Research: Psi



Testing the Psi Hypothesis

Our approach does not assume that psi exists, but treats the existence of psi as a hypothesis that can and should be tested scientifically. Included in the KPU’s psi research are meta-analytic reviews of the wider experimental literature, including ganzfeld ESP (Milton & Wiseman, 1999) and a comparison of clairvoyance and precognition (Steinkamp, Milton, & Morris, 1998).  Methodological guidelines have also been produced for ESP testing (Milton & Wiseman, 1997), and issues about the validity of different research approaches have been discussed (Stevens, 2004; Watt, 1994). This page describes two types of psi research that have been conducted at KPU: ganzfeld ESP studies, and studies into Experimenter Effects in Parapsychology. A reference list can be found at the foot of this page, and the Resources page contains an explanation of parapsychological terminology.

Ganzfeld ESP

What is the ganzfeld method? The ganzfeld is a mild sensory isolation procedure that is thought to be conducive to Extrasensory Perception (ESP).  Its use in parapsychology is based on a noise-reduction model of ESP that hypothesises that ESP functions like a weak signal that is ordinarily drowned out by surrounding well-understood signals such as somatic, visual, and auditory information.  By reducing external and internal sources of distraction, parapsychologists reasoned that any ESP “signal” would be more easily noticed.  Two individuals are usually involved in this procedure: the Sender will attempt to mentally communicate a randomly-chosen “target” to the Receiver.  The sender and receiver are placed in separate acoustically shielded rooms.  The receiver wears translucent eye-shields and is bathed in red light.  The receiver also reclines in a comfortable chair and wears headphones that play “white noise” (an unpatterned sound like radio static).  Thus the aim is for the receiver to become mentally and physically relaxed, and to have unpatterned sensory stimulation.  Under such conditions, thoughts and mental images become more salient to the receiver. 

A computer is used to randomly choose a target such as a one-minute video-clip from a large selection of possible targets (e.g., from movies, documentaries, cartoons), and plays that clip repeatedly to the sender.  At the same time, the receiver reports out loud any thoughts or images that come to mind (the “mentation”), and these verbal reports are recorded.  Of course neither the experimenter nor the receiver has any idea of what target the sender is viewing.  At the end of the sending period, the sender remains in their room while the receiver views four video clips – the target plus three decoys.  The receiver’s task is to compare each clip to their mentation, and to select which of the clips most closely matches the mentation.  If no information transfer is taking place (this is the null hypothesis), then we would expect the receiver to correctly identify the clip that was viewed by the sender 25% of the time by chance alone.  If the target clip is correctly identified, this counts as a “hit”.  Over a number of trials, usually with different sender-receiver pairs and with different sets of targets and decoys, the actual hit-rate is compared with the chance expectation using standard statistical techniques.  Provided that other normal sources of information transfer have been ruled out, such as cheating or sensory leakage, ESP is inferred to have taken place if the target is correctly identified more often than chance expectation.

There are good methodological reasons for presenting the target along with three decoys.  Firstly, it controls for the process of subjective validation (Marks & Kamman, 1980) whereby when there is just a single target one is easily able to find similarities between aspects of the target and various mentation items.  Similarities will of course occur by chance alone, but with four different target possibilities there will be chance matches to each of the possible targets.  However, if ESP is taking place, one would expect there to be a greater number of matches (i.e., more similarities) between the actual target and the receiver’s mentation.  Secondly, having four target possibilities enables parapsychologists to know the exact likelihood of obtaining a hit by chance alone, and this enables statistical tests to be used to quantify the outcome of the study.  In typical real-world situations, the factors leading to a coincidence, say between a person’s dream and real-world events the following day, are so complex that it is practically impossible to give an accurate calculation of the odds of that coincidence.  This is one reason why parapsychologists tend to focus on laboratory methods such as the ganzfeld to investigate ESP.

What are the findings? Using this procedure, a number of KPU studies have looked at individual differences in scoring on the ganzfeld ESP task.  One theme that appears to be emerging is that individuals who regard themselves as “creative” (e.g., artists, musicians) tend to score more “hits” (to correctly identify the target from a set of four possibilities) than less creative individuals (e.g., Dalton, 1997; Morris, Cunningham, McAlpine, & Taylor, 1998; Watt, 2006).  This line of research may throw some light on the question of the conditions needed to demonstrate psi, and further studies are needed to understand why creative individuals seem to perform well at ESP tasks.

Experimenter Effects in Parapsychology

What are experimenter effects and why are they important? Many parapsychologists have suggested that the belief of the experimenter may influence the outcome of their study – such that sceptics tend to find what they expect, and so do believers. Indeed, some have claimed that the experimenter’s own psi may affect the outcome of the study. This is an important issue for parapsychology because without an understanding of what causes experimenter effects, parapsychologists will not be able to specify the conditions under which other scientists can replicate their findings.

