Fig. 1: OC symptoms are linked to increased information gathering in a smartphone population study.
From: Indecision and recency-weighted evidence integration in non-clinical and clinical settings
a, The participants’ task was to determine which of two possible gems was more abundant (screenshot of the task on the Brain Explorer app, https://brainexplorer.net/). They were free to uncover as much evidence as they wished, by tapping on locations on the grid, until they decided to commit to a binary choice between one of two gems. An example sequence of samples and corresponding evidence accumulation process for cumulative evidence strength at the previous draw ESd−1 and the evidence strength update ΔES is depicted on the right. Here evidence strength at draw d ESd is quantified as the cumulative difference in evidence for the two gems with respect to the gem that is more abundant at draw d (for example, here 3 diamonds minus 2 yellow gems constitutes an evidence strength of 1 at draw 5). Cumulative evidence strength at the previous draw ESd−1 is defined as the lagged ES by one draw, and evidence strength update ΔES is quantified as the signed difference between the cumulative ES at draw d−1 and draw d. Participants rated their confidence after every binary choice, then received 100 points for correct responses or lost 100 points for incorrect responses. b, Within a sample of the general population (N = 5,237), individuals with higher OC symptoms sampled more information before committing to a decision (ρs = 0.040, P = 0.004). Data points pertaining to individuals with a self-reported OCD diagnosis are shown in orange. c, The probability of making a decision p(decide) was predicted from experimental factors as well as individual differences in OC symptoms using a general linear mixed model (GLMM, N = 5,237; data are presented as beta coefficients ± standard error (s.e.), scale adjusted to show main effects of experimental factors on the left and effects associated with individual differences in OC on the right y axis). The experimental factors (left y axis) comprised current draw number, cumulative evidence from the start of the game until the previous draw ESd−1, and the evidence strength update ΔES. More draws and higher ESd−1 and ΔES increased the probability of making a decision p(decide) (all P < 0.001). Individual differences in obsessive–compulsive symptoms (right y axis) were quantified as the main effect of the OCI-R score and the interactions of the OCI-R score with the experimental factors. The analysis showed that OC symptoms per se did not have a significant effect on p(decide) (P = 0.265), and that people with high OC symptoms did not weight the current draw number differently in deciding to commit to a choice (P = 0.406); instead, they weighted both previous and current evidence less (P < 0.001, highlighted by shaded area). The latter effect is illustrated in d. d, To obtain estimates of sensitivity to evidence at a single-participant level, we fit a general linear model (GLM) predicting p(decide) separately for each participant. Convergent with the significant interaction effect in the GLMM, those with higher OC symptoms had reduced beta coefficients for ΔES, that is, they weighed ΔES less in deciding to commit (ρs = −0.121, P < 0.001, N = 3,901 after removing outliers, defined as any participant with any beta values 1.5× interquartile range (IQR) above the third quartile or below the first quartile). e, Individuals with higher OC symptoms also rated their confidence in their decisions as being lower on average (ρs = −0.054, P = 0.002, N = 3,293, see exclusion criteria in Methods). f, A mediation analysis suggests that sensitivity to ΔES partially mediates the association between OC symptoms (OCI-R total score) and confidence ratings, such that lower confidence is partially explained by altered evidence integration (N = 3,293). All tests are two-sided and not corrected for multiple comparisons.
