Perceptual-learning systems as conservative structures: is economy an attractor?

Area of Science:

• Cognitive Science
• Theoretical Psychology
• Computational Neuroscience

Background:

• The law of Prägnanz, a core Gestalt principle, describes perceptual organization towards simplicity.
• Existing computational models (information-processing, connectionist) face challenges when considering physical realization constraints.

Purpose of the Study:

• To evaluate information-processing and connectionist models of the law of Prägnanz under physical constraints.
• To propose and assess an alternative Gestalt-inspired model for explaining the law of Prägnanz.

Main Methods:

• Comparative analysis of information-processing and connectionist models.
• Contrasting computational approaches with thermodynamic principles.
• Development and theoretical evaluation of a Gestalt-inspired model.

Main Results:

• Information-processing and connectionist models may not adequately explain the law of Prägnanz under physical limitations.
• A Gestalt-inspired model shows potential for physical realization.
• The proposed model has relatively moderate substrate requirements for information-processing resources.

Conclusions:

• The law of Prägnanz may be better explained by models directly inspired by Gestalt principles.
• Physical realizability is a critical factor in evaluating computational models of perception.
• The proposed Gestalt-inspired model offers a promising avenue for understanding perceptual organization.