Overview

When Dunne arrived at Princeton in 1979 to establish the PEAR laboratory, the field of parapsychology lacked standardized experimental protocols and faced persistent accusations of methodological sloppiness. Her background in developmental psychology, combined with her exposure to remote viewing research, positioned her to design experiments that would meet rigorous scientific standards while remaining sensitive to the subtle, context-dependent nature of anomalous phenomena.

Dunne’s methodological philosophy rested on several principles: (1) large-scale data collection to establish statistical baselines; (2) systematic variation of experimental conditions to isolate operator and environmental factors; (3) transparent reporting of all trials, including null results; and (4) collaborative engagement with statistical critics to address methodological concerns. This approach made PEAR’s work both influential and controversial, influential because the protocols were detailed and reproducible, controversial because skeptical statisticians identified potential vulnerabilities in the analytical framework.

Over 28 years, Dunne and her PEAR colleagues published more than 60 papers documenting experiments in remote viewing, random event generation, and human-machine interaction. Many of these publications included detailed methodological appendices, statistical tables, and responses to published critiques. This transparency, while intended to strengthen the field’s credibility, also created a permanent record of methodological choices that critics could scrutinize.

Remote viewing protocols and statistical analysis

Dunne’s most visible methodological contribution was the refinement of remote viewing protocols. Remote viewing (the attempt to describe a distant or hidden target using only mental imagery) had been studied at Stanford Research Institute (SRI) in the 1970s, but lacked standardized analytical procedures. Dunne developed a binary database method that converted qualitative descriptions into quantitative scores, allowing for statistical analysis across large numbers of trials.

The PEAR remote viewing protocol involved a sender (or automated target selection system) and a percipient who attempted to describe the target location or image. Dunne introduced several analytical refinements: (1) blind judging, in which an independent judge matched the percipient’s written description against a set of possible targets without knowing which was correct; (2) ranking procedures that assigned numerical scores based on the judge’s confidence; and (3) binary encoding of target features (e.g., “indoor” vs. “outdoor,” “natural” vs. “artificial”) to enable statistical comparison across diverse target sets.¹

These innovations allowed Dunne to analyze thousands of remote viewing trials using standard statistical tests. The PEAR database became one of the largest remote viewing datasets in parapsychology, providing a foundation for meta-analytic studies and effect-size estimation. Dunne’s analytical methods were adopted by other laboratories and remain cited in contemporary remote viewing research.

However, the shift from qualitative assessment to binary scoring introduced new methodological questions. Critics asked whether the binary encoding captured the full richness of the percipient’s descriptions, whether the choice of target features was arbitrary, and whether the blind judging procedure was truly blind when judges had access to metadata (e.g., trial dates, percipient identities) that might correlate with target properties.

RNG experimental design and operator effects

Dunne’s RNG experiments represented a shift from remote viewing toward direct mind-matter interaction. In these experiments, human operators attempted to influence the output of random event generators, electronic devices that produced random binary sequences (0s and 1s). The hypothesis was that focused intention could bias the statistical distribution of the random sequence toward a pre-stated target (e.g., more 1s than 0s).

The PEAR RNG experiments were massive in scale. Dunne and her colleagues collected millions of individual RNG trials across hundreds of operators, creating a database that allowed for fine-grained analysis of operator effects, gender differences, and temporal patterns. Dunne’s own first-authored work documented consistent gender effects, with female operators showing different statistical patterns than male operators in RNG experiments.³

Dunne also identified series position effects, a phenomenon in which RNG results were stronger at the beginning of an experimental session and declined over time.² This finding suggested that operator attention, motivation, or fatigue played a role in the anomalous effect. Rather than dismissing the effect as an artifact, Dunne incorporated it into her analytical model, treating it as evidence that the phenomenon was sensitive to human psychological factors.

The RNG methodology included several controls intended to rule out conventional explanations: (1) the RNG devices were tested for hardware bias before and after experiments; (2) baseline trials (without operator intention) were collected to establish chance performance; (3) operators were blind to the RNG output during the experiment; and (4) statistical analysis was pre-specified to avoid post-hoc data mining.⁴

Despite these controls, the RNG effect sizes were small, typically on the order of a few percent deviation from chance. This raised a fundamental methodological question: were the observed effects genuine anomalies, or were they artifacts of subtle biases in the experimental apparatus, data analysis, or statistical testing procedures?

