Individual traits and experiences predict the content of dreams

Abstract Dreams are universal yet highly idiosyncratic experiences. While memories and personal concerns are known to influence dream content, how such influences evolve over time and how stable individual traits shape dreaming remain unclear. Here, we systematically quantified the semantic structure of dreams in a large, multimodal dataset comprising 3366 reports of dreams and waking experiences collected from 207 adults between 2020 and 2024, alongside demographic, cognitive, psychometric, and sleep measures. To this end, we combined large language model-assisted evaluation of hypothesis-driven semantic dimensions and a data-driven lexical domain approach. Relative to waking reports, dreams shifted from self-referential, thought-centered narratives to perceptual experiences dominated by visuo-spatial details, multiple characters, and bizarre events. Stable traits, including attitude toward dreaming, mind-wandering propensity, and subjective sleep quality, selectively influenced dream content. A second, independent dataset collected during the first 2020 COVID-19 lockdown (80 participants) allowed us to examine the impact of a major external stressor on dream semantics. During lockdown, dreams showed increased references to limitations and heightened emotional intensity, effects that gradually normalized over the following years. These findings demonstrate that stable individual traits and incidental experiences jointly shape dream semantics.