Overview

Daryl J. Bem occupies a unique position in contemporary parapsychology: he is a mainstream social psychologist with a distinguished career in non-psi domains who published a high-profile psi study in a top-tier psychology journal, thereby bringing anomalous cognition research into direct contact with the broader experimental-psychology community. His 2011 paper in the Journal of Personality and Social Psychology was not a parapsychology publication; it was a mainstream venue, which amplified both its impact and the scrutiny it received.

Bem’s approach to psi research differs fundamentally from traditional parapsychology. Rather than designing novel paradigms to detect psi, he “time-reversed” well-established psychological effects (priming, habituation, recall facilitation, and affective approach/avoidance) so that the stimulus events occurred after the participant’s response was recorded. If psi exists, he reasoned, the same psychological effects should emerge in the retroactive direction. This methodological strategy grounded his work in mainstream experimental psychology while testing for anomalous retroactive influences.

His subsequent meta-analytic work, conducted in collaboration with Patrizio E. Tressoldi, Thomas Rabeyron, and Michael Duggan, synthesized 90 experiments from 33 laboratories across 14 countries, providing a large-scale assessment of the replicability and effect size of anomalous anticipation effects. This work has become central to debates about the empirical status of psi and the robustness of effect sizes in controversial domains.

Life and career

Bem is a Professor Emeritus in the Department of Psychology at Cornell University. Before his engagement with psi research, he established himself as a prominent social psychologist through theoretical contributions that shaped the field. His self-perception theory, developed in the late 1960s, offered an influential account of how individuals infer their own attitudes and beliefs from their behavior. Later, his exotic-becomes-erotic hypothesis provided a developmental theory of sexual orientation that generated substantial empirical and theoretical discussion.

These pre-psi contributions established Bem’s credibility and standing within mainstream psychology. When he turned his attention to anomalous cognition in the early 2000s, he brought with him both methodological sophistication and institutional legitimacy. His decision to publish psi research in mainstream psychology journals rather than parapsychology venues was deliberate: it positioned the work where mainstream psychologists would encounter it and where it could influence broader conversations about replication and statistical practice.

Bem’s entry into psi research was gradual. In the 1990s, he collaborated with Charles Honorton on a meta-analytic examination of the ganzfeld ESP database, published in Psychological Bulletin. This work, which found replicable evidence for an anomalous process of information transfer, demonstrated that Bem could engage with parapsychology data while maintaining the analytical rigor expected in mainstream psychology venues. His subsequent individual experiments on retroactive influences built on this foundation, extending the methodology into new domains of psychological effect.

Research

The 2011 “Feeling the Future” experiments

Bem’s landmark 2011 paper reported nine experiments designed to test for retroactive influences on cognition and affect by time-reversing established psychological effects.¹ The experiments tested four categories of retroactive effects: precognitive approach to erotic stimuli, precognitive avoidance of negative stimuli, retroactive priming, retroactive habituation, and retroactive facilitation of recall.

Across all nine experiments, involving more than 1,000 participants, the mean effect size was 0.22, and all but one experiment yielded statistically significant results.¹ The experiments employed standard psychological paradigms (word lists, image presentations, reaction-time tasks) but reversed the temporal sequence so that the stimulus event occurred after the participant’s response had been recorded. If anomalous retroactive influences exist, Bem hypothesized, the same effects observed in the forward direction should appear in the reversed direction.

A key finding was the role of individual differences. Stimulus seeking, a component of extraversion, was significantly correlated with psi performance in five of the nine experiments.¹ Participants who scored above the midpoint on stimulus seeking achieved a mean effect size of 0.43, substantially higher than the overall mean. This suggested that personality variables might moderate anomalous cognition effects, a pattern consistent with earlier parapsychology research.

