Related Experiment Videos
Recognition by top-down and bottom-up processing in cortex: the control of selective attention
1Volen Center for Complex Systems, Brandeis University, Waltham, Massachusetts 02454, USA. dgraboi@cts.com
Journal of Neurophysiology
|April 19, 2003
Summary
This study models visual word recognition using a hierarchical cortical network. The model demonstrates how attention guides feature processing for rapid, efficient recognition of words.
Area of Science:
- Neuroscience
- Cognitive Science
- Computational Modeling
Background:
- Visual recognition relies on hierarchical cortical processing.
- Attention directs information sampling within a movable window.
- Understanding the mechanisms of attention and information integration is crucial.
Purpose of the Study:
- To model the process of visual word recognition.
- To explore how the cortex moves an attention window and integrates information.
- To provide a physiologically plausible account of bi-directional cortical signal flow in recognition.
Main Methods:
- Developed a computational model of visual word recognition.
- Modeled hierarchical cortical areas representing features, letters, and words.
- Simulated top-down (T-D) and bottom-up processing cycles guided by attention.
Main Results:
- The model successfully simulates word recognition through iterative cycles of attention and processing.
- Recognition of 950 words occurred in an average of 4.9 cycles.
- Model results align with experimental findings on recognition time, set size, and contextual effects.
Conclusions:
- Bi-directional cortical signal flow guides attention for efficient visual recognition.
- The model offers a plausible explanation for rapid word recognition (<200 ms cortical processing).
- This framework accounts for various empirical observations in visual word recognition.