Feeling the Future: Experimental Evidence for Anomalous Retroactive Influences on Cognition and Affect
Also known as: Bem precognition experiments, Bem 2011 precognition experiments, Feeling the Future
This dossier is a research synthesis sourced using AI, not documentary evidence. Use the reference leads to check important claims.
This dossier concerns psychologist Daryl J. Bem’s 2011 article, “Feeling the Future: Experimental Evidence for Anomalous Retroactive Influences on Cognition and Affect,” a widely debated report in the Journal of Personality and Social Psychology. Bem described nine laboratory experiments, reportedly involving more than 1,000 participants, intended to test whether events occurring after a participant’s response could be associated with that earlier response. The paper framed its proposed effects as “retroactive” influences on cognition and affect: a future random outcome was claimed to correlate with a prior choice, performance measure, or physiological/affective response. Its examples included a computer-based task in which participants selected one of two curtains before a randomly determined image appeared, with some trials involving erotic images, and memory or priming paradigms in which later practice or exposure was claimed to improve earlier test performance. The article is historically important not because a psi interpretation is established, but because it became an unusually visible stress test of routine psychological research practices at the beginning of the reproducibility crisis. The basic controversy arose from a tension between the paper’s surface conformity to then-common experimental conventions and the extraordinary implication drawn from its positive results. The studies generally used familiar techniques: undergraduate or community participants, computerized randomization, standard null-hypothesis significance tests, multiple experiments, and conventional thresholds such as p < .05. Bem reported statistically significant deviations in the predicted direction across the series. On an extraordinary interpretation, these deviations indicated precognition, presentiment, or a reversal of ordinary temporal ordering in psychological influence. Critics instead argued that the observed pattern could emerge from ordinary sources of false-positive evidence: flexibility in stopping data collection, outcome selection, exclusions, analysis choices, repeated testing, selective reporting, and the tendency of journals to publish surprising positive findings more readily than null results. The case consequently figures in methodological discussions even among scholars with no interest in parapsychology. The experiments were associated with Cornell University and usually situated in Ithaca, New York, although the paper should be checked for the precise location, recruitment pools, and laboratory arrangements for each study. The participants were reportedly exposed to visual, verbal, and memory tasks rather than to a séance-like or field-based paranormal setting. This matters for comparison: the case is a claim produced through controlled behavioral research, not a witness tradition. The sensory material most frequently recalled is the computerized display of two curtained locations, followed by the revelation of an image selected by a random process. In the erotic-image anticipation experiment, the reported behavioral signature was a slightly above-chance tendency to choose the location where an erotic image would subsequently appear. In other paradigms, the proposed signature was better recall of words that would later be rehearsed, or a response pattern interpreted as retroactive priming. Effect sizes were generally characterized as small, so ordinary participant-level experience would not reliably resemble a dramatic subjective feeling of “knowing the future.” Several disagreements must be kept separate. One disagreement concerns whether the individual experiments were executed and analyzed exactly as reported; resolving that requires consulting the article, supplementary materials if any, later correspondence, and available data or protocols. A second concerns statistical inference: critics such as Eric-Jan Wagenmakers and collaborators argued that conventional p-values do not provide an adequate basis for strong belief in an implausible hypothesis, and they used Bayesian reasoning to argue that the evidence was weak when prior improbability was considered. Defenders replied that Bayesian conclusions depend on priors and that the reported results merited further empirical testing rather than summary dismissal. A third disagreement concerns reproducibility. Some later studies, including high-profile direct replications, reportedly failed to reproduce key effects, while some pooled or meta-analytic treatments claimed a residual small effect. Different conclusions may reflect differences in which experiments are included, whether unpublished results are sought, whether the replication was exact, and how study dependence and publication bias are modeled. The most consequential transmission history is methodological rather than occult. Bem circulated the work before or around publication in a disciplinary environment where preregistration, registered reports, data-sharing norms, and systematic correction for researcher degrees of freedom were less established than they later became. The resulting dispute was frequently used to illustrate why a sequence of nominally significant studies can be misleading when analysts retain many unrecorded choices. It helped motivate demonstrations of how common practices can produce false positives, including work by Joseph Simmons, Leif Nelson, and Uri Simonsohn on “false-positive psychology.” It also supplied a memorable public example for arguments over whether psychology should demand stronger evidential standards for surprising claims, use sequential methods, adopt preregistration, publish null replications, and distinguish exploratory work from confirmatory tests. Commercial and reputational incentives are relevant but should not be converted into allegations of misconduct. A counterintuitive finding in a prestigious journal can attract citations, media attention, book discussion, classroom use, and reputational stakes for both proponents and critics. Conversely, a conspicuous failed replication can become valuable as a methodological case study. These incentives can amplify conflict and retrospective certainty without demonstrating that any researcher deliberately manipulated a result. The scientifically cautious conclusion is that Bem reported disputed statistical findings under laboratory conditions; those findings do not establish that human cognition receives information from future random events. The case is best used in Lattice as a structured comparison point for small effects, extraordinary hypotheses, randomization claims, replication failures, analytic flexibility, publication selection, and the transformation of a contested result into a broader cultural emblem of a field’s reform debates.
