A spatial frequency spectral peakedness model predicts discrimination performance of regularity in dot patterns

The mechanisms underlying human perception of visual order remain incompletely understood in cognitive science. Prior research has shown that regularity functions as an adaptable dimension within the visual system. That uncertainty drove interest in how neural populations encode these specific spatial arrangements. No prior work had resolved whether a simple peakedness metric could account for performance across varying stimulus parameters. This gap motivated the current investigation into dot lattice structures. It was already known that receptive field sizes influence how observers process complex visual inputs. Scientists previously suggested that neural response distributions across different scales might represent regularity. This study builds upon those foundations to test the predictive power of spectral peakedness models.

The primary aim of this study is to determine if a spectral peakedness model can predict human discrimination performance for regularity in dot patterns. Researchers sought to resolve whether a simple metric derived from neural response distributions could account for subjective assessments. This objective addresses the broader question of how the visual system encodes order in complex stimuli. The team investigated if discriminability correlates with peakedness across diverse presentation conditions such as dot number and spacing. By testing this hypothesis, the authors intended to provide a computational basis for regularity perception. The motivation stems from the need to understand how the brain processes adaptable visual dimensions. They specifically examined jittered square lattices to isolate the variables influencing perceptual accuracy. This work aims to bridge the gap between physical stimulus properties and the neural mechanisms of visual order.

The researchers employed a filter-rectify-filter computational framework to analyze the spatial frequency characteristics of various dot lattices. This approach involved calculating response distributions across multiple scales for different presentation conditions. To gather human performance data, the team conducted two distinct psychophysical experiments. Participants performed a 2-alternative forced-choice task to judge the relative regularity of the stimuli. The first experiment utilized a single reference pattern to assess discrimination sensitivity. The second experiment implemented Thurstonian scaling to evaluate patterns spanning the entire range of regularity. These methods allowed for a direct comparison between the model predictions and observed human behavior. The study systematically varied dot number, size, and average spacing to ensure the robustness of the findings.

The study demonstrates that discriminability is highly correlated with the spectral peakedness measure across a wide range of presentation conditions. Two distinct peaks consistently appeared in the spectral analysis of the stimuli. A lower frequency peak corresponds to the spacing between dots in the regular lattice. A higher frequency peak relates to the size of the individual pattern elements. The researchers defined peakedness by comparing the relative heights of these two spectral peaks. This metric successfully predicted human performance in both the single reference and Thurstonian scaling experiments. The high correlation values indicate that the model captures the essential information used by observers to judge regularity. These results provide strong empirical support for the hypothesis that neural response distributions encode order.

The authors conclude that spectral peakedness provides a robust predictor for human regularity discrimination performance. Their findings suggest that the visual system relies on relative peak heights across spatial scales. This synthesis implies that regularity coding is tied to the distribution of neural responses. The evidence supports the hypothesis that the brain utilizes this specific spectral information. These results demonstrate that discriminability remains consistent across diverse presentation conditions like dot spacing. The study highlights the utility of filter-rectify-filter models in explaining complex perceptual tasks. Researchers suggest that this mechanism accounts for how observers judge order in various dot patterns. This work clarifies the relationship between physical stimulus properties and subjective visual regularity assessments.