Meta-analysis of 90 experiments

To encourage independent replication, Bem made all experimental materials available on request. In 2016, he published a meta-analysis of 90 experiments conducted by 33 laboratories in 14 countries, examining the replicability and effect size of anomalous anticipation effects.²

The meta-analysis yielded an overall effect greater than 6 sigma, with z = 6.40 and p = 1.2 × 10⁻¹⁰, and an effect size (Hedges’ g) of 0.09.² A Bayesian analysis produced a Bayes Factor of 5.1 × 10⁹, greatly exceeding the criterion value of 100 for “decisive evidence” in support of the experimental hypothesis.² When Bem’s original experiments were excluded, the combined effect size for independent replications was 0.06 (z = 4.16, p = 1.1 × 10⁻⁵), with a Bayes Factor of 3,853, again exceeding the criterion for decisive evidence.²

The meta-analysis also addressed concerns about publication bias and p-hacking. The number of potentially unretrieved experiments required to reduce the overall effect size to a trivial value of 0.01 was 544, and seven of eight additional statistical tests supported the conclusion that the database was not significantly compromised by selection bias or intense p-hacking.² P-curve analysis, a recently introduced statistical technique, estimated the true effect size to be 0.20 for the complete database and 0.24 for independent replications, virtually identical to the effect size of Bem’s original experiments (0.22) and closely related presentiment experiments (0.21).²

Ganzfeld meta-analysis

In 1994, Bem and Charles Honorton published a meta-analytic examination of the ganzfeld ESP database in Psychological Bulletin, a mainstream psychology venue. The analysis found replicable evidence for an anomalous process of information transfer.⁴ This work demonstrated that Bem could engage with parapsychology data while meeting the standards of mainstream psychology journals, and it established a precedent for his later psi publications.

International replication and expectancy effects

A 2021 international collaboration involving multiple laboratories sought to replicate and extend Bem’s precognition priming task.³ Two experiments were conducted with preregistered confirmatory hypotheses. The first experiment’s confirmatory hypothesis was not supported; however, exploratory analyses indicated that participants who completed the English-language version showed a significant effect, consistent with the original study, while no significant departure from chance was found in non-English translations.³

The second experiment tested whether a pro-psi or anti-psi statement at the beginning would modulate the effect. The primary psi hypothesis was not supported, and there was no effect in the English-language sample. However, mixed support emerged for the effect of the psi statement: participants who received the pro-psi statement had a greater psi score than those who received the anti-psi statement.³ These results highlighted the challenges of international collaboration in controversial science and the potential role of language and expectancy in task performance.

Skeptical critiques

Bem’s 2011 paper provoked substantial skeptical scrutiny, much of it focused on statistical methodology and the interpretation of effect sizes in the context of the broader replication crisis in psychology.

One major critique concerned the use of frequentist statistics and the interpretation of p-values. Wagenmakers, Wetzels, Borsboom, and van der Maas (2011) reanalyzed Bem’s data using Bayesian methods and argued that, contrary to his conclusion, the data did not yield evidence in favor of the psi hypothesis. They contended that Bem had incorrectly selected an unrealistic prior distribution for their analysis. Bem, in collaboration with Bayesian statisticians Jessica Utts and Wesley O. Johnson, responded that Wagenmakers et al. had employed an unrealistic prior and that a Bayesian analysis using a more reasonable distribution yielded strong evidence in favor of the psi hypothesis.⁵ This exchange highlighted the role of prior assumptions in Bayesian analysis and the importance of transparent specification of statistical assumptions.

A second line of critique focused on the effect size itself. An effect size of 0.22 in the original experiments, while statistically significant, is modest by the standards of much of psychology. Critics argued that such small effects are difficult to replicate reliably and may reflect methodological artifacts or publication bias. The meta-analysis addressed this concern by demonstrating that independent replications yielded effect sizes consistent with the original experiments and that the database showed no evidence of severe publication bias.

A third critique concerned the role of expectancy and experimenter effects. The international replication study revealed that effects were present in English-language versions but absent in translated versions, raising questions about whether language, cultural context, or experimenter expectancy might account for the observed effects rather than genuine psi phenomena.³

Responses and evaluation

Bem and his collaborators have engaged substantively with skeptical critiques, providing evidence-based responses grounded in statistical analysis and replication data.

