The architecture of spontaneous thoughts and experiences: a graph theory approach.
Peter Coppola, Pilleriin Sikka, Katja Valli, Jarno Tuominen, Antti Revonsuo, Ville Loukola, Ryan Bernstein, Nanna Strid, Manuela Kirberg, Alejandro Ezquerro‐Nassar, Jennifer Windt, Valdas Noreika, Natalia Mota, Tristan Bekinschtein
PsyArXiv April 21, 2024 preprint DOI: 10.31234/osf.io/wejs4 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Observational study using repeated measures and graph-theoretic language analysis Qualitative |
|---|---|
| Sample size | 176 |
| Population | 176 individuals providing mind-wandering and dream reports |
| Key findings | Dream reports had more complex structure, with more thematic repetitions, local cliques, and global integration, whereas mind-wandering reports had fewer word repetitions and a more verbose structure. Report structure carried individual-specific information, and word centrality in dreams predicted depression symptoms. |
Abstract
The language people use in everyday life provides a window into the mind. Mind-wandering and dreams have been thought to reflect unique individual differences and mental health. Here we use a large dataset of mind-wandering (n=1619) and dream (n=1434) reports from 176 individuals in conjunction with graph theory applied to natural language. We find that dream reports have a more complex structure, while mind-wandering reports have fewer word repetitions and more verbose structure, with essential nodal points in the narrative flux. Dream reports tend to have more thematic repetitions, local cliques, and global integration. Capitalising on a repeated measures design, we found that the structure of dream and mind-wandering reports contains individual-specific information. Finally, we find that word centrality in dreams is predictive of depression symptoms. Thus, this approach is sensitive to individual differences, quantitatively differentiates two distinct contents of consciousness, and seems promising for cost-effective analyses of large naturalistically occurring qualitative datasets.