Abstract
It is generally assumed that the metrics of saccadic eye movements reflect the distributed nature of spatial coding in the Superior Colliculus (SC). Neurons in the sensory and motor maps of the SC have large and overlapping receptive / movement fields, and it is the simultaneous activity of populations of neurons coding for neighboring locations in space which determines where the eyes move. This drives the eyes relatively precisely to a singleton target, but as a counterpart, biases the eyes to land at an intermediate location between two simultaneously-displayed visual stimuli. Here, I will argue that distributed coding, as a general property of the saccadic system, accounts not only for saccades to singleton target stimuli, but also for eye movements in reading. First, I will report new data showing that the eyes move to different locations in words depending on the visual context, thus confirming that saccade averaging generalizes to reading. Second, I will provide converging evidence from several of our studies using words as well as singleton stimuli, that saccade averaging operates over a limited area. As we have previously shown, this area is best defined in millimeters of collicular surface, thus when the non-homogenous afferent/efferent mapping property of the SC is taken into account; I will show here that it is nearly of the same extent irrespective of the stimulus material. Third, I will report that the greater representation of foveal compared to peripheral input in the collicular map shapes the distribution of landing sites in the vicinity of an isolated, singleton target, but also influences where the eyes move in more complex stimulus configurations. I will then delineate the basic principles of a novel theory of saccade generation in reading, discussing the possibility that distributed coding might be relayed by lateral interactions within the collicular map.
Meeting abstract presented at VSS 2012