Delusion-like beliefs and data quality: Are classic cognitive biases artifacts of carelessness?

Abstract

There is widespread agreement that delusions in clinical populations and delusion-like beliefs in the general population are, in part, caused by cognitive biases. Much of the evidence comes from two influential tasks: the Beads Task and the Bias Against Disconfirmatory Evidence Task. However, research using these tasks has been hampered by conceptual and empirical inconsistencies. In an online study, we examined relationships between delusion-like beliefs in the general population and cognitive biases associated with these tasks. Our study had four key strengths: A new animated Beads Task designed to reduce task miscomprehension, several data-quality checks to identify careless responders, a large sample (n = 1,002), and a preregistered analysis plan. When analyzing the full sample, our results replicated classic relationships between cognitive biases and delusion-like beliefs. However, when we removed 82 careless participants from the analyses (8.2% of the sample) we found that many of these relationships were severely diminished and, in some cases, eliminated outright. These results suggest that some (but not all) seemingly well-established relationships between cognitive biases and delusion-like beliefs might be artifacts of careless responding. General Scientific SummaryResearch suggests that cognitive biases play a key role in the development of delusion-like beliefs. For instance, participants who endorse such beliefs have been reported to "jump to conclusions" when performing abstract data-gathering tasks and to display a "bias against disconfirmatory evidence" when determining the best explanation for a scenario. However, the present study suggests that some (but not all) seemingly well-established relationships between cognitive biases and delusion-like beliefs might, in fact, be spurious-driven by careless responding in a subset of research participants.

Hallucination- and delusion-like experiences are associated with increased precision of sensory evidence in perceptual inference

“…We also attempted to replicate previous studies showing reduced sampling (e.g., fewer samples of evidence) in probabilistic reasoning, namely in the BT, associated with delusional ideation. After controlling for performance attentiveness, we reproduced Sulik et al’s (2023) 25 results in our subsample where we found no evidence of reduced sampling in probabilistic reasoning for participants higher in delusional ideation. This suggests that in subclinical psychosis, alterations in inferential processes attributable to sensory precision might be more prominent and detectable in the perceptual rather than in the cognitive domain.…”

“…We did not find any significant relationship between psychotic-like experiences and ADJ. Although this BT behavioural pattern has been shown in schizophrenia patients with delusions 17,19,45 , it has repeatedly failed to be replicated in individuals from the general population with high scores for delusional ideation 25,40,46 .…”

“…We used Amazon's Mechanical Turk service to invite a subsample of 200 participants from a larger cohort (N=1002) recruited by Sulik et al (2023) to take part in our online study. We limited our invitations only to participants who successfully answered at least two of three attention checks in Sulik et al (2023).…”

“…We used Amazon's Mechanical Turk service to invite a subsample of 200 participants from a larger cohort (N=1002) recruited by Sulik et al (2023) to take part in our online study. We limited our invitations only to participants who successfully answered at least two of three attention checks in Sulik et al (2023). The data from the draws to decision and graded estimates versions of BT (details below) and the Peters et al Delusion Inventory (PDI) 24 , a measure of delusion-like experiences, were collected by Sulik et al (2023) 25 .…”

“…We limited our invitations only to participants who successfully answered at least two of three attention checks in Sulik et al (2023). The data from the draws to decision and graded estimates versions of BT (details below) and the Peters et al Delusion Inventory (PDI) 24 , a measure of delusion-like experiences, were collected by Sulik et al (2023) 25 . In the present study, participants performed the RDM and completed the Cardiff Anomalous Perception Scale (CAPS) 26 and the Aberrant Salience Inventory (ASI) 27 , respectively measuring hallucination-like experiences and life attitudes related to aberrant salience.…”