Abstract
Purpose :
To understand visual behavior in natural conditions, we developed a novel active visual search task in which participants saccade freely across an evolving display of Gabor patches. Using data mining approach, we identified a characteristic visual search behavior and examined the correlated brain activity using high-density EEG recording with advanced EEG signal processing.
Methods :
Participants were 19 healthy young adults. Five Gabor patches were present on a gray background. A sequence of 1200 patches were pseudo-randomly selected from four patch sizes and five orientations (SOA 2 sec, duration 10 sec). The participant’s task was to detect infrequent (15%) instances in which the size and orientation of the newest patch matched those of another patch that was still visible, prompting a button press. Eye-tracking data were obtained from the left eye while 205-channel, 512-Hz scalp EEG data were recorded. EEG data were preprocessed using independent component analysis and subsequently their equivalent current dipole were estimated.
Results :
Approximately 50% (100 trials) of successful target detection trials followed this visual search pattern: they fixate near the Matching (new) stimulus at 900ms (the latency is relative to the button press), then fixate near the Matched (old) stimulus at 400ms, then fixate again near the Matching (new) stimulus at 150ms (Figure 1). Here, the returning to the Matching (new) stimulus as a second fixation emerged as an interesting question because 150ms is too short to change the behavior. Event-related potential (ERP) analysis showed that even though both the first and the second fixations showed sharp ERP spike relative to fixation onsets, the second fixation showed 600ms early ERP time course within temporal and occipital sources. Linear regression was used to separate the overlapping fixations and button press events, but the 600ms preceeding time course for the second fixation remained the same.
Conclusions :
Our interpretation for the preceeding ERP time course in the occipital visual areas is that it is due to reconstructive perception based on visual memory. Evidence for this type of neural activity is known as remapping; animal electrophysiology showed that receptive filed predictively fires before the beginning of saccade (Duhamel et al., 1992). The double take may consist of constructive confirmation of the visual reality.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.