Discovery Lab Answers

Caroline Watt, PhD holds the Koestler Chair in Parapsychology at the University of Edinburgh, succeeding Robert Morris in 2004 ^[2][3]. Her research spans laboratory ESP, precognitive dreaming, paranormal belief, science communication, and methodological reform — a breadth reflected across the sources retrieved for this question.

Experiments

Precognitive ganzfeld (Watt et al., 2020). Watt led a pre-registered study at Edinburgh testing whether selected participants could identify a future video target under ganzfeld (sensory-attenuation) conditions ^[4]. Collaborators were Emily Dawson, Alisdair Tullo, Abby Pooley, and Holly Rice. The study was designed as the first contribution to a registration-based prospective meta-analysis of ganzfeld ESP studies, building on Watt’s own 2017 registry protocol. A secondary hypothesis tested whether altered states of consciousness (ASC) during the ganzfeld session would predict psi task performance ^[4].

Sender-receiver meta-analysis (Pooley, Murray, and Watt, 2023). Watt co-authored, with Abby Pooley and Aja Murray, a meta-analysis examining five previously uninvestigated moderating factors related to the sender-receiver dynamic in the telepathy ganzfeld ^[1]. The analysis drew on a publicly available ganzfeld database (Tressoldi, 2019) and the final dataset comprised 41 studies ^[1].

Comics and skepticism (Wiseman, Collver, Worth, and Watt, 2021). Watt co-authored a study with Richard Wiseman, Jordan Collver, and Rik Worth investigating whether comics could promote skeptical attitudes toward paranormal claims ^[3]. This sits within a skeptic-proponent collaborative tradition that Watt has explicitly argued for ^[3].

Free-response ESP training (with Delanoy and Morris). The ESP-Nexus page for Deborah L. Delanoy documents a free-response training study by Delanoy, Morris, and Watt that found no significant ESP scoring on any formal measure, while informal practice trials scored above chance — a result the authors declined to claim because those trials lacked security against sensory access to targets.

Statistical reform (Watt and Kennedy, 2019). Watt and J. E. Kennedy published a correspondence in the Journal of Parapsychology responding to a special issue of The American Statistician on statistical inference, arguing that confirmatory research designs in parapsychology need to be strengthened, falsifiable hypotheses more clearly specified, and preregistration treated as obligatory for confirmatory (not just exploratory) work ^[5].

Science communication (personal essay, 2022). Watt’s autobiographical essay in Zeitschrift für Anomalistik documents her career trajectory, including winning the Perrott-Warrick Senior Researcher award in 2010 and her promotion to the Koestler Chair ^[2].

Methodology

Pre-registration and prospective meta-analysis. The 2020 ganzfeld study was pre-registered and designed explicitly as a contribution to a prospective meta-analysis ^[4]. The pre-registration specified two hypotheses: (1) that participants would exceed chance on the precognition task; and (2) that ASC scores would correlate positively with psi performance. Three experimenters each conducted 20 trials, for 60 trials total, with participants being female final-year undergraduate psychology students rated as believing the psi hypothesis would be supported ^[4].

Target pool and direct-hit scoring. The ganzfeld protocol used a four-alternative forced-choice design: participants viewed and ranked four targets, with a “direct hit” defined as the target ranked first. Mean Chance Expectation (MCE) is 25% ^[4].

Participant selection. Participants were selected — recruited on criteria suggesting psi-conduciveness — rather than drawn from an unselected population, following the tradition of testing individuals who self-identify as having psi experiences or express positive belief ^[4].

Consciousness measurement. The Phenomenology of Consciousness Inventory (PCI) was administered to capture 12 dimensions of altered state experience, linked to Hypothesis 2a ^[4].

Meta-analytic design (Pooley, Murray, and Watt, 2023). The 2023 meta-analysis used two parallel mixed-effects models — one treating study hit rates as a binomial mean, one treating them as a proportion — with five binary moderators representing sender-receiver factors ^[1]. Only studies conducted after the Joint Communiqué, with one session per day and receivers rating targets, were included ^[1]. Homogeneity analysis was computed automatically by the model function ^[1].

Science communication design (Wiseman et al., 2021). The skepticism study used a three-condition between-subjects design (text, magic comic, plain comic) with pre-planned analyses, between-groups ANOVAs, and power calculated for a small-to-medium effect (d = 0.35, power = 0.9, two-tailed) ^[3]. The study was pre-planned following a similar design used in Wiseman, Houstoun, and Watt (2020) ^[3].

