The human visual system utilizes attention to direct processing towards areas of interest. In particular, certain objects in a visual scene can be salient, meaning they attract attention rather than being the targets of some search process. Visual salience appears to be driven by the formation of visual proto-objects, which have been hypothesized to cause an increase in synchronous firing between neurons encoding parts of an object. This thesis approaches proto-objects both from a behavioral level and at a low level of analyzing synchrony. At the behavioral level, existing studies of visual salience rely on many repetitive trials or task instructions to tell study participants what to do, which can influence attentive behavior in a top-down manner, confounding the measurement of salience. I introduce an experimental paradigm that records attentional selections from subjects without any such information, and used this paradigm to analyze whether proto-objects interact in the determination of salience. The results show that uniqueness of an object does indeed attract attention, and I develop a model that normalizes among proto-objects to explain the measured data. At the neuronal level, I develop a more rapid method to perform jitter hypothesis tests regarding detecting the presence of synchronous spiking between pairs of neurons. While the detection of synchrony does imply some connection between neurons, I also show that the inference of a change in common input from changes in synchrony is not possible.
【 预 览 】
附件列表
Files
Size
Format
View
Neural and Behavioral Consequences of Perceptual Organization using Proto-Objects