Definition
Connectionism is an approach in artificial intelligence and cognitive science that models mental or behavioral phenomena as the emergent processes of interconnected networks of simple units. It emphasizes massively parallel sub-symbolic processing, contrasting with the high-level symbolic manipulation of GOFAI.
Why It Matters
It shifted AI from rigid, rule-based systems to flexible architectures that excel at processing the messy real world.
Core Concepts
- Artificial Neural Networks (ANN): The primary technical tool of connectionism. Units (neurons) are connected with variable weights; learning involves adjusting these weights based on input/output patterns.
- Sub-symbolic Processing: Information is represented not by discrete symbols (like “dog”) but by patterns of activation across many units.
- Graceful Degradation: Unlike brittle symbol-based systems, connectionist networks typically lose performance gradually if damaged or given incomplete input, rather than crashing entirely.
- Learning from Experience: The ability of networks to find statistical patterns in large data sets without explicit human rules (e.g., distinguishing sonar signals of mines vs. sea life).
- Backpropagation: The key algorithm that enabled the connectionist renaissance in the 1980s by allowing the training of multi-layered (“deep”) networks.