Overview

Sheldrake’s work on automated telephone telepathy testing represents a systematic effort to move parapsychological research into the digital age. Rather than relying on laboratory settings or experimenter-supervised trials, he developed computer-based systems that could be deployed online, reaching large and diverse participant populations while maintaining rigorous randomization and blinding. The shift from videotaped experiments to fully automated protocols addressed practical constraints: supervised trials are time-consuming and expensive, whereas automated systems allow participants to conduct tests in their own homes at their own pace.

The core methodological innovation was the use of mobile phones, email, and SMS messaging, communication channels already embedded in everyday life. By asking participants to guess the identity of an incoming caller or sender before answering or opening the message, Sheldrake created a naturalistic testing environment that did not require participants to adopt unfamiliar laboratory tasks. This approach also allowed researchers to collect data across geographic distances and time zones, expanding the potential sample size and diversity of the research.

Prevalence of Telephone Telepathy Experiences

Before developing automated tests, Sheldrake conducted questionnaire surveys in the United Kingdom, United States, Germany, and Argentina to establish the baseline prevalence of reported telephone telepathy experiences. The results were striking: an average of 85% of respondents reported having experienced apparent telephone telepathy, typically with people they knew well such as family members, friends, or close colleagues.¹ In the same surveys, 68% of respondents reported experiencing other kinds of telepathy.

These survey findings suggested that if telephone telepathy were a genuine phenomenon, it would be relatively common and accessible to ordinary people rather than confined to rare gifted individuals. This observation motivated Sheldrake to develop testing methods that could accommodate unselected, untrained participants and capture the conditions under which the effect naturally occurs, namely, communication with emotionally close contacts.

Early Videotaped Experiments

Before moving to fully automated systems, Sheldrake conducted a series of carefully controlled videotaped telephone telepathy experiments. In these trials, participants nominated four potential callers (typically people they knew well) and for each trial, the experimenter randomly selected one caller and asked them to call the participant. The participant then guessed the caller’s identity before answering the phone or hearing the caller’s voice.⁴

Across 271 videotaped trials, participants achieved a hit rate of 45%, significantly above the 25% chance level expected by random guessing (p < 1×10⁻⁶).² Notably, callers and participants were sometimes thousands of kilometers apart, ruling out explanations based on sensory cues or proximity. A replication conducted at the University of Amsterdam also yielded hit rates significantly above chance, providing independent confirmation of the effect.²

Sheldrake extended this approach to email telepathy, in which participants with four potential emailers attempted to guess the sender’s identity one minute before the email was sent. In 137 videotaped trials, the hit rate was 47%, again significantly above the 25% chance level (p < 1×10⁻⁶).² These early results provided strong evidence for the phenomenon and justified investment in developing automated systems to test it at larger scale.

Development of Automated Testing Systems

Sheldrake’s first automated email telepathy test, conducted with Leonidas Avraamides, allowed participants to register online with the names and email addresses of three senders.² A computer program randomly selected a sender and asked that person to send an email to the participant. The participant was then prompted to guess the sender’s identity, and the email was delivered only after the guess was recorded. Tests consisted of 6 or 9 trials per session.

Across 419 trials, including data from incomplete tests, the hit rate was 41.8%, significantly above the 33.3% chance level (p = .0001).² Interestingly, hit rates in incomplete tests were higher than in complete tests, suggesting that fatigue or habituation may reduce the effect. The highest hit rates were observed in participants aged 20–29 years. The effect size in these automated tests was lower than in the earlier videotaped experiments, possibly because participants were guessing about an email sent several minutes earlier rather than thinking about the sender simultaneously.

Sheldrake and colleagues subsequently developed automated tests for telephone telepathy using mobile phones.⁵ These systems maintained the core design: participants nominated known contacts, the computer randomly selected one to call, and the participant guessed the caller’s identity before answering. The automation eliminated experimenter bias and allowed unsupervised testing across large participant samples.

