idea that mental phenomena arise from connections between neurons
- assumes that both neurons and connections can be represented mathematically
if that's true, then artificial neural networks should be able to create consciousness?
- do we need a "critical mass" of neurons?
tensions with computationalism
- connectionism models at "low-level" (neurons), where computationalism models at "symbolic level"
- connectionism focused on learning from stimuli (back-propagation in neural networks), where computationalism focuses on learning as "elaborating possibilities through symbolic processing"