Here are some email exchanges that addresse some of the subtleties of the
anomalous correlations we see in the GCP database. We don't know much
about the mechanism or the sources of the effects, but it is clear that
it is not as simple as a single source, even though most of the talk and
writing tends to express this assumption. The following is a note
from Nova Spivack, asking thoughtful questions about these issues,
with my responses
interleaved. After that exchange is one with Prof. Johannes Hagel
that focuses more specifically on
the "experimenter" effect.
No real answers, but an attempt to become more clear about
what we are looking at.
----- Original Message Nova Spivack -----
Sent: Saturday, April 21, 2001 10:46 AM
Subject: Re: Hi, a few questions....
On Fri, 20 Apr 2001, Nova wrote:
> Hi, I came across your research several years ago, and today ran into it
> again. I have a formal background in cognitive science, CS and philosophy of
> mind---in general my academic focus was on consciousess and artificial
> intelligence. In addition, I have been studying Tibetan Buddhist meditation
> for around 16 years. I am in Princeton occaisionally---I advise Sarnoff Labs
> on Internet strategy. I would enjoy stopping by and seeing your lab, meeting
> your team, and perhaps participating in your studies. In addition, I am very
> close to several extremely senior Tibetan yogis with very advanced
> concentration and other abilities---most of them quite old now---who might
> be willing to participate as well. If you are interested, we could discuss
> that later.
Thanks for your interest. There seems to be lots we might
wish to discuss, so a visit when you are in Princeton would
be a good idea. I am at the lab usually on Mon - Wed, and
it will be best to coordinate with the rest of the group in
any case, so I will copy this note to Arnold Lettieri, who
keeps us organized with regard to visits. I should also
mention that the Global Consciousness Project which inspires
some of your questions is independent of the PEAR lab.
I'll give some brief responses to your questions here, but
many of them deserve actual discussion.
> Anyway, a few questions/ideas...
> - Have you ever experimented to see whether there are any "relativistic
> effects" on your random number generators? For example, are there any
> observable changes in the statistics when a generator is moving "fast"?
No. Interesting, but I think not feasible to test at this point.
> - Have you tested the effects of light on your generators? For example, if
> one were to illuminate a generator with various frequencies of light vs.
> total darkness, would that have any observable effects on the output
No. The REGs are enclosed for mechanical and EM shielding,
so any ordinary effect of light would be precluded. We have
run one experiment with a bare REG board, hence with
ordinary broad-spectrum light imposed, but it did not
show differentiable responses.
> - Have you tested the sensitivity of number generators under conditions of
> extreme cooling or absense of thermal noise? It would be interesting to use
> some sort of a super-sensitive superconducting device as the core of a
> random number generator. Perhaps that would yield more sensitive results?
Not extreme cooling, only the possible range of ambient
temperatures in which they might actually be used. Though
the possibility of a more sensitive detector might be based
on differences (e.g., superconduction) in the physics, the
range of qualitatively different random sources we have used
over the years appears not to matter to the size of
anomalous effects. We think this means that the important
conditions are more in the subjective realm, and that
information is a more likely "agent" than energy.
> - Have you tested for any differences that might take place when a generator
> is in a vacuum vs. not in a vacuum?
> - How can you be sure that observed statistical anomalies are causally
> related to particular current events and not merely coincident? Given that
> there are significant "global" or "mass consciousness" events almost every
> day if not more frequently than that, there should be anomalies on a nearly
> constant basis. If this is observed, how does it correlate with the
> ocurrence of "major" news stories over some period of time?
I cannot be sure. What I think makes sense is to continue
to gather data testing predictions that imply the linkage we
hypothesize, while also thinking creatively about tests that
may exclude some of the alternative sources. In the end, I
think we will progressively come to understand that we are
looking at a complex, broadly integrated system. Addressing
your more specific question, we can do "resampling" of the
data stream outside the defined period of the global event,
but using the parameters of the focused analysis. If the
results on average do not show the same deviation, we have a
prima facie case that the global event was at least meaningfully,
if not causally related to the anomaly. (Note that other
sources such as an "experimenter" effect are not excluded by
such statistical backgrounding.)
> - Could we do a more targeted experiment?---For example, let's situate a
> random number generator in a public company headquarters on or next to the
> desk of of the CEO. Is there any correlation with the statitical output and
> changes in the company stock price or volume, or other company events? What
> if we did an experiment where we had several such random number generators
> placed in different public company CEOs' offices, thus expaning our sample:
> would we find that each random number generator was correlated with the
> major events of its particular company and not the others? We could then do
> various experiments---start with nobody knowing about the device, then just
> informing the CEO of the presence of the device, then add the senior
> management, finally add the entire company: is there an observable
> difference in statistical properties as more of the collective consciousness
> of the company is made aware of the device?
We have in the PEAR lab's large FieldREG database quite a
number of targeted experiments, some of which are analogous
to what you propose here. Because we have found that the
category of situations producing an anomalous response tend
to share a numinous quality, I would not predict the CEO's
office and the concerns with business success to have an
effect. However, that is an empirical question. Would be
possible to do with appropriate support.
