2x2 Bar Graph Reading Accuracy Robust to Low-Pass Spatial Frequency Filtering

Poster Presentation: Tuesday, May 20, 2025, 8:30 am – 12:30 pm, Banyan Breezeway
Session: Perceptual Organization: Aesthetics

Nicholas Fencl1, Nestor Matthews1; 1Denison University

Introduction: College science courses often require that students develop perceptual organization skills to identify statistical effects in 2x2 graphs. For any given 2x2 graph, identifying Factor A’s main effect, Factor B’s main effect, and the interaction effect entails three different perceptual organization tasks. Here, we tested whether these three distinct perceptual organization tasks differentially rely on a 2x2 bar graph’s high-spatial frequency content. Method: Forty-five college aged students from an introductory psychology course viewed black and white 2x2 bar graphs. During training, students received instructions on classifying their randomly assigned target factor (Factor A, Factor B, or Interaction) as either statistically significant or not. During testing, students made their statistical-significance judgments on 2x2 graphs that differed randomly across trials in spatial frequency: full spectrum graphs vs low-pass filtered graphs. The black and white bars had sharp edges in full spectrum graphs, and blurry edges in low-pass filtered graphs. Stimuli from the two spatial frequency conditions had the same root-mean-square contrast. Results: Thirty-nine (86% of) students performed the task with greater-than-chance accuracy (binomial probability p < 0.001). Among those students, Wilcoxen signed rank tests for the Factor A, Factor B, and Interaction conditions each showed non-significant differences in accuracy (% correct) between the full spectrum graphs and the low-pass filtered graphs. Median accuracy across the six experimental conditions (three target factors x two spatial frequency conditions) ranged between 85.9 and 92 percent correct. Conclusion: Under the conditions tested here, our data suggest that 2x2 bar graph reading accuracy remains unaffected by the removal of high spatial frequency information. This finding appears to hold regardless of whether the perceptual organization task pertains to Factor A’s main effect, Factor B’s main effect, or the interaction. Visual mechanisms tuned to low spatial frequencies provide ample spatial information for accurate 2x2 bar graph reading.

Acknowledgements: Denison University Lisska Center for Intellectual Engagement