Network properties of dream sources

Abstract

Vast research in the last decades has shed interesting light on a variety of heterogeneous systems, including the human brain, by studying their network properties. Our investigation aimed to see whether the system of dream sources and of the semantic links between them shares the network properties of the brain. The investigation was carried out by means of an appropriate graph representation of data obtained according to a protocol oriented to eliciting episodic dream sources. The main results were the following: most dream sources belonged to compact clusters; important hubs, i.e. sources very closely connected to numerous other sources, were present; the vertex-degree distribution presented two significant peaks; small-world properties were valid for most pairs of sources; forms of self-similarity or of partial self-similarity at different scales were observed; the graphs of dreams sources exhibited a very high level of integration between overlapping clusters.