Visual representations outside of conscious awareness can support sensory preconditioning

Abstract

When one of two previously-paired neutral visual stimuli starts to subsequently predict reward, subjects treat the other neutral item as if it similarly predicts reward despite the lack of direct reinforcement for that particular visual stimulus. Termed sensory preconditioning, this phenomenon is often used to explore model-based learning. However, it has been suggested that sensory preconditioning may happen even for stimuli that are not consciously perceived. We tested this hypothesis in a decoded neurofeedback (DecNef) experiment. To form an association between a viewed stimulus and an unrelated non-conscious visual representation, online fMRI data was analyzed through a MVPA classifier while participants (N=5) viewed a fullscreen dot-motion display at 100% motion coherence. Visual feedback was given based on the likelihood that BOLD activity while participants viewed the dot-motion display represented an unrelated target image category (DecNef target), which was never consciously seen. After 3 days of neurofeedback, participants completed a betting task with feedback which reinforced participants to value the previously presented dot-motion display as a significant financial loss. Following this, participants completed another round without feedback in which a critical decision was made between two previously unseen objects: the DecNef target and a neutral control. Participants bet on the neutral control significantly more often than the DecNef target (individual 𝝌2’s, all p’s< 0.05 with 3 subjects showing the maximum possible effect at 𝝌2(1, N=30) = 30, p< 0.001), indicating successful preconditioning of a visual stimulus outside of conscious awareness. These results suggest associations can indeed be formed and conditioned between visual stimuli outside of conscious awareness, questioning whether consciousness is necessary for model-based learning. This opens a discussion on how these neurofeedback-driven subliminal visual presentations may complement traditional methods of rendering visual stimuli unconscious, such as: masking, continuous flash suppression, crowding, and other methods.