Separating experience reporting from content recall in dream reports across sleep stages

Abstract

Abstract Contentful dream reports are less frequent after non-rapid eye movement (NREM) than rapid eye movement (REM) awakening. This gap could reflect fewer reports of any experience or less recalled content within those reports. Distinguishing them requires retaining white dreams, experiences reported without recalled content. We audited the questions and codes behind all 20 datasets in the Dream EEG and Mentation database. Fourteen supported comparisons of any reported experience, ten supported content recall within experience reports, and one mapping merged white dreams with negative reports. A Bayesian hierarchical model decomposed the gap within six datasets supporting both comparisons at both stages. Averaged equally across datasets, contentful reports were 21.4 percentage points less frequent after NREM awakening. Most of this primary gap came from fewer reports of any experience: 16.6 points (95% credible interval 9.6 to 22.9), compared with 4.9 (− 4.1 to 12.3) for less recalled content. The experience-report contribution remained positive across weighting, sample and model checks. Content contributions depended on the datasets and weights averaged; both contributions were clearly positive in a four-dataset subset with at least ten records per stage. Uncertain-answer coding shifted the balance between negative reports and white dreams. Contrasting contentful with negative reports combined experience reporting with content recall. Theories must explain why any experience is reported less often after NREM awakening. Neural classifiers targeting any reported experience should retain white dreams in the positive class and assess content recall separately.