Purchase this article with an account.
Adam Larson, Lester Loschky, Elise Matz, Scott Smerchek, Pheasant Weber, Lindsey Berger; The roles of central versus peripheral visual information in recognizing scene gist. Journal of Vision 2008;8(6):737. doi: https://doi.org/10.1167/8.6.737.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Scene gist recognition is a critical early scene perception process. Viewers reach asymptotic gist recognition with masked stimulus durations of 100 ms, and most important information is gathered within 40–50 ms. The rapidity of gist recognition raises the question, where does the most important information for gist come from, central vision or the periphery?
Method: We briefly flashed scenes in either “window” or “scotoma” conditions. A “window” is a circular region showing the central portion of a scene, and blocking out all peripheral information. A “scotoma” is an inverse window—a circular region blocking out all central information of a scene, and showing only information outside the circle. In several experiments, we varied window and scotoma sizes and measured their effects on scene gist recognition to determine the importance of central versus peripheral visual information. Experiment 1 compared windows and scotomas with radii of 1°, 5°, 10.8° and 13.5° in a 27° × 27° image. Experiment 2 compared windows and scotomas with equal areas of 4, 10, 40, or 100% in circular images. Experiment 3 estimated the cross-over point for window and scotoma accuracy functions, to determine the critical radius producing equal gist recognition.
Results: Our data suggests that central information is more useful than peripheral information for recognizing scene gist. Fewer centrally located pixels are needed to achieve the same scene gist recognition accuracy as peripherally located pixels. However, scenes can be recognized at an asymptotic level without any information from the central 5° of vision (the fovea and parafovea), if [[gt]] 50% of image pixels are shown in the periphery. Furthermore, we identified a critical radius, perfectly dividing scenes into a central region and peripheral surround, each producing equal accuracy. The ratio of pixels in the central versus peripheral regions was approximately 3:7, suggesting parallels with cortical magnification.
This PDF is available to Subscribers Only