Discovery Lab Answers

Anomalous Cognition / Remote Viewing

Before reading these results: the library’s share of the published literature on remote viewing has not been measured. What follows summarizes the studies currently held in the ESP-Nexus library — treat it as a sample, not a settled account of the field. Because the coverage bias is unmeasured, it is not possible to say which way the picture is skewed relative to the full literature.

What the terms mean

“Remote viewing” (RV) is the experimental protocol in which a participant attempts to describe a physically distant or concealed target — a location, photograph, or object — without any conventional sensory access to it. The term was introduced in the 1970s at Stanford Research Institute (SRI) by Hal Puthoff and Russell Targ. “Anomalous cognition” (AC) is the neutral laboratory term later adopted at SRI and the government’s Stargate Program as a replacement label for the same phenomenon. The two terms are used interchangeably in the literature; AC also covers related phenomena such as forced-choice ESP and ganzfeld work.

The evidence base: pattern across 21 studies / 23 result rows

The rows span roughly five decades — from an early nine-trial series reported by Targ (1975) through a 2026 survey by Katz, Prather, and Akin — and they use five different, non-comparable metrics (standardized effect sizes, raw hit rates, z-scores, p-values only, and one correlation). Because these metrics are not interchangeable, no single pooled “bottom-line effect size” can be drawn across all rows; the evidence is described metric by metric and study by study below.

Positive results

The majority of rows report positive directions:

• Tressoldi (2023) conducted a random-effects meta-analysis across 40 studies and found a positive standardized effect with a confidence interval that does not include zero, using a z-over-√N metric — this is one of the few rows supplying a confidence interval, making it among the more formally precise entries in the set.
• Dunne (2003) reported a positive correlation across 336 formal binary trials with 30 participants, with a z and p both strongly against chance.
• Dunne (2005) reported two separate results: a meta-analytic z across 13 REG/remote-perception experiments combined, and a p-value for 653 remote-perception trials — both positive.
• Mueller (2021) found a positive hit rate across 100 trials pooled across present and future targets, with p well below conventional thresholds.
• Mueller (2017) reported direct hits (target ranked first) across 36 sessions at a hit rate above chance expectation.
• Roe (2020) combined three ganzfeld-condition experiments (110 trials) and found a positive z-over-√N effect with a hit rate above chance.
• Targ (2019) described the Pat Price formal nine-trial series with a high hit rate, citing a chance probability of roughly 1 in 100,000.
• Targ (1975) reported the earlier nine-experiment Pat Price series using judges’ selection analysis, also in a positive direction.
• Escola-Gascon (2023) reported three result rows: a total-sample result across 634 participants and over 20,000 trials, a “believers” subgroup result, and a per-experiment breakdown — all positive and with Cohen’s d effect sizes above chance.
• Piao (2023) reported a positive outcome on its primary metric (trials meeting a dual-grade threshold) for a single participant across 100 trials.
• Schwartz (2019) and Schwartz (2020) each reported positive p-values for location-prediction outcomes in applied remote viewing contexts.
• Schwartz (2025) reported a high self-reported accuracy proportion for objectively verifiable concepts in a precognitive RV subset.
• Katz (2026) surveyed experienced remote viewers and found that a high proportion reported a “sense of correctness” later verified — this is a prevalence figure from self-report, not a direct experimental hit rate.
• Brown (2012) found positive clarity-score distributions across 11 targets in 86 sessions.
• Mossbridge (2024) found a positive chi-square result for target-matching across moon phases in 792 trials.

Null and below-chance results

Three rows run in a different direction, and the grounding policy requires these be given equal weight:

• Wahbeh (2021) found a null overall result on the Quick Remote Viewing task across 534 participants — the largest participant count in the set by a substantial margin.
• Wahbeh (2022) found a below-chance (psi-missing) result on the same Quick Remote Viewing task in a pre-post intervention design (481 participants), with a negative Hedges’ g.
• Wiseman (2010) found a null result across four blind-judging trials using majority-vote scoring.

