Quantum Mechanics,
translated by Robert T. Beyer (Princeton University Press, Princeton,
NJ, 1955).
10
H. Stapp, Mindful Universe: Quantum Mechanics and the Participating
Observer (Springer, New York, 2007).
11
E. J. Squires, European J. Phys. 8, 171 (1987).
12
S. Goldstein, Phys. Today 51 (4), 38 (1998).
13
C. Fuchs and A. Peres, Phys. Today 53 (3), 70 (2000).
14
A. Zeilinger, Rev. Mod. Phys. 71, S288 (1999).
15
B. d’Espagnat, Found. Phys. 35, 1943 (2005).
16
B. Rosenblum and F. Kuttner, Quantum Enigma: Physics Encounters
Consciousness (Oxford University Press, Oxford, UK, 2006).
17
B. Rosenblum and F. Kuttner, Found. Phys. 32, 1273 (2002).
18
F. Crick, The Astonishing Hypothesis: The Scientific Search for the Soul
(Touchstone, New York, 1994).
19
W. Tittel, J. Brendel, B. Gisin, T. Herzog, H. Zbinden, and N. Gisin,
Phys. Rev. A 57, 3229 (1998).
20
D. J. Chalmers, The Conscious Mind: In Search of a Fundamental
Theory (Oxford University Press, New York, 1996).
21
M. Ibison and S. Jeffers, J. Sci. Explor. 12, 543 (1998).
22
D. Radin, Explore 4, 25 (2008).
23
F. D. Richard, C. F. Bond, Jr., and J. J. Stokes-Zoota, Rev. Gen.
Psychol. 7, 331 (2003).
24
J. Cohen, Statistical Power Analysis for the Behavioral Sciences, 2nd
ed. (Erlbaum, Hillsdale, NJ, 1988).
25
J. Houtkooper, J. Sci. Explor. 16, 171 (2002).
26
W. van Winsum, J. Sergeant, and R. Geuze, Electroencephalogr. Clin.
Neurophysiol. 58, 519 (1984).
27
A. Tellegen and G. Atkinson, J. Abnorm. Psychol. 83, 268 (1974).
28
G. R. Schmeidler and G. Murphy, J. Exp. Psychol. 36, 271 (1946).
29
See Federal Reserve Bank of Atlanta Working Paper 2003-5b (A.
Kriveloyova and C. Robotti, 2003), available at http://www.frbatlanta.
org/filelegacydocs/wp0305b.pdf (Accessed March 8, 2012).
30
N. M. Fournier and M. A. Persinger, Percept. Mot. Skills 98, 1219
(2004).
31
C. Gordon and M. Berk, South African Psychiatry Review 6(3), 24
(2003).
32
S. Dimitrova, I. Stoilova, T. Yanev, and I. Cholakov, Arch. Environ.
Health 59, 84 (2004).
33
G. B. Schaut and M. A. Persinger, Percept. Mot. Skills 61, 412 (1985).
34
See http://www.ngdc.noaa.gov/stp/geomag/kp_ap.html for more in-
formation about the ap index (Accessed March 8, 2012).
35
O. Freire, Jr., Stud. Hist. Phil. Sci. B 40, 280 (2009).
36
H. B ¨
osch, F. Steinkamp and E. Boller, Psychol. Bull. 132, 497 (2006).
37
D. Radin and R. Nelson, Found. Phys. 19, 1499 (1989).
38
S. Schmidt, R. Schneider, J. Utts, and H. Walach, Br. J. Psychol. 95,
235 (2004).
39
H. Schmidt, J. Parapsychol. 45, 87 (1981).
40
D. J. Bierman, Mind and Matter 1, 45 (2003).
41
D. J. Bierman and S. Whitmarsh, 27, The Emerging Physics of
Consciousness, edited by Jack A. Tuszynski (Springer, New York,
2006).
42
D. Radin, Entangled Minds (Simon & Schuster, New York, 2006).
43
T. Duane and T. Behrendt, Science 150, 367 (1965).
44
R. Targ and H. Puthoff, Nature 252, 602 (1974).
45
R. G. Jahn, Proc. IEEE 70, 136 (1982).
46
J. Wackermann, C. Seiter, H. Keibel, and H. Walach, Neurosci. Lett.
336, 60 (2003).
47
E. Girden, Psychol. Bull. 59, 353 (1962).
48
D. Radin, R. Nelson, Y. Dobyns, and J. Houtkooper, Psychol. Bull.
132, 529 (2006).
49
H. Gritton, B. Sutton, V. Martinez, M. Sarter, and T. Lee, Behav.
Neurosci. 123, 937 (2009).
50
P. Valdez, C. Ram ´
ırez, A. Garc ´
ıa, J. Talamantes, and J. Cortez,
Chronobiol. Int. 27, 393 (2010).
51
D. Benton, D. Owens, and P. Parker, Neuropsychologia 32, 595
(1994).
52
K. Barton, J. Fugelsang, and D. Smilek, Thinking & Reasoning 15, 250
(2009).
53
F. Travis and J. Shear, Conscious. Cogn. 19, 1110 (2010).
Phys. Essays 25, 2 (2012) 171
