Abstract
Contrasting hypotheses for the interaction of edge and region information in object conspicuity advocate additive or nonlinear strategies. Results from studies that assess the interactive contributions of edge and region information to target conspicuity via search time often report nonlinear feature combination, whereas the results from paradigms employing subjective evaluations of conspicuity such as pair-wise comparisons often report linear summation. We suggest that this discrepancy is due to the compressive influence of a floor effect in search paradigms, where subjectively discriminable conspicuities in search tasks manifest subadditive feature summation due to a generalised floor effect. In an investigation of the relative contributions of edge and texture information to object conspicuity, we employed both a search task and a subjective conspicuity ranking task to explore this hypothesis. We developed filtered noise textures that enabled the independent manipulation of a target’s edge and region information. Edges were manipulated via logistic blurring (95% of the blur function occurred over 0.105, 0.315, 0.630 or 2.520 deg), resulting in edge strengths from sharp to unnoticeable. Region texture content was manipulated both in scale (0%, 15% and 30%), and orientation (0, 8, or 16 degrees), relative to background texture. Results from both paradigms suggest a critical role for edge information, a strong contribution of scale contrast, and a relatively weaker contribution of orientation contrast. Results from the subjective ranking paradigm suggest that features are combined via linear summation. Results from the search paradigm also suggest summation once the compressive nonlinearity induced by the performance ceiling has been accounted for.
Meeting abstract presented at VSS 2015