How did you study this? A series of KPU studies (e.g.Watt & Ramakers, 2003) have looked at the question of experimenter effects in parapsychology. We selected a number of individuals who scored extremely high or extremely low on a paranormal belief questionnaire, and then trained them to administer a psi task to naive participants. So, the 'experimenters' were either strong believers or disbelievers in the paranormal. The psi task was a simple 'remote helping' task involving two sensorially isolated individuals - the 'helper' and the 'helpee'. The helpees sat in a sound-shielded room and were asked to focus their attention on a candle and to press a button every time they noticed they had become distracted from this focus. A computer recorded the number of self-reported distractions and the time that they occurred during the session. At the same time, in a distant room, the helper was following a randomised schedule of 'help' and 'no help' periods. During the help periods, the helper was asked to attempt to mentally assist the distant helpee to have fewer distractions on the task. Since the experimenter and the helpee did not know the times when the helper was attempting to help, one would expect there to be no systematic relationship between the helpee's distractions and whatever the helper was doing. The psi hypothesis, on the other hand, would predict that the helpee would have fewer distractions during those randomly-scheduled periods when the helper was thinking of them. The results for all sessions combined showed overall significant positive scoring on the psi task - that is, fewer distractions during help periods. More interestingly, when comparing sessions conducted by believer experimenters with sessions conducted by sceptics, the effect was entirely limited to those participants tested by believer experimenters. Participants tested by sceptical experimenters obtained chance results on the psi task.

What does this mean? The positive psi result could not be due to subtle cueing of the experimenters or helpees, because all were blind to the randomised condition manipulations that were taking place during the psi task. Sensory leakage was also ruled out by locating helpees and helpers in separate isolated rooms. Questionnaire measures suggested that participants’ expectancy and motivation were unaffected by their experimenters’ paranormal belief, raising the possibility that it was the experimenter’s psi that influenced the outcome of the study. Note, however, that other researchers have not yet attempted to replicate this finding. So, although it is statistically significant, the study's findings should be regarded as suggestive but not conclusive. If experimenter psi effects are real then this raises challenging questions not only for parapsychology but also for science in general. Traditionally the experimenter is regarded as an objective observer of the data, rather than being another participant in the study.



Dalton, K. (1997).  Exploring the links: Creativity and psi in the ganzfeld.  Proceedings of the 40th Annual Convention of the Parapsychological Association (pp. 119-134).  Brighton, UK, 7-10 August.

Marks, D., & Kammann, R. (1980).  The Psychology of the Psychic.  Buffalo, NY: Prometheus Books.

Milton, J., & Wiseman, R. (1997).  Guidelines for Extrasensory Perception Research.  Hatfield: University of Hertfordshire Press.

Milton, J., & Wiseman, R. (1999).  Does psi exist?  Lack of replication of an anomalous process of information transfer.  Psychological Bulletin, 125, 387-391.

Morris, R. L., Cunningham, S., McAlpine, S., & Taylor, R. (1998).  Toward replication and extension of autoganzfeld results.  In N. L. Zingrone, M. J. Schlitz, C. S. Alvarado, & J. Milton (Eds.)  Research in Parapsychology 1993, pp. 57-61.  Laneham, Md.: Scarecrow Press.

Steinkamp F., Milton J., & Morris R. L. (1998).  A meta-analysis of forced-choice experiments comparing clairvoyance and precognition. Journal of Parapsychology, 62, 193-218.

Stevens, P. (2004).  Experimental evaluation of a feedback-reinforcement model for dyadic ESP.  Journal of Parapsychology, 68, 65-92.

Watt, C. (1994).  Making the most of spontaneous cases.  In S. Krippner, (Ed.)  Advances in Parapsychological Research 7 (pp. 77-103).  Jefferson, NC: McFarland.

Watt, C., & Ramakers, P. (2003).  Experimenter effects with a remote facilitation of attention focusing task:  A study with multiple believer and disbeliever experimenters.  Journal of Parapsychology, 67, 99-116. Download pdf (170KB)

Watt, C. (2006). Research assistants or budding scientists? A review of 96 undergraduate student projects at the Koestler Parapsychology Unit. Journal of Parapsychology, 70, 335-355. Download pdf (162KB).

Rabeyron, T. & Watt, C. (2010). Paranormal experiences, mental health and mental boundaries, and psi. Personality and Individual Differences, 48, 487-492. Download pdf (Draft: 505KB).


Back to main Research page