> - The previous idea brings up the question of whether there is a difference
> between the output of a random number generator that is "known" vs. one that
> is "unknown." Essentially we would do an experiment in which we compared the
> output of a generator before anyone was informed of its existence and after
> it was made known to someone or some group. This would also require that the
> experimenters create a double blind situation where even they do not know
> which generators involved in the experiment. This could be accomplished by
> randomly turning on some of a set of possible generators at a random time,
> or for random intervals during a period of time. Nobody would know which
> generators were involved and whether or not they were "on" at any given
> time. Compare the results of this situation to a later experiment in which
> the identities and status of generators in the experiement are known to the
> experimenters. Is there any change in output?
I don't think there is likely to be a difference, given the
experience of the GCP. It is difficult to say, though,
because the experimenters always are in there. Even with
the double blind setup you envision, there is nothing to
preclude the desires and expectations of those who have an
interest from "invading" the data. We have a little
experience with blinding, and it is not easy to parse into
the neat categories we all are used to from ordinary
physical, biological, medical research. Consciounsess (as
your Tibeten friends would likely say) is remarkably
elusive, while it seems also to be virtually all-pervasive.
PS: Love the name of your company. > Lucid Ventures
-------------- Second exchange, Nova Spivack
Date: Wed, 25 Apr 2001 07:21:51 -0400 (EDT)
Subject: Re: Hi, a few questions....
On Mon, 23 Apr 2001, Rigpa wrote:
> Thanks Roger, I appreciate the response. I have spent a great deal of time
> studying consciousness from a variety of perspectives. You guys are doing
> some of the most important/interesting work in the field.
> Where can I find some more detailed online summaries and analyses of your
> experiemental results? There seemed to be limited resources available from
> your site.
Don't know if you are referring to PEAR's site or the GCP's.
The PEAR site, http://www.princeton.edu/~pear has abstracts
for about 40 papers, and pdf files for half a dozen of the
more comprehensive or otherwise useful papers. The GCP site
http://teilhard.global-mind.org/ has everything there is
concerning the experimental results, from the raw data to
individual event investigations to grand summaries. What are
you looking for?
> Thanks for the compliment on my company name. I invest in
> futuristic early-stage technologies with important global
> implications (room-temp. superconducting, zero-point energy
> devices, machine learning, new internet technologies. etc.).
> I am interested in what you are up to, although it seems we are
> a ways from any commercial applications. Still, it would be
> interesting to think about this a little...
I suspect we have some potential for educating people to be
more sensible about energy use (because we are interdependent),
but not much for generating electricity (or fueling SUV's).
> Even if there aren't any commercial applications, I am also interested in
> the theoretical implications. In particular, if you can "prove" an effect
> that can only be explained by appealing to consciousness, then that may
> finally convince the scientific materialists that consciousness is worthy of
> scientific study.
Yes. I'm there, although my personal vision is more to help a
broad, popular "knowing" take root, that moves us toward a more
conscious style of living. I think maturing toward Teilhard's
noosphere is possible, but more importantly it is essential.
We're wasting (that is, destroying) so much of the earth's promise,
and we must stop and think. We are so focused on self that we
are tearing up our home for the sake of "gratifications" that are
not even satisfying.
> One question: Since you can't eliminate the "experimenter" effect on your
> experiments, how can you effectively create control-group scenario? Without
> a way to establish a real control scenario that cannot be influenced, how
> can you be sure that you are measuring what you think you are measuring?
It is a good question that has been around forever, but
because it is difficult, it's often not discussed. On the
GCP site there is some discussion here and there, but no
resolution. Here is a bit from the FAQ: "The source of anomalous
effects probably has several dimensions, including the "experimenters",
and geographically local groups as well as dispersed interest
groups. The good news is that the GCP database offers some
potential for answering this question. The issues are subtle,
but there are sensible criteria such as correlation of effect
size with numbers of people potentially involved, and timing of
knowledge about the event, that may help distinguish
between experimenter effects and local versus general
influences. The bad news is that we still have
only a tiny statistical effect, so that it is always hard to
distinguish signal from noise. This means that
every "success" might be largely driven by chance, and every
"null" might include a real signal overwhelmed by noise.
In the long run, the signal can be seen, of course, but to separate
the signal into its components (global consciousness, experimenter
effects, local group influence, and so on) will require a still
longer run as well as well-thought-out analysis strategies."
> Consider this: since the experimenters always know about the experiment,
> they are influencing it (assuming that knowing = influencing). In this base
> condition, is the distribution random? It seems to me that, according to
> your recent work on network REGs, there are always deviations from
> randomness over time, so even the base case is not random.
I favor the theoretical distribution as the source for expectation,
partly for this reason. All the REGs we use have a logical
stage that guarantees a mean deviation of zero over the long
run, and lots of design to ensure they get there randomly.
