CEREBRAL MECHANISMS OF VISUAL FEATURE INTEGRATION

点击翻译按钮获取中文摘要

Recordings from primate cortex have shown that different features of visual stimuli are analyzed in distinct cortical fields. How does the perceptual system recombine these features to form a coherent percept? Clues are available from studies of visual search . Subjects rapidly detect targets distinguished from distracting stimuli by single features (e.g., color) but are slow and error-prone when targets are distinguished by conjointed features (e.g., color and orientation). According to Treisman's feature integration theory, single features of stimuli are processed automatically and in parallel in different feature maps. However, visual feature integration (VFI) requires a serial process in which objects must first be localized before their features can be bound together by an attentional scan. Unfortunately, it is difficult to study the physiology of VFI in search tasks since the experimenter can never be certain which stimulus the subject is processing. A new paradigm, rapid serial visual presentation (RSVP), permits VFI to be studied for stimuli presented one-by-one. Four experiments are proposed using RSVP tasks. For comparison, a fifth experiment utilizing visual search is also included. In all experiments, behavioral analysis will be combined with multi-channel recording of event- related brain potentials (ERPs). ERP difference waves associated with the processing of individual stimulus features and with feature integration will be isolated to provide insight into the localization and timing of VHI. Preliminary results show ERP difference waves related to feature conjunction over frontal and parietal regions at latencies of 200-250 ms. Event-related difference spectra (ERDs) will also be used to localize cortical activity associated with VFI. Preliminary data show ERD changes related to visual feature processing over the right parietal and occipital scalp. In all experiments, behavioral and electrophysiological results from normal subjects will be compared with data from patients with focal cortical lesions of the lateral parietal lobe, dorsolateral prefrontal cortex, or posterior superior temporal plane. Lesions in the regions have been shown to impair VFI in visual search tasks. Behavioral and electrophysiological results from normal subjects and patients will be used to test a model in which different VFI operations are localized to distinct cortical regions; the "master map" of locations in lateral parietal cortex, the scan of the master map controlled by dorsolateral prefrontal cortex with a variable aperture of attentional scan controlled by the temporal/parietal junction.