COGNITIVE FIDELITY IN DATA REPRESENTATION: A PSYCHOPHYSICAL ANALYSIS OF VISUAL PERCEPTION CONSTRAINTS IN ANALYTIC DASHBOARDS

Abstract

Background: As data volume increases, the design of analytic dashboards often prioritizes information density over cognitive accessibility. This study investigates the psychophysical limitations of human visual perception—specifically visual working memory (VWM) and feature binding—within the context of modern business intelligence interfaces. Methods: We conducted a multi-stage experiment (n=240) comparing three visualization modalities: static aggregated views, interactive filtering dashboards, and narrative "scrollytelling" formats. Participants performed low-level analytic activities and complex insight generation tasks while under varying cognitive loads. Stimuli were assessed using the Chromatic Vision Simulator to ensure accessibility. Results: Quantitative analysis revealed that while interactive dashboards offer the highest theoretical retrieval capacity, they induce significantly higher cognitive fatigue. The "scrollytelling" format resulted in a 34% improvement in long-term information retention and superior performance in risk assessment tasks involving proportion estimates. Furthermore, the use of anthropomorphic icons (stick figures) significantly improved probability comprehension among participants with lower baseline numeracy compared to abstract geometric shapes. Conclusion: The findings suggest that the bottleneck in data analytics is no longer computational but perceptual. Effective dashboard design must account for the "binding capacity" of VWM. We propose a "Cognitive Fidelity" framework that prioritizes narrative structure and perceptual grouping over raw data density to enhance decision-making accuracy.