Judy Fan's Insights on Cognitive Tools

Jun 5, 2025

Lecture Summary: Judy Fan on Cognitive Tools and Visual Abstraction

Introduction

  • Judy Fan is a cognitive scientist and assistant professor at Stanford University.
  • She has won awards like the Glushkow dissertation prize and NSF career award.
  • Judy's research spans from visual psychophysics and computational neuroscience to cognitive processes like artistic expression and education.
  • Focus on cognitive tools and technological innovation that make invisible concepts visible.

Key Concepts

Cognitive Tools

  • Cognitive tools like the number line are human inventions that aid in understanding and reasoning.
  • These tools include graphical representations like rectangular coordinates, which link algebraic expressions with geometric curves.
  • The development of these tools is crucial for mathematical and scientific discovery.

Human Innovation

  • Human innovation involves creating tools and abstractions that make underlying concepts visible and understandable.
  • Examples include Darwin’s finches and Feynman diagrams, which make complex scientific concepts more comprehensible.
  • Visual abstraction is key in communicating and understanding scientific knowledge.

Visual Abstraction

Visual Communication

  • Visual communication involves perception, production, and communication using visual abstraction.
  • Research on drawings shows how people adjust visual detail based on context (e.g., drawing games with varying object similarity).

Mechanistic vs. Depictive Drawings

  • Study on drawings differentiating between mechanistic explanations (focusing on causality) and ordinary depiction (focusing on visual resemblance).
  • Findings suggest explanations focus more on causal parts and symbolic representation (e.g., arrows), while depictions are more about visual fidelity.
  • Visual explanations help in understanding mechanisms better but are less effective in object identification compared to depictive drawings.

Artificial Intelligence in Visual Abstraction

Sketch Generation and Understanding

  • AI models are tested for their ability to understand and generate sketches that humans find recognizable.
  • Models like CLIP show promise, but gaps exist between AI and human recognition and abstraction.

Human-Model Comparison

  • Comparative studies show a gap in how AI models understand and generate sketches compared to humans.
  • Human sketches under time constraints show different abstract characteristics compared to AI-generated ones.

Data Visualization and Statistical Reasoning

Importance of Data Visualization

  • Data visualization is a powerful tool for understanding large, complex data sets.
  • It is increasingly important in education for developing quantitative literacy.

Machine Learning and Visualization Understanding

  • Studies involve comparing human and AI performance in understanding data visualizations.
  • Current AI models like GPT-4 show performance gaps in understanding visual data despite advances.

Improving Visualization Skills

  • Research focuses on understanding how people choose suitable visualizations for specific data questions.
  • Studies demonstrate that people are sensitive to features that make some visualizations better suited for certain questions.

Conclusion

  • Judy Fan’s research aims to understand and improve cognitive tools for scientific learning and innovation.
  • Future directions include refining psychological models of visual communication and developing better educational resources for visual and data literacy.

Full transcript