Skeptical critiques

The PEAR methodology attracted sustained criticism from both mainstream statisticians and skeptical parapsychologists. The most prominent critique came from Hansen, Utts, and Markwick, who published a detailed statistical analysis of PEAR’s remote viewing experiments in the Journal of Parapsychology. Their critique raised several methodological concerns: (1) potential sensory leakage through the experimental protocol; (2) possible biases in the blind judging procedure; (3) the use of multiple statistical tests without appropriate correction for multiple comparisons; and (4) the possibility that the reported effects were artifacts of the analytical method rather than genuine anomalies.

Hansen and colleagues were particularly concerned about the binary encoding procedure. They argued that converting qualitative descriptions into binary scores involved subjective choices that could introduce bias. For example, the decision to code a target as “indoor” or “outdoor” might depend on the judge’s interpretation of ambiguous descriptions, and this interpretation might be influenced by knowledge of the percipient’s identity or the trial date.

Another line of criticism focused on the statistical analysis itself. Skeptics noted that PEAR used multiple statistical tests (e.g., comparing hit rates across different target categories, analyzing temporal patterns, examining operator subgroups) without applying standard corrections for multiple comparisons. This raised the possibility that some reported effects were statistical flukes, false positives that would not replicate in independent studies.

The small effect sizes also drew criticism. Even if the PEAR effects were statistically significant, skeptics argued, the practical significance was minimal. An effect size of a few percent deviation from chance could easily be explained by subtle methodological biases or experimenter expectancy effects, rather than by genuine anomalous phenomena.

Responses and evaluation

Dunne and her PEAR colleagues responded to the Hansen, Utts, and Markwick critique with a detailed technical reply published in the Journal of Parapsychology. In this response, Dunne and her co-authors addressed each methodological concern, providing additional data and clarifications about the PEAR protocols.¹

Regarding sensory leakage, Dunne emphasized that the PEAR remote viewing protocol included multiple safeguards: the percipient was physically isolated from the target location, the target was selected by a randomization procedure that the percipient could not influence, and the percipient’s descriptions were recorded in real-time without opportunity for revision. She argued that the specific details in percipient descriptions (architectural features, colors, spatial layouts) were unlikely to be guessed through sensory leakage or prior knowledge.

On the question of bias in blind judging, Dunne acknowledged that the procedure was not perfectly blind, as judges had access to metadata. However, she argued that the PEAR team had conducted additional analyses using different judging procedures (e.g., computer-based matching, independent judges from outside the laboratory) and obtained similar results. This convergence across different analytical methods suggested that the findings were not artifacts of a single judging bias.

Regarding multiple comparisons, Dunne noted that PEAR had pre-specified its primary statistical tests before analyzing the data. While the team had conducted exploratory analyses (e.g., examining gender effects, temporal patterns), these were presented as secondary findings requiring replication rather than as primary evidence for anomalous effects. She also pointed out that many of the reported effects (e.g., gender differences, series position effects) had been replicated across independent experiments and operator samples, suggesting they were not statistical flukes.

The question of effect size remained more difficult to address. Dunne acknowledged that the PEAR effects were small, but argued that this did not necessarily indicate methodological artifact. Small effects could reflect the genuine nature of consciousness-matter interaction, a subtle, probabilistic phenomenon rather than a dramatic, easily detectable one. She also noted that small effects in physics and psychology are often accepted as genuine when they are consistent, replicable, and theoretically meaningful.

Dunne’s engagement with methodological criticism was notable for its depth and specificity. Rather than dismissing skeptical concerns as closed-minded, she treated them as legitimate scientific questions and provided detailed technical responses. This approach elevated the methodological discourse in parapsychology, even though it did not fully resolve the underlying disagreements about the interpretation of the PEAR data.

From a contemporary perspective, Dunne’s methodological contributions can be evaluated on several dimensions. First, her protocols and analytical procedures were substantially more rigorous than those used in earlier parapsychology research. The PEAR laboratory established standards for experimental control, data transparency, and statistical reporting that influenced the field. Second, her work demonstrated that anomalous effects, if they exist, are subtle and context-dependent, they require large sample sizes and careful experimental design to detect reliably. Third, her engagement with skeptical criticism, while not resolving fundamental disagreements, advanced the quality of methodological discourse in parapsychology.

However, the persistent methodological critiques also suggest that Dunne’s protocols, while innovative, did not fully eliminate the vulnerabilities that skeptics identified. The small effect sizes, the complexity of the analytical procedures, and the difficulty of achieving truly blind conditions in consciousness research remain unresolved challenges. Whether these challenges reflect genuine limitations of the phenomenon or limitations of current experimental methods remains a matter of scientific debate.