Poster A16, Saturday, March 25, 5:00 – 7:00 pm, Pacific Concourse
Observing model-based control of emotion-triggered attention with steady-state visual evoked potentials
deborah talmi1, Matthias Wieser, Martina Slapkova; 1University of Manchester, 2University of Rotterdam, 3University of Manchester
The investigation of distinctions between model-based and model-free learning has benefitted greatly from an innovative paradigm where the two processes can be distinguished (Daw et all. 2011). Here we describe a novel paradigm that targets underlying differences between these two modes of learning. Through aversive classical conditioning, participants learned to fear some Navon figures. During learning, the local and global dimensions of the figures was consistent. At test, novel figures were presented crossing the local and global dimensions of threatening and safe figures. While a model-free system cannot distinguish between the aversive value of these figures, the model-based system could, in principle, utilise the verbal instruction participants received just before test, stating which dimension was relevant to the conditioning contingency (e.g. ‘shape’). We asked whether we could see differences between instructed threatening and safe novel shapes in the steady-state visual evoked potential, an EEG signal known to be modulated by visual attention to threat. Replicating previous work, the SSVEP was modulated by model-free learning; its amplitude was higher for learned threatening figures compared to learned safe figures. The SSVEP was also modulated by model-based learning, with a higher amplitude for instructed threatening compared to instructed safe figures, but only when the global dimension of the figure was the relevant dimension for the conditioning contingency. These results point to a previously unknown constraint on the ability of the model-based system to modulate attention.
Topic Area: EMOTION & SOCIAL: Emotion-cognition interactions