CGSC 1001 Lecture Notes - Lecture 6: Cognitive Architecture, Hebbian Theory, Explicit Memory
15 Apr 2016
School
Department
Course
Professor
Document Summary
Representaion of something that excludes unimportant detail and informaion: a scale model of a home made of cardboard, a categorizaion scheme for the students in my class, a simulaion of a hurricane. A programming environment or set of tools for making cogniive models. Typically includes constraints on how cogniion works in all people speed of learning, memory retrieval (ignoring cultural and other learned aspects) Someimes easier to make a model in architecture, other ways more diicult. Symbolic: operates at the level of discrete symbols: producion systems, ex: unemployment (ei, characterisics: Declaraive/procedural memory disincion: our inability to consciously retrieve and relect on procedural memories. Sub- symbolic: operates using number representaions, which in aggregate consitute symbols: uses hebbian learning to learn paterns. When it gets incomplete input, can complete it based on the weights between the nodes. Unsupervised learning: connecionism: supervises learning, using backpropogaion, great for classiicaion. Needs thousands of examples to work, unlike humans.
