Abstract
Segregating objects from backgrounds is one of vision's most important tasks. A long-standing hypothesis links motion-based object segregation with suppressive center-surround mechanism. This proposed functional link has theoretical and neurophysiological support, but little direct behavioral evidence exists. Here, we investigated the role of background motion suppression in rapid object segregation. We hypothesized that perceptual insensitivity to large, background-like moving stimuli (Tadin et al., 2003) has a functional consequence of enhancing the relative visibility of moving objects on moving backgrounds. METHODS: At both low and high contrast, we measured subjects' ability to (1) detect moving stimuli (6°/s) across different sizes (1.5°-16° diameter) and (2) segregate a small moving object (1° diameter) presented on different sized moving backgrounds (also 1.5°-16°). Given our aim to study rapid motion segregation, we measured duration thresholds using a custom-built 360Hz display system. RESULTS: Size and contrast influenced both motion detection and segregation. Importantly, the relationship between the two tasks was negative. At high contrast, duration thresholds for motion detection gradually increased from ~6 ms for 1.5° stimuli to over 12 ms for 16° stimuli. Over the same range of background sizes, thresholds for motion segregation decreased from 12 to 5.5 ms. Remarkably, subjects were considerably better at detecting a moving object on a moving 16° background than detecting whether the same background stimulus was moving or stationary. At low contrast, we found pronounced spatial summation in the motion detection task (thresholds improved with increasing size). As predicted, the segregation performance was poor, and, unlike at high contrast, considerably worse than motion detection. CONCLUSION: These findings link spatial suppression with rapid motion segregation. Spatial suppression effectively accomplishes background subtraction and, as a result, increases visual saliency of object motion. This is a computationally and metabolically efficient strategy because object segregation is accomplished by suppressing irrelevant background information.
Meeting abstract presented at VSS 2014