A New Theoretical Model of Human Attachment Based on Active Inference and Psychotherapeutic Applications in Virtual Environments

Focused on attachment theory, the doctoral thesis presents an innovative model of the human attachment system grounded in the Active Inference framework. The work offers a critical review of the state of the art concerning attachment development, from the prenatal period to the first two years of life. Through a combination of neurobiological, cognitive, and genetic perspectives, the thesis explores how early experiences influence the formation of attachment patterns, with significant implications for mental health throughout life. The Active Inference framework, founded on principles of homeostatic and predictive regulation, is used to unify under a neat theoretical framework the biological and psychological phenomena involved in attachment. The application of the Active Inference framework enables the translation of theoretical concepts into practical applications, particularly in the field of digital technologies and artificial intelligence. This leads to the proposal of Virtual Psi, a virtual agent integrated within the Virtual Hospital model. In this context, Virtual Psi and virtual patients interact with real patients and therapists, with the specific goal of strengthening secure attachment patterns and providing daily, personalized, and accessible support. The proposed care model is not intended to replace traditional therapy but to complement it, overcoming some of the limitations of classical psychotherapeutic settings.