How Cannabis Causes Paranoia: Using the Intravenous Administration of Δ9-Tetrahydrocannabinol (THC) to Identify Key Cognitive Mechanisms Leading to Paranoia

Daniel Freeman; Graham Dunn; Robin M. Murray; Nicole Evans; Rachel Lister; Angus Antley; Mel Slater; Beata Godlewska; Robert Cornish; Jonathan Williams; Martina Di Simplicio; Artemis Igoumenou; Rudolf Brenneisen; Elizabeth M. Tunbridge; Paul J. Harrison; Catherine J. Harmer; Philip Cowen; Paul D. Morrison


Schizophr Bull. 2015;41(2):391-399. 

In This Article


Paranoia is a key psychotic experience, distributed as a quantitative trait in the population, which requires explanation in its own right. This is the first experimental study testing the causal effects of THC on paranoia specifically, and it also determined the mechanism of action, taking advantage of the 90-minute period for which the drug was active in participants. It included an extensive assessment battery of paranoia, using a real life social situation, an immersive virtual reality test, self-report questionnaires, and an interviewer assessment. The study clearly establishes that THC causes paranoia in vulnerable individuals.

THC also led to the occurrence of anomalous experiences, anxiety, worry, depression, and negative thoughts about the self. These factors—highly plausible candidates derived from a theoretical model—fully explained the increase in paranoia. This is evidence for their role in causing paranoia. The clear clinical implication is that reducing negative emotion in patients with delusions, eg, by reducing the tendency to worry, testing out anxious fears, and increasing self-confidence, will lead to improvements in paranoia.[4] Also, the identification, normalization, and reduction of subtle anomalies of experience (eg, by reducing triggers and learning to tolerate the confusing sensory experiences) are clinically warranted. It is intriguing that negative affect and anomalous experience were so closely tied together. As in previous studies, THC also impaired working memory,[14] but there was no evidence that changes in working memory were responsible for the occurrence of paranoia.

There is a note of caution: the validity of the mediation analysis is dependent on the assumption that an increase in component 1 (anomalous experiences, anxiety, worry, depression, and negative thoughts about the self) is the intermediate variable on the pathway from THC administration to increases in paranoia. The data cannot exclude the possibility that the change in paranoia leads to the change in component 1. However, if we fit a regression model for component 1 with paranoia, experimental condition and baseline levels of paranoia and anxiety as covariates, we obtain a less parsimonious result. Although there is a highly significant effect of paranoia on component 1 (estimated regression coefficient: +0.342 with SE = 0.033; t = 10.46; P < .001), there remains an unexplained effect of THC that is not explained by paranoia (+0.355 with SE = 0.152; t = 2.34; P = .021). We conclude from both a theoretical perspective and the relative simplicity of the first model that paranoia is likely to be the distal outcome.

By the inclusion of an awareness condition, the study also made the first experimental attempt to block the misinterpretation considered central to paranoia. Contrary to prediction, this psychological manipulation did not decrease paranoia but perhaps had a paradoxical effect of exacerbating it, though statistical significance was not reached. The awareness condition may have increased sensitivity to paranoid thoughts. Interestingly, some participants in this condition reported being aware that their suspiciousness was due to having had the cannabis, ie, awareness affected judgments of the cause of such fears. As one participant in the awareness condition put it: "I'm always a bit sensitive but this intensified totally because I'm full of cannabis." The study indicates that misinterpretations in paranoia do not shift readily simply with provision of an alternative verbal explanation and that unintended consequences could arise from this approach.

The study could not address several issues. It is impossible to rule out biases in the estimation of direct and indirect effects arising from hidden confounding. There would have been benefits for learning about the effects of the psychological manipulation by including an additional condition in which the cognitive awareness training was also provided with the placebo; however, adding further randomization conditions causes practical difficulties since recruitment was labor intensive, with almost 2000 people being screened for participation. Arguably, greater effects would have been observed by increasing the dose of THC, though attrition from side effects would have compromised data collection. The dose had ecological validity since it is equivalent to about one strong cannabis cigarette. It would also have strengthened causal claims to have included randomization to other constitutes of cannabis, such as cannabidiol,[18] which may show opposite effects on the mechanisms underlying paranoia. The statistical strategy of PCA lessens the problem of multiple testing, which is a difficulty for the evaluation of multifactorial models, but does not provide results for individual assessments. The conclusions need to be validated by further experiments. As experiences such as paranoia begin to receive research attention in their own right, we expect to see the emergence of similar experimental studies that determine the causes underlying individual psychotic experiences.