Download
1 / 12
120 likes | 275 Vues
This study explores how sharp and blurred targets are detected differently based on their visual context—whether they are perceived as figures or grounds. It discusses the processing differences between figure and ground, highlighting that sharp targets are easier to detect when presented as figures, while blurred targets are better identified against backgrounds. The findings stem from a series of experiments designed to analyze detection thresholds, luminance levels, and the effects of retinal eccentricity. This research sheds light on underlying visual processing mechanisms and offers insights into attention and image perception.
E N D
Sharp Targets Are Detected Better Against a Figure, and Blurred Targets Are Detected Better Against a Background Eva Wong and Noami Weisstein, 1983
Overview • Background • Figure vs. Ground • Reversal • Processing Differences • Experiments • Assumptions & Hypothesis • Research Question • Design & Measures • Results • Discussion
BackgroundFigure vs. Ground • Rubin’s illusion
BackgroundFigure-Ground Reversals • When “foreground” becomes “background” and/or vice versa • Widespread in art (according to Douglas Hofstadter, anyway)
BackgroundProcessing Differences • “Figure” aids detection of: • Contour discontinuity • Retinal image displacement • Line orientation • Possible Reasons: • Differential attention? • Differential resolution? • Differential sensitivity to spatial frequency?
The ExperimentAssumptions & Hypothesis • “Figure” and “Ground” represent different channels in the visual system • The channels have different functions: • “Figure” responsible for detail • “Ground” responsible for ‘global information’ • Therefore: • “Figure” channel more sensitive to high spatial frequencies • “Ground” channel more sensitive to low spatial frequencies
The ExperimentResearch Question • So, is the detection threshold: • lower in the figural regions for high spatial frequencies (such as a sharp target?) • lower in the ground region for low spatial frequencies (such as a blurred target?) vs.
The ExperimentDesign & Measures • First Experiment • Purpose • Find observers who hold their (monocular) gaze regardless of what’s figure or ground • Procedure • Half of subjects initiate trial when the faces are figure; the other half initiate the trial only when the goblet is figure • The stimulus then appears in the blind spot at 50% probability • Measure detection accuracy; if different than chance, they’re not fixating!
The ExperimentDesign & Measures (cont.) • Second Experiment • Purpose • Establish luminance level where TP = 70% for both blurred and sharp targets • Procedure • Display sharp target at fixation cross at 50% probability • Change luminance until 70% accuracy is achieved for each of three blocks • Measure the final luminance value for each observer • Repeat for blurred target
The ExperimentDesign & Measures (cont.) • Third Experiment: • Purpose • Determine accuracy of target detection against figure and against ground regions • Procedure • Target has a 50% probability of being presented • If target is presented (20 msec), it has • A 50% probability of being in the “goblet region” • A 50% probability of being in a “face region” • Measure TP and FP to estimate d’ and plot ROC
Results Discrimination improves: • When sharp targets displayed in figure • When blurred targets displayed in ground Off-fixation attenuates d’ by a “fixed magnitude” • Reflects an early processing constraint: retinal eccentricity • Caused by decreasing resolution with increasing distance from fovea
Discussion • Conclusions: • Different visual processes mediate the analysis of figure and ground • Accuracy not determined solely by attention, as defined by gaze or what is perceived as figure) • [Accuracy is also not determined solely by photoreceptor density] • “Global information extraction” may proceed faster than figure analysis