- Words
- 2,643
- Observations
- 12
- Reference leads
- 6
- Validation score
- 100/100
Chronology and publication context
The exact calendar dates should be verified from the original journal record, but Bem’s article is generally identified as a 2011 publication in the Journal of Personality and Social Psychology. It presented nine experiments that had apparently been conducted before publication, largely in a Cornell-associated research context.
In 2011, the paper quickly attracted methodological criticism because its claims were both extraordinary and supported through analytic practices then widespread in experimental psychology. Wagenmakers and colleagues published a prominent Bayesian critique, while Bem and other commentators defended continued empirical examination and disputed aspects of the skeptical framing.
By 2012, prominently publicized replication efforts, including work associated with Stuart Ritchie, Chris French, and Richard Wiseman, reported failures to reproduce selected Bem effects. Other critical work examined the consequences of optional stopping, flexible analysis, selective publication, and the evidential value of repeated significant results.
In subsequent years, the article remained a canonical example in replication-crisis teaching, debates over preregistration, and discussions of the appropriate burden of proof for low-prior-probability claims. Later meta-analytic and replication literature should be treated as a contested continuation rather than a final settlement unless its methods and corpus are independently checked.
People, organisations, and setting
Daryl J. Bem was the principal author and public advocate for the article’s interpretation that the results could reflect anomalous retroactive influence. Cornell University and Ithaca, New York, are the principal recalled institutional and geographic associations, though the procedural record should be checked for experiment-by-experiment details.
The research setting was reportedly a controlled psychology laboratory using computer-administered tasks and human participants, often described as students or participants drawn from a university-associated population. This is unlike spontaneous precognition testimony because the claimed anomaly was operationalized as a small statistical departure from chance across groups.
Important later interlocutors include Eric-Jan Wagenmakers and collaborators, who challenged the inferential force of the evidence, and replication researchers including Ritchie, French, and Wiseman. Joseph Simmons, Leif Nelson, and Uri Simonsohn are relevant to the broader methodological context because their demonstrations of false-positive-generating research practices became closely associated with debates triggered by cases such as this one.
Reported phenomena and measured responses
The reported phenomena were not principally vivid visions, dreams, or subjective prophecies. They were group-level behavioral effects in which responses made before a random future event were claimed to predict, or be influenced by, that later event at rates modestly above chance.
In the best-known “precognitive detection” style task, participants selected between two screen positions or curtains before a computer displayed an image behind one position. A future random selection was said to determine the image location, and the reported effect was a higher-than-chance rate of choosing the later erotic-image location, with no comparable effect necessarily expected for neutral material.
Other experiments reportedly used memory and priming designs. A characteristic proposed effect was that participants did better on an earlier recall or recognition measure for material that they would only later practice, rehearse, or receive as a prime.
The meaningful sensory features were therefore visual computer displays, image valence or arousal, words or stimuli for memory tasks, and response keys or choices. No special environmental manifestation, external entity, or independently observable physical disturbance is required by the reported design.
Investigation, replication, and methodological tests
Evaluation requires inspection of the original methods: the random-number generation method, concealment of future outcomes, experimenter access to condition information, participant exclusions, trial counts, stopping rules, planned contrasts, and whether each reported analysis was specified before seeing the data. A randomization procedure can be technically sound while the overall evidential process remains vulnerable to selection or analytic flexibility.
Direct replications are especially informative when they use the same materials, outcome definitions, recruitment approach, sample size rationale, and analysis plan. Null replications do not logically prove that an original sample fluctuation or genuine tiny effect is impossible, but repeated, adequately powered null results reduce confidence in a robust and readily detectable phenomenon.
