LODE – The Language Of Dreams

Type of project: National  |  Start date: 05/10/2023  |  End date: 04/10/2025

Every night, when we fall asleep, we progressively disengage from the external environment and cease to experience it. Yet, even during sleep, we may be exposed to a wide gamut of internally generated experiences, ranging from simple abstract thoughts to complex movie-like dreams. These experiences typically present clear aspects of continuity with those of our waking self. Indeed, our dreams draw on previously acquired memories and beliefs, and are tightly related to our mental and physical health.

Alterations in the frequency or content of dreams are in fact known to accompany, or even precede, the waking manifestation of clinical symptoms related to psychiatric and neurological disorders.

Nonetheless, until now, the analysis of dream experiences in pathological conditions has been mainly employed as a tool to test hypotheses about the generation and functions of dreaming, whereas its potential value as a clinical tool received little attention. Among the reasons for this is the fact that dreams have been long regarded as psychological objects, and as such have been investigated mainly through subjective scales and questionnaires, which entail relative risks of evaluation error and experimenter bias and also limit the possibility to study large data samples.

Here we propose that these limitations could be overcome through the adoption of standardised, information technology (IT-)based data collection approaches and the use of quantitative, automated and reproducible analysis methods borrowed from the fields of artificial intelligence (AI) and computational linguistics.

Objectives and results

The principal objectives of the present project are:

• to deepen our knowledge regarding the neurophysiology of dreaming, and in particular the anatomo-functional mechanisms involved in the generation and delineation of dream imagery
• to investigate the possible translational value of natural language processing applied to verbal dream reports, with a specific focus on neurological conditions including epilepsy, brain damage, and neurodegenerative disorders
• to develop an IT-infrastructure for the collection, storage and analysis of dream reports and of health-related information.

To these aims, we will collect and analyze multiple dreams and we will create an open access, expandable, high-quality dream-repository that will remain available for future basic and translational research and applications.

AI-based techniques will be used to automatically extract quantitative features reflecting the structure and content of dreams, and to determine whether changes of oneiric experiences present any reliable and reproducible association with cognitive and brain anatomo-functional alterations.

In this respect, the project could yield potential breakthroughs in key applications of social, clinical, and economic interest.