Connectionism is a neural network approach to artificial intelligence that focusses on how parallel processing and interconnectedness between neurons can help machines learn, while symbolic AI is an approach that uses preprogrammed symbols and rules to solve problems.