Related Experiment Videos
Attractors: architects of network organization?
1The Mark O. Hatfield Marine Science Center, Oregon State University, Newport, OR 97365, USA. gmpitsos@slugo.hmsc.orst.edu
Brain, Behavior and Evolution
|September 6, 2000
Summary
Neural attractors dynamically shape network structure and function. This study proposes that attractor dynamics guide activity-dependent learning, optimizing synaptic strengths and network parameters for efficient information processing in spike-activated networks.
Area of Science:
- Computational Neuroscience
- Neural Networks
- Systems Neuroscience
Background:
- Neural systems exhibit attractor dynamics, where activity settles into specific states.
- Information processing in neural networks is linked to these dynamic attractor states.
- Classical models often overlook the dynamic interplay between network activity and structure.
Purpose of the Study:
- To explore how attractor dynamics can influence neural architecture and learning.
- To propose a model where attractors guide activity-dependent synaptic and network parameter optimization.
- To investigate the role of temporal coding in spike-activated networks.
Main Methods:
- Conceptual framework based on attractor dynamics in neural systems.
- Discussion of pulse-propagated or spike-activated network models.
- Consideration of activity-dependent learning mechanisms.
Main Results:
- Attractors are identified as sources of information that can shape neural architecture.
- A conjecture is presented: attractor relaxation dynamics may guide optimal, interrelated setting of network parameters.
- Temporal coding in spike intervals is crucial for emergent cooperative activity.
Conclusions:
- Attractor dynamics offer a novel perspective on neural information processing and self-organization.
- This framework suggests a more integrated approach to neural network design compared to traditional models.
- The proposed ideas are testable in both biological and simulated neural networks.