Abstract
Previous work in our lab shows that the N300 event-related potential component is sensitive to statistical regularity (e.g., exemplar representativeness) under conditions of full attention, but the degree to which this effect depends on attention remains unknown. We address this gap by measuring brain responses to unattended natural scene stimuli. Twelve subjects played an attentionally demanding continuous motion tracking game at fixation using a joystick while images of “good” and “bad” exemplars of natural scenes (beaches, cities, forests, highways, mountains, and offices) were presented in the background. Good exemplars were rated by a large sample of independent observers as highly representative of their category whereas bad exemplars were rated as belonging to the category but poorly representative of it. We also manipulated exemplar category frequency within block by using a completely randomized trial structure in Experiment 1 (50% good exemplars) and a scene category blocked trial structure in Experiment 2 (80% good exemplars). All subjects participated in both experiments. Preliminary results replicate our prior finding of a larger N300 in response to bad compared to good exemplars, and further extend this effect to ignored stimuli. The results indicate that full attention is not necessary to assess statistical regularity in scene structure. Moreover, the N300 effect seems to be larger when we combine temporally global statistical regularity (exemplar representativeness) with temporally local statistical regularity (predictability within the experiment: Expt. 1 vs. 2). Together, these results suggest that the N300 may index a form of implicit template matching that benefits from both immediate context and longer-term prior experience.