Ron Ferguson

Assistant Professor Intelligent Systems and Cognitive Science Groups College of Computing Georgia Institute of Technology 801 Atlantic Avenue Atlanta GA 30332 Office hours: Wednesday 1:30-3:00 TSRB 220, Tech Square Email: rwf@cc.gatech.edu

Research interests

The research of the Visual Thinking Group (VTG) concerns those aspects of reasoning and cognition that deal with symmetry, repetition and analogy, especially with respect to spatial reasoning tasks. By modeling how these analogical reasoning processes are used in spatial reasoning, we gain both models that can be used to understand  human performance in these areas and practical tools for tasks such as diagram interpretation.

An Advanced Visual Reasoning Architecture (AVRA). AVRA is a diagrammatic reasoning architecture in development by the Visual Thinking Group. AVRA will differ from typical diagrammatic reasoning systems in several ways:

• Incremental and dynamic. Visual elements can be dynamically added, removed, or modified with only limited processing at the interpretation level.
• Knowledge-driven. Visual relations are linked to knowledge structures, and can be influenced by or influence those structures.
• Regularity-sensitive. Individual shapes or the entire system can be checked for visual repetition and symmetry, and interpretations can use regularity as part of their reasoning.

Impact areas: Self-correcting diagrams, knowledge acquisition through diagrams, and diagram-based educational software.

A New Model of Symmetry Detection (MAGI). MAGI is a cognitively-driven model of qualitative symmetry detection. It has a number of interesting characteristics:

• MAGI makes novel psychological predictions, showing a hitherto untested role of visual structure in symmetry perception.
• It links perceptual and conceptual symmetry forms, showing how one can support the other (as is typically found in diagrams that utilize juxtaposition).
• It makes novel predictions about classical symmetry phenomena, including the preference for vertical symmetry, a century-old puzzle in symmetry perception.

Impact areas: Models of cognition and perception (including models of similarity and analogy), self-correcting diagrams.

Systems constructed as part of my research

The Incremental Structure-Mapping Engine (I-SME), a model of incremental analogical mapping. Using I-SME, it is possible to show how analogies are processed over an extended period -- for example, how a student may solve a physics problem by creating an analogy between the problem and a previously-worked example, and then extend the analogy as the student works through the problem.

GeoRep, a spatial representation system. GeoRep, given a vector graphics file (i.e., a line drawing) constructs a cognitively-plausible representation of the visual relations found in the drawing. This representation can then be used as the foundation of of a domain-specific spatial representation.

MAGI, a structure-mapping model of symmetry detection. MAGI models symmetry detection as structural alignment of visual relations, making symmetry-detection itself a form of analogical reasoning. This leads to several novel and psychologically-testable predictions about the nature of symmetry-detection in humans.

JUXTA, a diagrammatic reasoner. JUXTA uses GeoRep and MAGI to analyze "juxtaposition diagrams" -- diagrams in which an implicit side-by-side comparison of similar situations drives the diagram's point.

Organizational Activities

Member of the American Association for Artifial Intelligence, the American Psychological Association , the Association for Computing Machinery, and IEEE's Computer Society.