Abstract
Introduction
That delusional and delusion-prone individuals “jump to conclusions” on probabilistic reasoning tasks is a key finding in cognitive neuropsychiatry. Here we focused on a less frequently investigated aspect of “jumping to conclusions” (JTC): certainty judgments. We incorporated rigorous procedures from experimental economics to eliminate potential confounds of miscomprehension and motivation and systematically investigated the effect of incentives on task performance.
Methods
Low- and high-delusion-prone participants (n = 109) completed a series of computerised trials; on each trial, they were shown a black or a white fish, caught from one of the two lakes containing fish of both colours in complementary ratios. In the betting condition, participants were given £4 to distribute over the two lakes as they wished; in the control condition, participants simply provided an estimate of how probable each lake was. Deviations from Bayesian probabilities were investigated.
Results
Whereas high-delusion-prone participants in both the control and betting conditions underestimated the Bayesian probabilities (i.e. were conservative), low-delusion-prone participants in the control condition underestimated but those in the betting condition provided accurate estimates. In the control condition, there was a trend for high-delusion-prone participants to give higher estimates than low-delusion-prone participants, which is consistent with previous reports of “jumping to conclusions” in delusion-prone participants. However, our findings in the betting condition, where high-delusion-prone participants provided lower estimates than low-delusion-prone participants (who were accurate), are inconsistent with the jumping-to-conclusions effect in both a relative and an absolute sense.
Conclusions
Our findings highlight the key role of task incentives and underscore the importance of comparing the responses of delusion-prone participants to an objective rational standard as well as to the responses of non-delusion-prone participants.
Acknowledgements
We thank Dr Michael Naef and Dr Bjoern Hartig for the opportunity to use the behavioural economics lab at Royal Holloway, University of London, and for their help with programming our task. We also thank Dr Victoria Bourne for advice on statistical analyses and Professor Todd S. Woodward for providing images that formed the basis of our stimuli.
Notes
1. The Bayesian probability (Bayes, Citation1763) of a particular lake, given a displayed fish of a particular colour, is given by:
2. The same pattern of results (significant interaction, robust when accounting for risk aversion) was found when PDI groups were defined by endorsement of PDI items or by scores on the separate PDI dimensions distress, preoccupation or conviction.