Segmentation, Binding, and Illusory Conjunctions

Area of Science:

• Computational Neuroscience
• Cognitive Neuroscience
• Neural Network Modeling

Background:

• The brain's ability to bind different attributes of an object (e.g., shape, color) into a unified perception is a fundamental cognitive process.
• Neural network models offer a powerful framework for investigating the mechanisms underlying complex cognitive functions like binding.
• Oscillating neural networks, composed of excitatory and inhibitory cell assemblies, provide a biologically plausible substrate for information processing.

Purpose of the Study:

• To investigate the neural mechanisms of attribute binding using a computational model of oscillating neural networks.
• To determine if a model of interconnected excitatory and inhibitory cell assemblies can replicate the phenomenon of attribute binding.
• To explore the limitations of such a model when processing multiple object attributes, specifically examining the emergence of illusory conjunctions.

Main Methods:

• Developed a computational model comprising two interconnected oscillating neural networks.
• Excitatory cell assemblies within each network represented object attributes (e.g., shape, color).
• Simulated the segmentation of input data containing mixed attribute pairs and observed the phase locking of oscillating activities to demonstrate binding.

Main Results:

• The model successfully demonstrated attribute binding for pairs of objects, indicated by the phase locking of relevant neural activities.
• When presented with more than two objects, the model exhibited faulty correlations, conjoining attributes from different objects.
• These faulty correlations mirror the phenomenon of illusory conjunctions observed in human visual perception.

Conclusions:

• Oscillating neural networks with interconnected excitatory and inhibitory assemblies can model the binding of object attributes.
• The model's limitations in handling multiple objects highlight potential neural mechanisms contributing to illusory conjunctions in vision.
• This research provides insights into the computational principles underlying perceptual binding and its potential failure modes.