Abstract
Introduction: Feature-based attention boosts the gain of neurons tuned to task-relevant features, even for matched features in unattended locations. Thus, feature-selectivity in unattended regions carries information about attended features, and this can be measured through frequency tagging of the unattended region (steady state visually evoked potentials, or SSVEP; Bridwell et al. 2012; Painter et al., 2014). In this study, we measure the feature-selectivity of anticipatory SSVEPs during a detection task to test whether neural signals for attentional filtering predict behavioral performance. Methods: Subjects monitored for the presence of a point-light human walker embedded in uniform random dynamic noise in the center of the screen (a detection task in the attended region). The target was surrounded by peripheral annulus with 100% coherent noise moving left, right, up or diagonal, and flickering at 15 Hz (the SSVEP-tagged unattended region). To measure attentional (not perceptual) modulations of the SSVEP, we analyzed only the EEG data collected 1 second before the onset of the target. We computed tuning curves (normalized SSVEP power) from this pre-target interval using the electrodes with highest SSVEP power. A Partial Least Squares regression model determined the extent to which cortical coherence maps generated with those electrodes as seeds predict individual participant's d-prime sensitivity. Results: The highest SSVEP power electrodes over posterior parietal cortex (PPC) were modulated by the direction of the unattended feature, strongest on trials in which the unattended motion matched the dominant local motion of the "backstroke" of the feet. Coherence between PPC and dorsolateral prefrontal cortex on those same trials successfully predicted individual participant's d-prime sensitivity. Participants with higher d-prime values had higher coherence, and vice versa. Conclusion: The results from this experiment demonstrate importance of the backstroke of the feet as a cue for detecting bipedal motion of humans, and attention to that feature predicts subsequent perceptual sensitivity.
Meeting abstract presented at VSS 2016