Discovery Lab Answers

Presentiment: What the Data Show

Presentiment research asks whether the body registers a physiological response to an emotionally significant stimulus before that stimulus is randomly selected and presented — a potential signature of unconscious precognition. The lead time is short: typically 0.5 to 15 seconds between the prestimulus physiological shift and the future event ^[1].

The Core Empirical Picture

The effect has been observed across multiple physiological systems — electrodermal activity (EDA/skin conductance), pupillary dilation, EEG, cardiac measures — and in humans and animals ^[1]. Dean Radin’s EDA program, launched in 1997, is among the most cited lines of work; his eye-tracking study with Borges examined anticipatory pupillary dilation and gaze behavior before randomly selected emotional versus neutral photos across two experiments ^[2].

Two meta-analyses have estimated the effect size for presentiment in the range of d ≈ 0.21 to 0.28 (Mossbridge et al., 2012; Duggan & Tressoldi, 2018) ^[1][3][4]. Duggan and Tressoldi’s estimate carries a 95% confidence interval of 0.18–0.38, and the cumulative signal across pooled studies departed from chance by more than six standard deviations ^[3][4]. That is the most precisely cited aggregate figure in the retrieved sources.

Radin and Sheehan (2011) situate presentiment within the broader 75-year record of precognition experiments and emphasize the design logic: physiological data are recorded and locked in computer memory before the stimulus is randomly selected, so the prestimulus record cannot be contaminated by foreknowledge of the outcome ^[5].

Scale Extension: Beyond the Laboratory

Levin (2024, 2026) references Radin’s analysis of 13 years of daily Twitter sentiment data in 10 languages, examining sentiment in the two weeks prior to events rated as significantly negative and unpredictable (acts of terrorism, mass shootings, unexpected celebrity deaths). The sources describe an apparent predictive pattern — in one instance a predictive signal up to one whole minute before a stimulus — extending the presentiment framework well beyond the standard 0.5–15 second window ^[3][4].

Roger Nelson’s Global Consciousness Project work adds a related signal at the largest scale: GCP data surrounding the September 11 attacks showed a large deviation centered on that day, with apparent structure beginning approximately one to two days before the first plane struck. Nelson estimated the ratio of global-scale response time to human neural response time at roughly 20,000:1, which he noted predicts a precursor on that order, consistent with what was observed.

Mechanistic Accounts and Competing Explanations

Mossbridge (2023) reviews two proposed models for short-lead-time unconscious precognition and notes that the presentiment paradigm is among the more tractable laboratory designs for testing them ^[1]. However, the sources also note that May et al.’s Decision Augmentation Theory (DAT) offers a conventional-anticipation alternative account of the EDA presentiment data — the idea that participants unconsciously bias their choices using existing environmental information rather than accessing future states. Radin published a direct response contesting the sufficiency of DAT as a full explanation [Dean Radin profile page].

Levin (2026) explores an orthodox quantum-mechanics justification for the presentiment effect, proposing parameter bounds within which such a model would be physically reasonable ^[3].

Unsettled Signals — What Is Not Resolved

• Effect size heterogeneity: the 0.21–0.28 range spans two independent meta-analyses; neither the sources nor the meta-analyses claim consistency across all modalities or labs.
• Alternative explanations remain live: DAT and conventional anticipation mechanisms are not ruled out by any single study; Radin’s response contests DAT but does not close the debate.
• Physiological system specificity: the effect appears in EDA, EEG, pupillometry, and cardiac data, but whether these are all measuring the same underlying phenomenon or different ones is unresolved ^[1].
• Scale generalizability: extending the effect from seconds in the lab to days in GCP or Twitter data involves assumptions about mechanism that are not empirically settled.
• Effect size magnitude: at d ≈ 0.21–0.28, the effect is small and embedded in strong noise; individual studies are easily underpowered ^[1][3][4].

The Presentiment and Electrodermal Anticipation page on Radin’s hub and the Roger D. Nelson profile carry the deeper synthesis on these two research programs.