Structural and longitudinal analysis of cognitive social networks in dreams

Hye Joo Han, Purdue University

Abstract

Cognitive social networks were constructed from long dream series of five individuals and from another individual, whose waking-life social network was also constructed from a questionnaire. Network parameters for the cognitive social networks of the first five individuals indicate that dream social networks are structured systematically; transitivity, assortativity and giant component proportion were monotonically related. Negative assortativity found in the dream social networks differentiated dream social networks from real-world social networks, which tend to have positive assortativity. For the sixth individual, dream and waking-life social networks composed of the same people were compared. Discontinuity between dreams and waking life was found in comparisons of edges (i.e., acquaintances of people) and comparisons of network parameters. However, continuity between dreams and waking life was found in edges with higher weights and centrality measures for vertices; strong relationships considerably overlap, moreover, the importance of people correlates between the dream and waking-life social networks. Lastly, dynamics of dream social networks were investigated using longitudinal analysis and link predictions. Five types of longitudinal analysis on four networks in different time intervals found persistent people and relationships and changes in persistent people and overall networks. Link prediction methods predicted remaining and newly formed edges in dream networks from earlier dream networks more precisely than random predictors. Varied performances of the different link prediction methods gave better understanding about how cognitive social networks evolve in dreams.

Degree

Ph.D.

Advisors

Schweickert, Purdue University.

Subject Area

Quantitative psychology|Cognitive psychology

COinS