Abstract
In vision and neuroscience there is great interest in the mechanisms that underlie object perception, and a common approach to their study is the characterization of neural structures that appear more responsive to certain object images than others. However, it may be difficult to dissociate neural activity driven by low-level features of these images from neural activity that drives the higher-level percepts of the objects depicted. We explore an alternative approach, simultaneously measuring changes in perceptual and neural activity as visual stimuli undergo systematic transformations (e.g., as recognizable objects first evolve from, then dissolve into, randomness) while important low-level image properties are held constant. Subjects' object percepts arise quite sharply in such cases and thereafter exhibit a marked hysteresis (i.e., percepts persist at levels of image degradation far worse than those which support their original formation).
These properties, along with the ability to objectively verify subjects' conscious percepts, offer distinct advantages when this technique is used in conjunction with neuroimaging (e.g., magnetoencephalography) to isolate neural activity that corresponds specifically with the object percept — such activity should demonstrate a similar time course, one with a relatively sharp onset, despite potentially gradual image evolution, and a delayed offset consistent with the perceptual hysteresis. Using this approach, we can indeed distinguish, spatially and temporally, certain MEG signal components that appear to demonstrate such characteristics, whether the visual stimuli depict faces or other objects. Experiments such as these may help further our understanding of the neural mechanisms that underlie a number of important perceptual phenomena, from those associated with the priming and formation of an object percept to those involved in its enhancement and maintenance, as seen, for example, in perceptual learning and hysteresis.