Discovery Lab Answers

Harold E. Puthoff, PhD is a physicist who co-led the remote-viewing research program at Stanford Research Institute (SRI) with Russell Targ, and whose work spans laboratory psi experiments, government-sponsored intelligence applications, and theoretical physics frameworks for anomalous cognition. His profile is at Harold E. Puthoff, PhD.

Experiments

Puthoff’s experimental record at SRI falls into several distinct lines of work, all conducted collaboratively with Targ.

Remote viewing — the core SRI program. The foundational series, published in Nature in 1974, tested whether individuals could describe remote geographical locations while shielded from ordinary sensory access ^[3]. A larger synthesis of this program appeared in the Proceedings of the IEEE in 1976, which tabulated 51 double-blind remote-viewing experiments involving nine subjects: six formal study participants and three visiting government scientists ^[2].

Neurophysiological signal transmission. An embedded sub-experiment within the SRI program tested whether a remote sender’s flickering visual stimulus could influence the receiver’s EEG alpha activity — an attempt to find a physiological correlate of whatever channel was being studied ^{[2] [3]}.

Precognition and ESP teaching-machine work. Puthoff and Targ also reported experiments using an ESP teaching machine, in which subjects attempted to predict which target would be illuminated before it was selected, with one subject eventually reaching a level where the probability of her scoring pattern under the null hypothesis fell below 10⁻³ per run ^[1]. A precognition series — in which the same subject was graduated from forced-choice to precognitive trials — produced results across 672 trials at p < 0.005 ^[1].

Pavel Stepanek analysis. The 1976 IEEE paper also included an information-theoretic reanalysis of Pavel Stepanek’s long-running card-calling series, applying Shannon entropy measures to calculate the channel capacity of the reported effect ^[2].

Classified government research. Much of the SRI program was conducted under U.S. government sponsorship. A 1980 SRI final report co-authored by Puthoff was produced under the classified “Grill Flame” special access program and has since been declassified.

Uri Geller experiments. The Nature paper also reported separate experiments with Uri Geller, who was isolated inside an electromagnetically shielded room (attenuating RF radiation across 15 kHz to 1 GHz, and magnetic fields by 68 dB at 15 kHz) while attempting to reproduce target drawings selected by personnel blind to the experiment ^[3].

Methodology

The core remote-viewing protocol shared several consistent design features across experiments:

• Sensory shielding of the receiver. Receivers remained at SRI while senders traveled to remote geographical locations, typically several kilometers away, chosen from a target pool unknown to both the receiver and the experimenters present during the session ^[3].
• Double-blind judging. Transcripts of receiver narratives were evaluated by independent judges who visited the actual target locations and rank-ordered the transcripts against them, without knowledge of which transcript corresponded to which location ^[3].
• Sum-of-ranks evaluation. Statistical significance was assessed by a conservative rank-matching procedure. The 1976 IEEE paper explicitly notes that this procedure “vastly underestimates the statistical significance of the individual descriptions” ^[2], because it reduces rich qualitative descriptions to ordinal rankings.
• Target pool design. Targets were selected by laboratory personnel not otherwise associated with the experiment, and both experimenter and subject were kept blind to the target pool contents ^[3].
• Information-theoretic framing. Puthoff and Targ applied Shannon entropy and channel-capacity mathematics to quantify the information content of correct responses — treating remote viewing as a signal-detection and coding problem rather than a simple hit/miss count ^[2].
• Neurophysiological controls. In the EEG sub-experiment, alpha power was measured across 12-trial averages under 0 Hz, 6 Hz, and 16 Hz flash conditions randomly intermixed, with the receiver unaware of which stimulus the sender was observing ^{[2] [3]}.

Puthoff also argued, in the theoretical companion work, that psi phenomena could be framed within information-theoretic models compatible with quantum mechanics and thermodynamic principles — attempting to constrain claims to what existing physics could accommodate rather than positing new forces ^[1].

