Discovery Lab Answers

Presentiment: Data and Commentary

Coverage note: The evidence table below reflects studies currently held in the ESP-Nexus library. The library’s share of the published presentiment literature has not been measured, so treat this as a summary of what the library holds rather than a settled account of the field as a whole. The chart below the table plots z-scores against publication year for the subset of results reporting that metric.

What is presentiment?

Presentiment research tests whether the body registers physiological responses to emotionally significant stimuli before those stimuli are randomly selected and presented. It is operationalized as a measurable difference in autonomic or electrocortical activity in a pre-stimulus window, not as a conscious prediction. The phenomenon sits at the intersection of psychophysiology and parapsychology and is sometimes framed as a potential signature of unconscious precognition.

The evidence base

All 17 result rows in the structured evidence carry dir=positive — no null or below-chance result appears among the library’s holdings for this phenomenon. That is itself a datum worth examining carefully, because publication bias and selective reporting systematically favor positive outcomes, and the library’s coverage of this literature is unmeasured. With that qualification in place, here is what the rows actually report.

Meta-analytic and pooled results

Mossbridge et al. (2012) is the landmark meta-analysis of the physiological anticipation literature, pooling 26 studies under a fixed-effect model and returning a standardized effect of Cohen’s d = 0.21 with a confidence interval of [0.15, 0.27]. Radin (2011)’s sign-test aggregate across 101 studies found 85 returning positive results.

Individual experiments

Metric discipline: the table mixes standardized effect sizes (d, z/√N), raw z-scores, and p-only results. These are not comparable across rows and cannot be averaged into a single “overall” figure.

Physiological modalities covered

The library’s holdings span multiple measurement channels:

• Electrodermal activity (EDA/GSR) — the original Radin (1997) paradigm and its replications
• Pupillometry and eye-tracking — Radin (2009)
• Slow cortical potentials (SCP) — Radin (2007)
• EEG phase-synchronization — Kittenis (2011)
• EEG pre-stimulus classification — Mossbridge (2017, 2023)
• Social-media sentiment aggregates — Radin (2022, 2023), extending the paradigm to collective-level data via Twitter

The convergence across modalities is frequently cited by proponents as strengthening the inference; critics counter that shared methodological conventions (stimulus-set construction, epoch selection, motion artifact handling) could propagate a common confound across modalities.

The scatterplot renders z-scores against publication year for the 11 results across 9 papers that report on this metric, with an OLS trend line. Note that z is a test statistic that grows with sample size — it reflects statistical significance over time, not effect magnitude.

Unsettled signals (what the evidence does not settle)

Several questions remain open within the library’s holdings and in the broader literature:

• All rows are positive-direction. No null or negative result appears in the structured evidence. Whether this reflects a genuine absence of null findings in the literature, publication bias, or the library’s current holdings is unknown. The library’s share of the full presentiment literature has not been measured, so the possibility that null results exist and are underrepresented cannot be assessed from this table alone.
• Effect-size heterogeneity. The three rows reporting standardized effect sizes span ES = 0.064 (Radin 2004) to ES = 0.21 (Mossbridge 2012) — a threefold range. The Mossbridge (2012) meta-analysis itself covered 26 studies under a fixed-effect model, and the appropriateness of that model (vs. random-effects) when heterogeneity is present is a standing methodological debate.
• Moderator questions. Radin (2007) reports a significant result specifically for female participants on SCP; Tressoldi (2010) reports a significant result specifically for high-absorption participants. Whether presentiment is moderated by sex, trait absorption, meditation experience, or other individual-difference variables has not been resolved.
• Social-media extension. Radin (2022, 2023) extend the paradigm to collective Twitter sentiment — a methodologically distinct approach using natural language processing and circular-shift permutation statistics, not controlled psychophysiology. Whether this constitutes a replication of the laboratory presentiment effect or a separate phenomenon is not established.
• Decline over time. Radin (2011) and Mossbridge (2012) cite stability of the effect over time; whether z-scores in the chart below show a meaningful trend is for the reader to assess from the fitted line in the caption.

Skeptical critiques

What critics argue. Edwin C. May and colleagues have advanced Decision Augmentation Theory (DAT) as an alternative account of EDA presentiment data. DAT proposes that the observed pre-stimulus physiological differences do not require real-time anticipation of a future stimulus; instead, a psi-mediated bias at the decision point (when the participant or experimenter selects trial parameters) could produce the same pattern without invoking moment-to-moment somatic anticipation. On this account, what looks like presentiment is a selection artifact from precognitive influence on earlier choices, not a body “feeling the future” during the trial itself.

What the experimental data show. Radin published a direct response contesting the sufficiency of DAT as an explanation of his EDA presentiment data, arguing that the timing and structure of the observed autonomic responses are not well-explained by a decision-selection mechanism alone. The exchange is documented in the ESP-Nexus May: Precognition and presentiment mechanisms page. The library’s structured evidence does not contain a DAT-positive empirical result that directly tests the two models against each other.

Analysis. DAT and the somatic-anticipation account make overlapping but distinguishable predictions about trial-level timing and about what happens when decision points are experimentally controlled. Published tests designed to distinguish them have not produced a widely agreed resolution. Independent replication of the basic EDA paradigm outside Radin’s group has been reported — Bierman (2000) and Mossbridge (2012) aggregate results from multiple laboratories — but the specific DAT-versus-presentiment contrast has received less direct experimental attention than the basic paradigm itself.

A second line of criticism concerns artifact control: whether pre-stimulus windows are genuinely free from post-stimulus contamination given epoch-definition choices, and whether the random stimulus sequences are sufficiently independent of any motor preparation or orienting response. These concerns appear in the Franklin et al. (2014) review of the precognition literature as ongoing methodological questions rather than settled refutations.

For a deeper synthesis of Radin’s presentiment program specifically, see the ESP-Nexus page Presentiment and Electrodermal Anticipation of Future Stimuli, and for May’s theoretical alternative, May: Precognition and presentiment mechanisms.