But we are fully cognizant that even "calibrations" are
subject to experimenter effects -- so we simply intend that
they be good calibrations, meaning they should tell us
whether the REG is performing as a proper random event
generator. It's a loop, but an interesting one to look out
of -- or into.
> Is this due to the experimenters knowing about the experiment? Now you add
> a set of subjects and have them further influence the REG--yielding further
> deviation from randomness. So the effect should be additive. Does adding more
> knowers = less randomness, is there a limit to the additive effects?
Clearly yes, assuming any of the effects we see are "real".
In FieldREG experiments, groups of half a dozen produce the
same effect sizes as groups of 1000. And the GCP takes this
a step further. Most of the effect sizes there are clearly
in the same range of magnitude (based on a sensible
comparison standard like how long it takes to develop the
> Perhaps, if left on their own, REG devices will inevitably behave
> non-randomly over time regardless of whether anyone is specifically knowing
> them or not---as your network experiments seem to indicate---not because of
> any influence from conscious intent, but rather because that is the
> underlying nature of the universe. What if your REG devices simply measure
> some strange chaotic property of quantum mechanics...something fundamentally
> "non-random" about our universe. Perhaps the fact that our universe doesn't
> behave randomly when put to the test is the ultimate reason it is here in
> the first place...maybe it started with a slight deviation from perfect
> randomness and that deviation self-amplified itself fractally into all sorts
> of complexity and dimensionality...perhaps your REG device is simply
> detecting the echoes of this fundamental "background non-randomness"...? Is
> this the mathematical equivalent of detecting "background radiation" from
> the early days of the Big Bang?
Interesting to think about. The logic stage of the REG
implementation ensures that (all else being equal) any residual
long- term strangeness in the data should be informative indeed.
Sorting out the meaning from the information is the hard part.
But, hey, there's nothing like a good puzzle to keep us on the
chaotic flight that provides an enchanting "butterfly" perspective.
> Another possibility: What if you are measuring quantum correlation between
> observation and reality aka the double slit experiment? Essentially, without
> recourse to consciousness, we might be able to explain what you observe
> simply on an information/measurement level. Does the act of measuring a REG
> over time change its quantum state enough to generate deviations from
> randomness? Perhaps what you've got here is something similar to the
> double-slit experiment.
Very interesting thoughts. I don't have any useful comment,
though I might note that we did run a double slit experiment
in the PEAR lab for a while, with the target being to increase
vs. decrease contrast in the interference fringes. Appeared to
have about the same yield as the rest of our Mind/Machine
experiments. This, and the other systematic exploration of
devices, persuades me that the M/M effect is information-based,
that is, that it does not necessarily touch directly any
physical, material entity. Instead, it touches the structure
of interactions, makes "decisions" where there would only have
been random movement, adds or extracts meaning.
----------- Exchange with Johannes Hagel
Date: Tue, 18 Feb 2003 08:07:54 -0500 (EST)
To: Prof. Dr. Johannes Hagel
Cc: Margot Tschapke ,
Subject: Re: How to distinguish "real" from "experimenter" effects ?
On Sun, 16 Feb 2003, Prof. Dr. Johannes Hagel wrote:
> I have the following question to you: I am sure that the question "war
> or peace in Irak" does touch you emotionally like other big events
> (Sept. 11-th a.s.o). How can you then distinguish in general the PK
> influence of your personal emotions (being political concerns and the
> situation of our world but also the ones of a success of the GCP as a
> project) on the RNG network (what I call experimenter effect) from
> global PK effects ? This distinction seems rather difficult to me
> specially in the case of the protest movements. They (unlikely to the
> September 11 event) have been foreseen and it was clear that they will
> take place. I ask you this question because I am faced to the same type
> of problem concerning our experiments with inanimate matter and I do not
> have a definite answer in the form of a deductive argument that will
> "survive" in a serious discussion.
It is a question that is increasingly being asked, and it is
a good one in many respects. The first point to make is that
the anomalous, nonlocal effect exists, and that is what is
most interesting. Once we are convinced there is a real
effect, then we want to consider mechanism, and the source
of the effect is, of course an important part of mechanism.
Most of the ideas I have about distinguishing experimenter
vs group or global consciousness are logical arguments,
rather than experimentally testable conditions.
0) I think, personally, that the effects are much too
impressive for me or the small group of experimenters engaged
in the GCP to be responsible.
1) Many effects (in the GCP, most of the cases) occur before
experimenters know about them or ask any formal questions.
So any "experimenter effect" would have to be unconscious.
1a) There is a parallel "analyser effect" or "observer
effect" that is similarly posed, but these are variants that
are logically the same as the "experimenter effect."
2) We have some evidence of a decrease in effect size for
eggs that are more distant from the nominal source or the
major focus of interest.
3) Many of the results do not conform to my expectations and
those of my colleagues, and there is adequate statistical
power to be sure this is not a matter of chance fluctuation.
4) Independent assessments of the same data, and applications
of different statistical tools yield essentially similar, and
consistent, confirmatory results.