On the Bayesian statistics debate, Bem’s response with Utts and Johnson demonstrated that the choice of prior distribution is critical to Bayesian inference and that reasonable priors yield strong evidence for the psi hypothesis.⁵ This response contributed to broader conversations in psychology about the proper use of Bayesian methods and the importance of transparent specification of assumptions. The exchange illustrated that disagreement about psi effects could not be resolved by statistical method alone; the underlying assumptions about prior probabilities and the plausibility of psi effects remained contested.

On the question of effect size and replicability, the meta-analysis of 90 experiments provided direct evidence that independent laboratories could replicate the effect, albeit with a smaller effect size (0.06) than Bem’s original experiments (0.22).² The consistency between the original effect size and the P-curve estimate of the true effect size (0.20 for the complete database, 0.24 for independent replications) suggested that the effect was not an artifact of selective reporting or p-hacking. However, the smaller effect size in independent replications raised questions about whether the effect was robust across all laboratory contexts or whether it depended on specific conditions or experimenter characteristics.

On the role of expectancy and language, the international replication study revealed a genuine puzzle: effects appeared in English-language versions but not in translations.³ Rather than dismissing this as evidence against psi, Bem and collaborators interpreted it as evidence that language, cultural context, or participant expectations might modulate psi effects. This interpretation is consistent with parapsychology research suggesting that belief, expectancy, and experimenter effects can influence psi performance. However, it also raises the question of whether the effect is genuinely anomalous or whether it reflects demand characteristics or experimenter bias. The authors acknowledged these challenges and emphasized the importance of continued investigation into the mechanisms underlying the observed effects.

Importantly, Bem has not claimed that his work proves the existence of psi in any absolute sense. Rather, he has argued that the data provide evidence for anomalous retroactive influences that warrant serious investigation and that cannot be easily dismissed as artifacts or statistical flukes. The meta-analysis, in particular, demonstrates that the effect is reproducible across multiple laboratories and that it is not simply a product of Bem’s own experimental practice.

Influence and reception

Bem’s 2011 paper had an outsized impact on experimental psychology, though not primarily because the field accepted the psi hypothesis. Instead, the paper became a catalyst for methodological reform and a focal point for debates about replication, statistical practice, and the credibility of psychological science.

The paper’s publication in Journal of Personality and Social Psychology, a top-tier mainstream venue, meant that it reached a broad audience of experimental psychologists who might never have encountered parapsychology research. The paper’s apparent demonstration of a statistically significant effect using standard experimental methods forced mainstream psychology to confront questions about the robustness of published findings and the adequacy of conventional statistical practices. This timing coincided with growing concerns about replication failures in psychology, and Bem’s work became emblematic of the broader crisis.

The paper’s reception was mixed. Some psychologists viewed it as evidence that the field’s statistical and methodological standards were inadequate and that many published findings might not replicate. Others argued that the paper demonstrated the importance of pre-registration, transparent reporting, and replication studies. The paper became a touchstone in discussions about p-hacking, publication bias, and the need for reform in how psychology conducts and reports research.

Bem’s subsequent meta-analysis and his engagement with international replication efforts demonstrated a commitment to transparency and replicability that earned respect even from skeptics. By making materials available and encouraging independent replication, Bem modeled best practices for controversial research. The meta-analysis, which synthesized 90 experiments from 33 laboratories, provided a large-scale assessment of the replicability of anomalous anticipation effects and contributed to broader conversations about how to evaluate evidence in controversial domains.

In parapsychology specifically, Bem’s work has been influential in establishing that anomalous cognition effects can be studied using mainstream experimental methods and that the results can be subjected to rigorous meta-analytic scrutiny. His ganzfeld meta-analysis with Honorton demonstrated that parapsychology data could be examined in mainstream psychology journals, and his subsequent work has continued this tradition. However, Bem remains a social psychologist first and a psi researcher second; his primary identity and institutional affiliation are in mainstream psychology, not parapsychology.

The broader impact of Bem’s work extends beyond parapsychology to questions about how science evaluates evidence for extraordinary claims. His work has contributed to discussions about the role of effect size, replication, meta-analysis, and Bayesian inference in assessing controversial findings. Whether or not psi exists, Bem’s methodological contributions have influenced how experimental psychology approaches questions of replicability and statistical inference.