The content and structure of dreams are coupled to affect

Abstract: Dreams offer a unique window into the cognitive and affective dynamics of the sleeping and the waking mind. Recent quantitative linguistic approaches have shown promise in obtaining corpus-level measures of dream sentiment and topic occurrence. However, it is currently unclear how the affective content of individual dreams relates to their semantic content and structure. Here, we combine word embedding, topic modeling, and network analysis to investigate this relationship. By applying Discourse Atom Topic Modeling (DATM) to the DreamBank corpus of >18K dream reports, we represent the latent themes arising within dream reports as a sparse dictionary of topics and identify the affective associations of those topics. We show that variation in dream report affect (valence and arousal) is associated with changes in topical content. By representing each dream report as a network of topics, we demonstrate that the affective content of dream narratives is also coupled to semantic structure. Positively valenced dream reports exhibit more coherent, structured, and linear narratives, whilst negatively valenced dreams have more narrative loops and dominant topics. Topic networks of high arousal dream reports are structurally dominated by few high arousal topics and incoherent topical connections, whereas low arousal dream reports contain more loops. These findings suggest that affective processes are associated with both the content and structure of dreams. Our approach showcases the potential of integrating natural language processing and network analysis with psychology to elucidate the interplay of affect, cognition and narrative in dreams. This methodology has broad applications for the study of narrated experience and psychiatric symptomatology.

• The paper reveals a strong coupling between affective dimensions (valence and arousal) and dream semantics using advanced NLP and network analysis.
• It employs Discourse Atom Topic Modeling to identify 130 interpretable topics, demonstrating how emotional states influence narrative coherence and structure.
• The findings have implications for psychiatric research and offer a novel framework for analyzing subjective experiences in dreams.

The Content and Structure of Dreams are Coupled to Affect

Quantitative Analysis of Dream Reports

This study applies computational linguistics and network analysis to explore the relationship between affective content, semantics, and narrative structure in dreams. By utilizing over 18,000 dream reports from the DreamBank corpus, a combination of NLP techniques—specifically, Discourse Atom Topic Modeling (DATM)—alongside network analysis reveals intricate links between the affective dimensions of dreams (valence and arousal) and their thematic content.

Using word embeddings and topic models, the authors identify 130 interpretable topics that represent dream semantics along valence and arousal dimensions. Correlations between dream report affect and semantic content suggest distinctive variations in topic prevalence according to their affective associations.

Dream Semantics and Structure

The study goes beyond lexical occurrences, examining semantic connections within dreams. Dream narratives are characterized by topic-based network structures. Positively valenced narratives demonstrate coherence, linear progression, and structured networks, whereas negative dreams are marked by fragmentation, prevalent thematic loops, and the dominance of specific topics.

Negative affect and high arousal are associated with semantic incoherence, confirming theories regarding affective dysregulation reflected in dream content. Moreover, the structural analysis of topic networks highlights an intriguing pattern: positive and coherent connections are predominant in less intense dreams, contrasting with the dominance and disarray observed under negative or high arousal conditions.

Network-Based Insights

The network analysis uncovers small-world characteristics of dream topics, indicating a balance of semantic connections. For example, the mean semantic coherence, path length, and clustering coefficients reveal how narrative organization shifts with affective states. Furthermore, network modularity and connectivity describe shifts in narrative linearity and organization depth, linked to affect regimens within dream sequences.

Relationships between affective content and semantic structures are validated using control analyses, which replicate these findings across model variations and using alternative datasets like waking reports from the "Diary" subreddit. Discrepancies between dream and waking narratives further emphasize the unique interplay between dream affect, semantics, and structure.

Implications and Future Directions

By combining DATM and network analysis, this paper proposes a robust methodological framework for examining narratives beyond traditional lexical analysis. This model has potential applications in psychiatric research, helping dissect subjective experiences, such as those documented during altered states or psychiatric evaluations.

Future work could aim to incorporate physiological data to enhance understanding of dream narratives in relation to neural correlates and sleep disorders. Additionally, exploring context-specific semantic and affective influences within dream reports could refine this approach's sensitivity and expand its application to related fields in cognitive and cultural studies.

Conclusion

This research emphasizes how integrating quantitative linguistic models with network analysis provides vital insights into the interplay of affect, cognition, and semantics in dream narratives. The findings not only advance methodologies for studying dreams but also set the stage for more comprehensive investigations into narrated experiences, contributing to mental health diagnostics and understanding complex cognitive phenomena.