How Can We Determine What’s Real? A Cognitive Assessment of User Trust Towards AI-Generated Images of Politicians

This is the file repository used for our study originally titled "How Can We Determine What’s Real? A Cognitive Assessment of User Trust Towards AI-Generated Images of Politicians," REB 2023-6848 at Dalhousie University, done through the NeuroCognitive Imaging Lab in partnership with the Cognition and Organizations Research Group. This study was later nicknamed "Political AI" for short, based on the phenomenon of interest. The repository consists of 23 (of 26 reported in the paper) electroencephalography (EEG) and behavioural recordings prepared while participants attended various images of politicians. The images came in one of four conditions: AI generated, AI generated with a truthfulness notification, human, and human with a truthfulness notification. The EEG recordings were used to compare event-related potentials, which ultimately revealed patterns associated with attentional and affective processes. The paradigm is provided and this set also contains 300 responses to an online behavioural study based on a similar paradigm called "online behavioral" provided in CSV format.