Mind-to-Mind: A Multi‑Phase Talent Selection Program and Experimental Test of Interconnected Minds

This registration documents the planned predictor analyses for Phase 4 of a multi-phase research program investigating reports of direct mind-to-mind communication under controlled conditions. The broader program is designed to identify emotionally bonded pairs who report repeated experiences consistent with telepathic connection and to evaluate their performance using experimental paradigms that minimize sensory cues and expectancy effects. Phase 4 employed a remote, intention-based task administered via videoconferencing. In each session, one participant (“Sender”) alternated between directing focused positive intention toward their partner and engaging in a cognitively demanding control task. The paired participant (“Receiver”) attempted to identify the Sender’s condition across 10 randomized trials. Each dyad completed two Receiver blocks (20 total trials per pair). Sessions were monitored by study staff to ensure physical isolation and protocol adherence. Individual- and pair-level accuracy were evaluated relative to a binomial chance expectation of 0.50 per trial. Aggregate accuracy analyses have already been completed and are reported separately. Data included in this preregistration were collected between July 16, 2024, and January 10, 2026, prior to conducting the predictor analyses specified herein. The purpose of the present preregistration is to formally specify exploratory predictor analyses examining variability in task performance. At the individual level, mixed-effects binomial logistic regression models will be used to evaluate whether psychological traits, experiential characteristics, personality dimensions, meditation history, and performance on related tasks are associated with Receiver accuracy. A random intercept for pair will account for non-independence within dyads. The binomial modeling framework evaluates deviations from the chance probability of 0.50 at the trial level and therefore assesses predictors of above-chance performance without dichotomizing accuracy. Predictors will be grouped into prespecified theoretical model families, and multiplicity within each family will be controlled using the Benjamini–Hochberg false discovery rate procedure. At the pair level, binomial logistic regression models will examine whether dyadic characteristics predict total pair accuracy (0–20 trials). Primary predictors include emotional bond strength, communication frequency, genetic relatedness, and physical separation during the task. Because a substantial proportion of pairs were co-located, physical separation will be modeled using a two-part parameterization that distinguishes co-location from non-zero separation and evaluates log-transformed distance among non-co-located pairs. Both the co-location indicator and the log-transformed distance term will be entered simultaneously into the same regression model. A secondary model will evaluate whether prior self-reported telepathic or telesomatic experiences are associated with performance, adjusting for relational characteristics. These analyses are hypothesis-generating rather than confirmatory and are intended to identify patterns of association that may inform subsequent preregistered laboratory replication studies incorporating physiological measures. The overarching aim is to determine whether stable individual or dyadic characteristics are associated with above-chance performance under controlled remote testing conditions and to refine selection strategies for future mechanistic investigations. Preregistered Analysis Plan Data Structure The dataset consists of 200 individual participants nested within 100 dyads. Each participant completed 10 Receiver trials in Phase 4. Pair-level analyses use a dataset collapsed to one row per dyad (100 observations), with total correct responses per pair ranging from 0 to 20. Primary Outcomes Individual-level outcome: Number of correct responses (0–10) per Receiver block. Pair-level outcome: Total number of correct responses (0–20) per dyad. All models will use binomial logistic regression with the number of trials specified as the binomial denominator. All regression tests will be two-sided. Individual-Level Models Individual-level analyses will be conducted using mixed-effects binomial logistic regression with a random intercept for PairCode to account for non-independence within dyads. Continuous predictors will be z-standardized prior to analysis. Age will be treated as continuous. Sex will be coded as binary. Race will be coded as a binary indicator of European ancestry endorsement versus non-endorsement. Meditation history will be evaluated as a prespecified focal predictor in a minimally adjusted model including demographic covariates (age, sex, race). Meditation will then be retained as an adjustment covariate in all domain-specific models. The following model families will be evaluated separately: Psychosocial Model Predictors: Boundary Thinness, Toronto Empathy Scale, Paranormal Belief, Paranormal Experience Covariates: Age, Sex, Race, Meditation Phenomenology Model Predictors: Noetic Signature Inventory factor scores Covariates: Age, Sex, Race, Meditation Personality Model Predictors: Big Five traits (E, A, C, N, O) Covariates: Age, Sex, Race, Meditation Interconnectedness Model Predictors: Interconnectedness measures Covariates: Age, Sex, Race, Meditation Performance Model Predictors: Objective performance on related tasks Covariates: Age, Sex, Race, Meditation No post hoc removal of prespecified predictors based on statistical significance will be performed within model families. All prespecified predictors will be retained unless model convergence fails. Multiplicity will be controlled within each model family using the Benjamini–Hochberg false discovery rate procedure applied to focal predictors within that family. Multiplicity correction is applied within, but not across, prespecified model families due to their conceptual independence. Pair-Level Models Pair-level analyses will use binomial logistic regression without random effects. Continuous predictors will be z-standardized prior to analysis. Physical separation during the task will be modeled using a two-part parameterization: Binary indicator of co-location (TaskDistance = 0 vs > 0) Log-transformed TaskDistance among non-zero distances Both components will be included simultaneously in the same regression model. Primary relational model predictors: Emotional Bond (continuous) Communication Frequency (continuous) Genetic Relatedness (treated as continuous percent shared genes and modeled as linear on the log-odds scale) Co-location indicator Log-transformed TaskDistance Secondary endorsement model predictors: Prior Telepathy (binary) Prior Telesomatic Experience (binary) These models will adjust for Emotional Bond, Genetic Relatedness, and distance parameters. Multiplicity will be controlled within each pair-level model family using the Benjamini–Hochberg false discovery rate procedure. Missing Data Models will be estimated using complete-case analysis for each model family. Participants or dyads with missing data on predictors included in a given model will be excluded from that specific model but retained for other analyses where data are complete. No data imputation will be performed. Model Reporting and Interpretation All model coefficients will be reported as odds ratios with 95 percent confidence intervals. Adjusted p-values following false discovery rate correction will be reported within each model family. Statistical interpretation will emphasize effect sizes and uncertainty rather than dichotomous significance thresholds. Results will be interpreted as exploratory associations rather than confirmatory hypothesis tests.