Data

Because no structured evidence block with audited statistics has been retrieved for this question, the figures below come directly from the numbered primary sources and are reported as the sources state them, without pooling across different metrics.

Remote viewing — the nine-subject, 51-experiment SRI series ^[2]:

Nature paper — nine-location series with Price ^[3]:

Six of nine descriptions were correctly matched by judges. Under the null hypothesis (random permutation without replacement), this outcome carries p = 5.6 × 10⁻⁴ ^[3].

Nature paper — preliminary series (five judges, nine locations) ^[3]:

Twenty-four correct matches were obtained against a chance expectation of five, significant at p = 8 × 10⁻¹⁰ ^[3].

EEG alpha-power sub-experiment ^{[2] [3]}:

At 16 Hz flash stimuli, the receiver’s alpha activity showed −24% average power reduction (p < 0.04) and −28% peak power reduction (p < 0.03) relative to no-flash periods. A similar but smaller reduction was observed at 6 Hz (−12% average power, −21% peak power) ^[2].

Precognition / ESP teaching machine ^[1]:

A subject’s best run produced 40 or more hits out of 96 trials (chance = 24), with p < 10⁻³ per such run. The precognition series ran 672 trials at p < 0.005 overall ^[1].

Stepanek reanalysis ^[2]:

Over 19,350 calls averaging 9 seconds each, Stepanek’s hit rate was 61.9% (11,978 hits, 7,372 misses), with the channel-capacity calculation yielding approximately 0.0053 bits per call — a small but reportedly non-zero information-transfer rate under the information-theoretic model ^[2].

One finding the sources report without ambiguity and that bears on generalizability: among the SRI subjects, some of the best results came from the inexperienced group, which Puthoff and Targ interpreted as supporting the hypothesis that remote viewing may be a widely distributed latent ability rather than a rare talent ^[2].

Skeptical critiques

What critics argue. The Nature paper drew immediate methodological scrutiny. The editors of Nature themselves, in a prefatory note accompanying publication, flagged that they were publishing the paper as a matter of record while expressing reservations about the experimental controls — specifically whether sensory leakage through ordinary channels had been adequately excluded. The Geller experiments in the same paper attracted particular criticism: Marks and Kammann (1980) argued in their book The Psychology of the Psychic that the remote-viewing transcripts contained embedded cues — references to prior sessions, weather, and ordinal information — that could have allowed judges to match transcripts to targets without genuine psi, and that the judging procedure was not adequately blind to these cues. Hyman (1985), in his contribution to the exchange with Honorton over the ganzfeld database, articulated broader objections about multiple endpoints and the reporting of favorable sub-series from larger datasets that did not uniformly reach significance — an objection applicable to the SRI data structure, where the full 51-experiment set and individual sub-series are reported selectively.

What the experimental data show. The Proceedings of the IEEE paper ^[2] explicitly acknowledges that the sum-of-ranks procedure it used is conservative and that sub-series results varied: the learner subjects S5 and S6 did not reach significance. The Nature paper ^[3] reports that Price’s descriptions “contain inaccuracies,” and the paper itself discloses the Faraday-cage shielding parameters for the Geller sessions — leaving open whether all ordinary information channels were controlled. The EEG sub-experiment ^[3] showed effects in the predicted direction at two stimulus frequencies, but the receiver was not explicitly stated to have been blind to the experimental schedule in the retrieved passages.

Analysis. The core methodological dispute — whether the judging procedure was sufficiently blind to extraneous cues in the transcripts — has not been resolved by a published independent replication of the specific SRI double-blind judging protocol by a group outside SRI. The information-theoretic reanalysis of Stepanek’s data ^[2] was conducted by Puthoff and Targ themselves rather than independently. The selective reporting concern (significant sub-series within a larger, unevenly significant dataset) is a structural feature of how the 51-experiment database is presented in ^[2], and the sources retrieved for this question do not include a published response from Puthoff or Targ that directly addresses the transcript-cue critique by name.