With the increasing reliance on data-driven decision-making, information visualizations have become critical tools across various domains. However, cognitive biases frequently distort users’ interpretation of visualized data, leading to suboptimal outcomes. While prior research has documented cognitive biases in specific visualization contexts, most studies remain at the behavioral level and lack a unified framework for systematic bias modification. This study proposes a novel framework for understanding and modifying cognitive biases in information visualizations using event-related potentials (ERP). The framework consists of four key stages: 1) identification of potential cognitive bias triggers, 2) neural and behavioral characterization of cognitive biases, 3) preliminary design plans for cognitive bias modification, and 4) design evaluation and design strategy optimization. A case study on automotive dashboard interfaces demonstrates the framework’s effectiveness in addressing anchoring bias. Experimental results show that combined angular and textual encoding induces anchoring bias, as evidenced by specific ERP components (P2, P300, N400). Three design strategies were derived: context-appropriate textual warnings, vertical alignment of elements, and semantically consistent information positioning. This work bridges theoretical insights with practical applications, offering a structured and actionable guidance for creating bias-modified visualizations.