Meta-analysis can aggregate noisy small studies but introduces its own vulnerability to publication bias, nonindependence, heterogeneous protocols, and disputed inclusion rules. For this case, a pooled positive estimate should not be treated as a simple confirmation without examining unpublished studies, registered reports, correction methods, and the degree to which the corpus consists of independent replications.
Interpretive disputes and ordinary explanations
The central proponent interpretation is that later events somehow influence earlier cognition or affect, a result sometimes classified as precognition, presentiment, or psi. This interpretation is disputed and would conflict with ordinary assumptions about causal temporal order if taken literally.
Mundane explanations include sampling variation, multiple testing, optional stopping, selective exclusion, post hoc selection of successful tasks or stimulus categories, data-contingent modeling decisions, and publication bias. None requires conscious fraud; they can occur through ordinary professional practices, incentives, and incomplete reporting.
A Bayesian dispute is also central. Skeptics argue that an implausible claim needs much stronger evidence than a conventional p-value threshold, whereas critics of that response argue that priors are contestable and should not prevent testing. Both positions concern how evidence changes belief, rather than supplying a direct observation of backward causation.
There is also a genre dispute. To parapsychology-oriented readers, the paper may be an instance of a long experimental psi tradition. To many psychologists, it is primarily a demonstration of how conventional statistical norms can generate persuasive but nonreproducible findings.
Transmission, retelling, and public meaning
The title “Feeling the Future” is memorable and has encouraged retellings that can overstate the experimental claim as if participants consistently sensed forthcoming events. The reported effects were small aggregate statistical differences, not a practical forecasting ability demonstrated for an individual participant.
The article circulated through journal discussion, blogs, mainstream science reporting, methods courses, replication-crisis essays, and debates over open science. Retellings often foreground erotic stimuli because they are vivid and easily summarized, while giving less attention to the multiple paradigms, statistical assumptions, and procedural detail.
The case also became symbolically useful to opposing communities. Psi proponents could cite peer-reviewed positive findings in a mainstream journal, while reform advocates could cite the same paper as evidence that publication prestige and standard significance testing do not guarantee reliable discovery.
Cross-case comparison motifs
A useful cross-case motif is the “small effect with large ontological implication” pattern. The numerical deviation is claimed to be modest, but accepting the preferred interpretation would require major revision of background theory.
Another motif is “random future target.” This includes studies in which a later random stimulus, image category, or reinforcement schedule is asserted to correlate with an earlier measure. Comparing randomization, blinding, target-generation timing, and data handling is essential.
A third motif is “methodological controversy becomes cultural case study.” A finding may become more influential as an example in debates about inference and reform than as support for its original substantive hypothesis.
Limits of this recalled dossier
This synthesis is a set of research leads from recalled knowledge, not a verification of the article, its raw data, later replication protocols, or the complete publication record. Exact sample sizes, effect estimates, experiment numbering, and dates should be checked against primary sources.
The dossier does not infer deception, misconduct, or motives from the controversy. Claims about bias here refer to structural risks in research design and dissemination, not findings about the character or intent of particular researchers.
The available recall is stronger for the article’s broad role in replication-crisis history than for fine-grained procedural particulars. Any comparative analysis should preserve that distinction and avoid treating a disputed statistical signal as verified evidence of paranormal causation.
Chronology
Cornell-associated laboratory studies are conducted.
Bem reportedly conducted a series of nine computerized cognition and affect experiments intended to test retroactive influence, but the precise dates and sequence for each experiment require checking in the original paper.
approximate“Feeling the Future” is published and publicized.
Bem’s article appeared in the Journal of Personality and Social Psychology and reported statistically significant results interpreted as anomalous retroactive influences on cognition and affect.
documentedStatistical and Bayesian criticism intensifies.
Wagenmakers and collaborators became prominent critics of the article’s inferential basis, arguing that ordinary significance tests did not justify the extraordinary interpretation.
documentedHigh-profile replications report null results for selected effects.
Later replication work, including a widely discussed study by Ritchie, French, and Wiseman, reportedly failed to reproduce several of the original paradigms’ effects.
documentedThe paper becomes a replication-crisis reference point.
The controversy is repeatedly invoked in arguments for preregistration, transparency, stronger evidential standards, publication of null results, and clearer separation of exploratory from confirmatory analyses.
documentedPeople and roles
Daryl J. Bem.
Psychologist and author of the 2011 article.Bem reported the nine experiments and argued that their aggregate results were compatible with anomalous retroactive influence, an interpretation that remains disputed.
Cornell University.
Institutional association and recalled research setting.The experiments are commonly associated with Cornell and Ithaca, New York, although exact recruitment and lab arrangements should be checked in the primary report.
