UNCO (Uncovering patterns of unconscious reactions to fake content)

This project aims to investigate the type of patterns that may manifest subconsciously when a person is reading fake news – or viewing fake images.

The intention is to understand what makes people feel that content is fake at an emotional level, although it may not surface to the conscious level – and what types of manipulations go un-noticed. Interviewed subjects will be exposed to both human- and AI-generated content, and we will investigate the question of the potential differencesobserved between university-level and older adults. Our approach will rely on gathering data from devices such as those for eye-tracking, sweat and heart-rate monitors, but also recordings of the face of the interviewees while watching or listeining to (fake) content – which will be analyzed via emotion recognition algorithms. This multi-modal data will be the input to train an AI model that can extract emotional patterns which may be related to whether the content is true, recognizably fake, or perceived as fake at a subconscious level only. Depending on the actual correlations uncovered, this approach may point to an unobtrusive way to alert readers of online content of their potentlly fake nature via, for instance, readings on their smart watches.