Neuronal activity in area V1 during contour grouping conforms to a growth-cone model of object-based attention

Attention can select elements of spatially extended objects and group them into coherent representations, but the processes responsible for object-based attention have remained unclear. Shape-selective neurons in higher areas of the visual cortex could contribute to object-based attention by providing feedback to lower areas to enhance the representation of image elements that belong to a relevant shape (Van der Velde and de Kamps, 2001). Another possibility is that attention gradually spreads according to the Gestalt rules, successively adding new elements to a growing object representation. Here we investigated the dynamics of object-based attention by recording the activity of area V1 neurons in a contour grouping task. Our results show that attention gradually spreads across the object representation as neuronal responses evoked by contour elements further along a relevant curve are enhanced at progressively later times, while responses evoked by the initial segments of the curve stay at an enhanced level. The speed of the attentional spreading process decreased at locations where the relevant curve came close to an irrelevant one. Our findings inspire a " growth-cone " model where attention spreads at multiple spatial scales by enhancing the responses of neurons with differently sized receptive fields, at a pace of approximately 50ms per receptive field.