Eric-Jan Wagenmakers.
Methodological critic.Wagenmakers and collaborators published a well-known Bayesian critique relevant to the evidential interpretation of Bem’s results.
Stuart Ritchie.
Replication researcher.Ritchie is associated with a later high-profile attempt to replicate selected Bem experiments that reported null results.
Chris French.
Replication researcher and anomalistic psychologist.French collaborated on a prominent reported failure to replicate selected effects.
Richard Wiseman.
Replication researcher and public science communicator.Wiseman collaborated on a prominent reported failure to replicate selected effects.
Joseph P. Simmons, Leif D. Nelson, and Uri Simonsohn.
Methodological researchers relevant to the controversy’s context.Their work on flexible research practices and false positives is often discussed alongside the Bem episode, though it is not itself a replication of the psi claim.
Connections to explore
Small statistical effect with extraordinary implication.
The claimed behavioral deviations were modest, but a literal retrocausal account would challenge ordinary causal assumptions and therefore requires especially rigorous corroboration.
Suggested search: extraordinary claims small effects Bayesian priors replication psychologyFuture random target paradigm.
The case can be compared with presentiment and precognition experiments in which physiological or behavioral measures are claimed to anticipate a later randomly selected stimulus.
Suggested search: presentiment random future stimulus physiological anticipation experimentsErotic or high-arousal stimulus moderation.
The use of erotic imagery links this case to designs proposing that motivationally salient stimuli produce stronger anomalous anticipation than neutral stimuli.
Suggested search: Bem erotic stimuli precognitive detection experiment replicationPreregistration and researcher degrees of freedom.
The controversy is a core comparison case for claims whose apparent support may depend on unrecorded choices in sampling, exclusions, outcomes, or analysis.
Suggested search: Bem feeling future optional stopping preregistration false positive psychologyNull replication versus pooled residual effect.
The case illustrates how direct replications and meta-analyses can appear to conflict because they answer different questions and rely on different assumptions about selection and heterogeneity.
Suggested search: Bem psi replication meta-analysis publication bias registered replicationUnretrieved reference leads
Feeling the Future: Experimental Evidence for Anomalous Retroactive Influences on Cognition and Affect.
Daryl J. Bem. · Peer-reviewed journal article.
This is the primary source for the nine experiments, reported procedures, results, and authorial interpretation.
Suggested search: Daryl J Bem 2011 Feeling the Future Journal of Personality and Social PsychologyWhy Psychologists Must Change the Way They Analyze Their Data: The Case of Psi.
Eric-Jan Wagenmakers and collaborators. · Methodological critique.
This is a major Bayesian and methodological criticism of Bem’s evidential claims that should be checked against the original article and responses.
Suggested search: Wagenmakers Why Psychologists Must Change the Way They Analyze Their Data The Case of PsiFailing the Future: Three Unsuccessful Attempts to Replicate Bem’s ‘Retroactive Facilitation of Recall’ Effect.
Stuart J. Ritchie, Christopher C. French, and Richard Wiseman. · Replication study.
This is a key discovery lead for prominent null replication evidence focused on one of Bem’s paradigms.
Suggested search: Ritchie French Wiseman Failing the Future three unsuccessful attempts replicate Bem retroactive facilitation recallFalse-Positive Psychology: Undisclosed Flexibility in Data Collection and Analysis Allows Presenting Anything as Significant.
Joseph P. Simmons, Leif D. Nelson, and Uri Simonsohn. · Methodological journal article.
This article supplies the broader account of how conventional analytic flexibility can generate false positives and is historically linked to discussion of the Bem controversy.
Suggested search: Simmons Nelson Simonsohn False-Positive Psychology undisclosed flexibility data collection analysisCorrecting the Past: Failures to Replicate Psi.
Julia Galak and collaborators. · Replication study.
This is a likely relevant lead for later failed replication work and for comparing protocols, outcomes, and claimed effect sizes.
Suggested search: Galak Correcting the Past Failures to Replicate Psi BemFeeling the Future: A Meta-Analysis of 90 Experiments on the Anomalous Anticipation of Random Events.
Julia Mossbridge, Patrizio Tressoldi, and Jessica Utts. · Meta-analysis.
This separate review provides comparison material on presentiment research and contested pooling methods, but it is not the same subject as Bem’s nine-experiment paper.
Suggested search: Mossbridge Tressoldi Utts 2012 Feeling the Future meta-analysis 90 experiments