CGSC170 Lecture Notes - Lecture 2: Artificial Neural Network, Fault Tolerance, Turing Machine

• School: University of Delaware
• Department: Cognitive Science
• Course: CGSC170
• Professor: Kaja Jasinka

Document Summary

One of principal tools for studying the mind. Ca(cid:374) u(cid:374)dersta(cid:374)d (cid:272)og(cid:374)iti(cid:448)e s(cid:455)ste(cid:373)s a(cid:374)d ho(cid:449) the(cid:455) fit together (cid:271)(cid:455) (cid:272)o(cid:374)stru(cid:272)ti(cid:374)g (cid:373)odels that (cid:862)fit(cid:863) the data: model a theory of cognition. Data can be many things: e. g. how humans use language for interacting with the world. Two visual systems hypotheses a model of the visual system designed to fit data (both neuro) False positive: i think i see a lion (it was really a gazelle) False negative: i think i see a gazelle (it was really a lion) Abstract mathematical tools for modeling cognitive processes (abstract from biological details of neural functioning instead capture general principles of how brain works) Aka connectionist networks or artificial neural networks (cid:862)a (cid:271)ridge (cid:271)et(cid:449)ee(cid:374) algorith(cid:373) a(cid:374)d i(cid:373)ple(cid:373)e(cid:374)tatio(cid:374)(cid:863) Concerns about rule-based and serial models of simple cognitive abilities: gap i(cid:374) the (cid:374)euros(cid:272)ie(cid:374)tist(cid:859)s tool kit -> the missing level of analysis, worries about biological plausibility of physical symbol systems.