Unsettled signals — what the evidence does not resolve

The following disagreements are present in the rows and must not be smoothed over:

• Direction disagreement. Wahbeh (2021, 2022) and Wiseman (2010) return null or below-chance results on the same general protocol category where other studies report positive effects. Wahbeh (2021) is also the highest-N individual study in the set, which means the null comes from the largest single-sample test.
• High heterogeneity. Tressoldi (2023) flags high between-study heterogeneity (I² ≥ 50%) in the pooled result, meaning the studies going into that meta-analysis do not behave as a uniform set — the positive pooled effect coexists with substantial variation across studies.
• Metric incompatibility. Standardized effect sizes, raw hit rates, p-values only, and correlations cannot be averaged; what looks like a consistent “positive” pattern across rows reflects partly the predominance of certain study designs and metrics, not a single replicated quantity.
• Sample and design heterogeneity. Rows range from single-participant case studies (Piao 2023: one participant, 100 trials; Targ 1975/2019: the Pat Price series) to large multi-participant surveys (Wahbeh 2021: 534 participants). Single-participant series with exceptional performers do not generalize the same way as population-representative samples.

The government program and institutional context

The Stargate Program (run through SRI and later the Science Applications International Corporation, SAIC) was a classified US government effort that ran for roughly two decades and produced hundreds of applied remote viewing tasks. Two formal assessments were commissioned in the mid-1990s: Utts (1996) concluded that the experimental evidence was sufficient to establish AC as a real phenomenon and that precognitive RV was particularly consistent; Mumford, Rose, and Goslin (1995) concluded that methodological problems and the concentration of results within a single laboratory prevented a definitive interpretation. These two assessments, reached from the same data set, frame a disagreement that subsequent work has not fully resolved.

Edwin C. May, who is profiled on ESP-Nexus, developed a theoretical framework linking target properties — specifically Shannon entropy — to AC performance quality, proposing that the information complexity of a target image influences the likelihood and accuracy of remote viewing responses. Patrizio E. Tressoldi, also profiled on the site, has conducted meta-analytic work across multiple AC paradigms; his 2023 meta-analysis is one of the rows in the table above. Dean I. Radin’s work on anomalous cognition, including mediumship and channeling research, is covered at his ESP-Nexus profile.

Skeptical critiques

What critics argue. Mumford, Rose, and Goslin (1995) argued that the SRI/SAIC body of work, while anomalous, did not constitute proof of a paranormal phenomenon because: (a) statistically significant results were concentrated within one laboratory using one set of methods, limiting independence of evidence; (b) no plausible causal mechanism had been identified; and (c) competing explanations had not been carefully ruled out. Their report concluded that the operational utility of RV for intelligence purposes was not demonstrated.

What the experimental data show. Utts (1996) countered from the same data that eight summary findings held across the SRI/SAIC work, including that precognitive RV was successful and that performance appeared consistent with other laboratories working in the same period. Tressoldi (2023) extended that cross-laboratory argument meta-analytically across 40 studies, finding a positive effect with a confidence interval excluding zero — though with the high heterogeneity noted above. Wahbeh (2021, 2022), using a standardized online protocol with a large unselected sample, obtained null and below-chance results, consistent with the concern that positive effects may depend on selected participants or specific laboratory conditions.

Analysis. Mumford et al. and Utts reached opposite conclusions from overlapping evidence in 1995–1996; that disagreement has not been resolved by a single decisive replication. The Wahbeh null results (particularly the 534-participant study) introduce a scale of negative evidence not present in the earlier program-era work. Tressoldi’s heterogeneity finding means that even a positive pooled estimate does not imply consistent replication across individual studies. Independent replication outside the original SRI/SAIC group using the same formal protocol and blind judging procedures, at comparable scale, has been attempted but not produced uniformly positive outcomes across groups.

For a broader synthesis of AC across paradigms — including forced-choice and ganzfeld designs — see the Anomalous Cognition page, and for the meta-analytic framing, Tressoldi’s spoke page.