Methodological advocacy. Watt and Kennedy (2019) argued explicitly for stronger confirmatory research standards: pre-registration should be mandatory for confirmatory research, and exploratory research not pre-registered should be presumed entirely exploratory ^[5]. They further argued that “statistical significance enables bad science; statistical thinking enables good science” ^[5].

Data

The structured evidence block is not present for this question, so no audited quantitative table is rendered below. Statistics below come directly from the numbered primary sources; where a surrounding sentence is intact and attribution is unambiguous, the figure is reported.

Watt et al. (2020) — precognitive ganzfeld.

Twenty-two direct hits were obtained across 60 trials, a 37% hit rate against a 25% MCE, with an exact binomial p = .03 (one-tailed) ^[4]. This constitutes a statistically significant result on the primary hypothesis. The secondary hypothesis — that PCI scores would correlate with psi performance — was not supported: correlations across the 12 PCI dimensions were either near-zero or small, and no significant relationship was found ^[4].

Pooley, Murray, and Watt (2023) — sender-receiver meta-analysis.

Both the binomial mean model and the proportion model indicated a significant effect of the five sender-receiver moderators overall, and two specific factors emerged as significant ^[1]. The retrieved excerpts do not carry the individual moderator-level statistics; the paper itself (doi:10.31156/jaex.23878) contains the full breakdown.

Wiseman, Collver, Worth, and Watt (2021) — comics and skepticism.

Total N was 517 across three conditions (Text: N = 171, Magic Comic: N = 179, Comic: N = 167). Groups did not differ significantly in age or time to complete the study ^[3]. The groups differed significantly in how interesting and entertaining they found the material; the Comic condition was rated significantly more entertaining than the Magic Comic ^[3].

Free-response ESP training (Delanoy, Morris, and Watt).

No significant ESP scoring was found on any formal measure. Informal practice exercises scored above chance, but no claim was made for that result given the absence of security controls against sensory leakage.

Prior ganzfeld landscape (as summarized in Watt et al., 2020).

The 2020 paper situates its own result within prior precognitive ganzfeld work:

These figures are drawn from Table 1 of Watt et al. (2020) ^[4] as context for their own study design, not as an independent meta-analysis.

Skeptical critiques

What critics argue. The 2020 ganzfeld study itself records that Milton and Wiseman (1999) had previously concluded, from a meta-analysis of post-autoganzfeld studies, that the combined results were at chance — a 27% hit rate against 25% MCE — and therefore did not support ESP ^[4]. This served as the baseline against which subsequent positive results, including Watt et al.’s own, have to be read. Within the Wiseman, Collver, Worth, and Watt (2021) science-communication study, the authors acknowledged that the lack of prior research in the area limited pre-study effect-size estimation, which in turn affected confidence in the adequacy of the sample size ^[3].

What the experimental data show. Storm et al. (2010), cited in Watt et al. (2020), reanalyzed ganzfeld studies published between March 1997 and 2008 and found an overall statistically significant result of 32% hit rate after discarding one study judged a statistical outlier ^[4]. When all studies across the Milton-Wiseman and Storm et al. databases are combined, Watt et al. (2020) reports an overall hit rate of 30% ^[4]. Watt et al.’s own 2020 study returned 37%, significant at p = .03, but with 60 trials it is a single, modest-N contribution and was designed from the outset to be combined with others in a prospective meta-analysis rather than interpreted in isolation ^[4].

Analysis. The pre-registration of Watt et al. (2020) was a direct methodological response to the kind of criticism that had attached to earlier ganzfeld research — namely, that post-hoc flexibility in analysis and publication bias inflated apparent effects. Whether pre-registration fully addresses those concerns, and whether the 37% result replicates in subsequent contributions to the prospective meta-analysis, are questions the published record at the time of these sources leaves open. The null finding on the ASC hypothesis within the same study is a within-study disconfirmation that the authors report candidly ^[4]. Independent replication of the specific sender-receiver moderators identified in Pooley, Murray, and Watt (2023) has not been published in the sources retrieved for this question.

For Watt’s full research profile, including her work on paranormal belief and distant healing, see her page at Caroline Watt, PhD — ESP-Nexus.