|Ph.D Student||Knafo Tal|
|Subject||Active Sensing in the Temporal Domain|
|Department||Department of Medicine||Supervisor||Professor Shimon Marom|
|Full Thesis text|
Why -- when presented with the exact same stimulus -- we sometimes notice it but other times miss it entirely? These response fluctuations emerge in all modalities and are fundamental for the understanding of perception, decision making, and other related cognitive processes. As such, considerable efforts have been invested over the years in an attempt to characterize and interpret variations in response. These were studied from a psychophysical perspective on an exclusively behavioral level (i.e. exploring the spatiotemporal relationship between the physical stimuli and the structure of responses they produce), as well as from a neurophysiological angle examining brain mechanisms that underlie perception and behavior (i.e. analyzing neuronal activity using imaging and other monitoring techniques).
This study is aimed at identifying temporal and spatial features underlying detection within the context of trial-by-trial variability. We explore the gap of information between the visual stimuli and the behavioral reports by examining the information flow at different levels organization in a visual detection task.
We combine both the psychophysical and the neurophysiological perspectives. This allows an explicit examination of the information flow between the temporal aspects of the stimuli, and the aware and unaware processes related to perception and detection. Specifically, subjects’ responses, eye movements, and EEG activity were measured while weak visual stimuli were presented in open and closed loop contexts.
We found two processes involved in signal detection: gaze-based and report-based detection. While both exhibited fluctuations in response to stimuli, their interrelation revealed to be somewhat complex. The performance of the eyes (i.e. gaze-based) in detecting the stimulus exceeded the performance of the subjects (i.e. report-based). Information gained through gaze-based process had no impact on report-based responses. Moreover, the threshold of report-based performance was unaffected by changes in the dynamics and performance of gaze-based detection. In contrast, gaze-based response time was correlated with the behavioral report.
Integration of eye tracking and EEG unraveled temporal and spatial components which are related to gaze-based and report-based responses. Using these components, a non-parametric classifier was constructed, achieving 84 percent accuracy in predicting responses on a single trial basis.
Taken together, this study presents converging behavioral and electrophysiological evidence for temporal and spatial features that play a key role in trial-by-trial fluctuations. Clearly, there is more (or less) than meets the eyes.