CGSC 1001 Lecture Notes - Lecture 6: Cognitive Architecture, Procedural Memory, Connectionism
Document Summary
Data: experiment, many participants, controlled task, statistics, inference, the model does the same task, the steps the model takes are the steps we think humans take (prediction, produces data and the data is the prediction. Kinds of cognitive architectures: symbolic, operates at the level of discrete symbols, sub-symbolic, operates using number representations, which in aggregate constitute symbols, hybrid symbolic/sub-symbolic, brain, models cognition at the level of biology, but speak to cognitive issues. Symbolic architectures: production systems, for example (assuming words beginning with capital letters are variables, and other words are constants), if. Typical characteristics of symbolic architectures: declarative/procedure memory distinction, justi cations: hm and other brain damages patients, our inability to consciously retrieve and re ect on procedural memories, goals are subset of declarative memory, production compilation models automatization. Sub-symbolic: associative network: uses hebbian learning to learn patterns, when it get incomplete input, it can complete it based on the weights between models, it is unsupervised learning.