Meta-Analytic Findings

In 2025, Sheldrake, Stedall, and Tressoldi published a comprehensive meta-analysis synthesizing results from 15 published papers describing 26 telecommunication telepathy experiments conducted between 2003 and 2024.¹ This analysis pooled data across telephone, email, and SMS message experiments, all using similar randomized designs in which participants attempted to identify the sender or caller before receiving the communication.

The overall hit rate across all studies was significantly above chance level (p = 1×10⁻⁷).¹ Notably, the effect sizes in telecommunication telepathy were higher than those reported in ganzfeld and dream telepathy experiments, suggesting that the naturalistic context of mobile communication may be particularly conducive to the expression of telepathic ability.¹

The meta-analysis revealed several important moderating factors. Selected participants (those with prior belief in or experience of telepathy) showed significantly higher hit rates than unselected participants.¹ Hit rates were also significantly higher when callers and participants shared an emotional bond, consistent with earlier findings that the effect is strongest with close relationships.¹ Importantly, there was no significant difference between results obtained by Sheldrake and colleagues and those from independent replications, indicating that the effect is not an artifact of experimenter expectancy or selective reporting.

A critical finding concerned precognitive conditions. When participants attempted to anticipate calls before the caller had been randomly selected, hit rates fell to chance level.¹ This pattern suggests that the effect depends on the caller’s intention or attention directed toward the participant, rather than on the participant’s ability to access future information. The distinction between simultaneous telepathy and precognition has important theoretical implications for understanding the mechanism underlying the phenomenon.

Conference Call Format and Limitations

In 2024, Sheldrake and Stedall published a detailed comparison of four new automated telephone telepathy test formats, exploring different ways to structure trials while maintaining user-friendliness and experimental rigor.³ One approach involved a “conference call” format in which three participants who knew each other were linked together continuously. In each trial, the receiver was selected at random, the other two participants were muted, and one was randomly selected as the caller and asked to think about the receiver before being connected.

Surprisingly, none of the conference call tests yielded hit rates significantly different from the 50% chance level expected when choosing between two callers.³ Sheldrake and Stedall hypothesized that the failure of this format stemmed from the fact that all three participants remained continuously engaged with the test system. This continuous engagement may have created cognitive interference or disrupted the natural conditions under which telepathic effects emerge.

By contrast, a second type of test in which trials were spaced out over longer time periods and callers and receivers went about their normal lives between trials produced significant results. Across 266 trials, the hit rate was 57% (p = .01).³ This finding suggests that the effect is enhanced when participants are not consciously focused on the experiment and when the communication occurs in a naturalistic context rather than a continuous laboratory-like setting.

Future Directions: Intuition Training Applications

Based on the patterns observed in automated testing, Sheldrake and Stedall proposed developing an intuition training application that would integrate with participants’ regular calls and messages.³ Rather than requiring participants to engage in a separate experimental task, such an app would prompt users to guess the identity of an incoming caller or message sender as part of their everyday mobile phone use. This approach would be more user-friendly, allow participants to practice their intuitive skills in a naturalistic context, and enable researchers to identify talented participants for more rigorous testing.

The meta-analysis concluded that “research on telecommunication telepathy could become an increasingly fruitful area for psi research, especially in conjunction with automated intuition training apps.”¹ This vision reflects Sheldrake’s broader strategy of moving parapsychological research out of specialized laboratories and into the contexts where anomalous experiences naturally occur, in this case, the ubiquitous mobile communication devices that structure modern social life.

The development of automated systems also addresses a longstanding challenge in parapsychology: the difficulty of recruiting and retaining large participant samples for extended experimental protocols. By embedding testing in everyday technology and allowing asynchronous, unsupervised participation, Sheldrake’s approach makes it feasible to collect data from thousands of participants over months or years. This scaling up of sample size and trial volume increases statistical power and reduces the likelihood that positive results reflect chance fluctuations or publication bias.