Behavioral and neural correlates of impaired scene perception following saccadic eye movements
Poster No:
2538
Submission Type:
Abstract Submission
Authors:
Yong Min Choi1, Tzu-Yao Chiu1, Julie Golomb1
Institutions:
1The Ohio State University, Columbus, OH
First Author:
Co-Author(s):
Introduction:
Across frequent ballistic eye movements (i.e., saccades), the visual input projected to the retina shifts drastically. Although people are not aware of it, visual perception is impaired around the time of saccadic eye movements (Burr et al., 1994; Golomb et al., 2014). Most of these previous studies have used simple visual stimuli to assess perisaccadic visual perception, but how the processing of complex visual scene images is influenced by saccades remains less explored. Therefore, the current study explored behavioral evidence of impaired post-saccadic scene perception, and then further investigated underlying neural mechanisms by assessing 1) scene category information in high-level visual cortex, and 2) low-level visual information (i.e., spatial frequency) in early visual cortex.
Methods:
In a series of behavioral and fMRI experiments assisted by eye-tracking, we presented scene images at different time points relative to saccade offset (Figure 1). Scene images contained either low- or high-spatial frequency information to investigate whether specific spatial frequency bands contribute differently to post-saccadic scene perception. In gaze-contingent behavioral experiments (Figure 1A), subjects followed a fixation dot and categorized briefly presented post-saccadic scene images (e.g., beach, mountain, forest, city, office, and highway). In Experiment 1, the scene image appeared either 5 ms or 500 ms after the saccade completion. Experiment 2 (pre-registered) included five post-saccadic delay conditions (5, 16, 50, 158, and 500 ms) to explore the time course of post-saccadic scene perception. In the fMRI experiment, instead of an explicit categorization task, subjects performed a 1-back task on scene images presented with jittered timing after the saccade cue (Figure 1B). Trials were sorted according to post-saccadic delay based on post-hoc analysis of eye-tracking data.
Results:
To answer how scene perception is behaviorally influenced by saccadic eye movement, we compared scene categorization accuracy between post-saccadic delay conditions. We found lower scene categorization accuracy in short than long delay trials, irrespective of spatial frequency of scene images (Experiment 1). Furthermore, Experiment 2 showed a lower scene categorization accuracy in the 5 ms and 16 ms post-saccadic delay condition compared to the baseline (500 ms), but not in the 50 ms and 158 ms delay conditions, suggesting rapid recovery within 50 ms after saccade completion. Additional analyses confirmed that the post-saccadic impairment was not attributable to eye movement velocity or saccade detection criterion.
Next, we investigated neural correlates underlying the behavioral impairment. We assessed scene category and spatial frequency information in scene-selective region (parahippocampal place area; PPA) and early visual cortex (EVC), respectively, by running RSA-based decoding analysis (Haxby et al., 2001). First, the posterior PPA showed lower scene category information in the short compared to the long post-saccadic delay trials. Moreover, spatial frequency information in early visual cortex was also decreased in the short compared to long post-saccadic delay trials.
Next, we investigated neural correlates underlying the behavioral impairment. We assessed scene category and spatial frequency information in scene-selective region (parahippocampal place area; PPA) and early visual cortex (EVC), respectively, by running RSA-based decoding analysis (Haxby et al., 2001). First, the posterior PPA showed lower scene category information in the short compared to the long post-saccadic delay trials. Moreover, spatial frequency information in early visual cortex was also decreased in the short compared to long post-saccadic delay trials.
Conclusions:
To summarize, we provided behavioral evidence for impaired perception of both high- and low- spatial frequency filtered scene images presented shortly after saccades. Neural evidence suggested that interrupted neural representations at different levels of visual hierarchies may underly impaired post-saccadic scene perception, affecting both high-level scene category information AND low-level visual information of a scene image. These results provide novel evidence for impaired perception of realistic scene images following saccadic eye movement, unlike seemingly intact visual perception across eye movements.
Modeling and Analysis Methods:
Classification and Predictive Modeling
Multivariate Approaches
Novel Imaging Acquisition Methods:
BOLD fMRI 2
Perception, Attention and Motor Behavior:
Perception: Visual 1
Perception and Attention Other
Keywords:
Vision
Other - eye movement
1|2Indicates the priority used for review
