Schmidtmann, G., Desjardins, A., Kingdom, F.A.A., RF shape channels: The processing of compound Radial Frequency patterns. VSS, 2017 [POSTER]
Radial Frequency (RF) patterns are quasi-circular contours that are frequently used to investigate intermediate stages of shape processing. Combinations of RF patterns have been used to construct more complex shapes such as head contours. Previous studies have suggested that complex shapes may be encoded by multiple, narrowly-tuned RF shape channels. The aim of this study was to test the hypothesis that complex shape processing may be based on multiple, independent RF channels and to demonstrate the limitations such shape descriptors. Thresholds were determined for detection (circle vs. RF compound) and discrimination (RF compound vs. RF compound) of various weighted combinations (symmetrical and asymmetrical) of two RF components (RF3&RF5; RF3&RF8; RF4&RF7). If both RF components were processed by a common broadband channel, one would expect a substantial increase in sensitivity as the information from both components would be summed within the same channel (additive summation: AS). If the two components were processed independently by separate channels, one would expect only a slight increase in sensitivity for the compound compared to the components (probability summation: PS). The data were analyzed by a model for probability (PS) and additive summation (AS) under Signal Detection Theory (Kingdom, Baldwin & Schmidtmann, Journal of Vision, 15(5):1). Results show that summation of information from different RF components is consistent with AS. This suggest that the shapes tested here are processed by a broadly tuned mechanism. In addition, we demonstrate the mathematical limitations of RF patterns which make them an unlikely candidate for universal shape descriptors.