Lecture 5 - Objections to Computationalism PDF

Preview

Summary

Professor Charles Yang
Fall 2017...

Description

Introduction to Cognitive Science CIS 140

Lecture 5 - Objections to Computationalism ● Searle objects to the whole idea of computationalism ○ Wrong to think that mental processes and program processes are identical ○ There is more to having a mind than having formal or syntactical processes ■ Because the mind has more than syntax, it has semantics ○ Chinese Room Experiment - computers can map the symbols to produce answers in Chinese but will never understand the meaning behind the symbols/answers ■ Arguments - the processor which maps the symbols does not understand Chinese but the system as a whole (processor, rules, symbols, room) understands Chinese ● Whole system still doesn’t have any semantics ● Objections might be ○ To the whole project (Searle) ○ To its exhaustiveness (maybe it can explain some things but not all) ○ To the minimization of the importance of the implementational level/brain ● Computationalism - not important to understand physical nature of mind ○ The mind could be generated by “old beer cans powered by windmills” ■ Even if implementation is irrelevant, it could be important discovery of the computational processes ■ Strict computationalism accept interdisciplinary work with neuroscientist is epistemically (necessary for discovery) necessary ○ Computationalism - focuses more on software rather than the implementation/hardware ● Can understand how brains generate minds without an intermediate computational explanation ○ Searle - doesn’t make the causal/generative distinction in this paper. ○ Says that brains cause minds ← not right ○ Brain structures “realize” minds and that we shouldn’t think of this any differently than we think of lower level processes making up biology kinds ● According to computationalism - thermostats, humans, laptops are pretty similar ○ Searle - there is a difference ○ Difference in kind view incompatible with naturalism? - No, swimming, walking, flying are all different kinds of natural processes ● Strong AI: “The brain is just a digital computer and the mind is just a computer program” ○ Symbols of a digital computer have no meaning ← not right ■ Computer symbols have meaning; they’re just causally irrelevant ● Turing Test for Intelligent - computer can beat human at imitation game ○ Person has to judge response comes from computer and from human ● Computational Theory of Mind Refinement (CTM, refinement) ○ Mental representations are symbolic states of a computer with content ○ Mental processes are formal transitions from one mental representation to another ● Refined CTM still says that content is causally still irrelevant. Even though we may understand

Chinese, we still communicate by converting symbols like the CRE ● Searle - maybe formal processes could stimulate thinking but they cannot duplicate it ● Searle is not discontinuing the usefulness of computational models of cognition, he thinks it may not be ideal ● Kahneman and Tversky - Linda question → showed people do not always think logically ○ Kahneman - people use heuristics based on representativeness and stereotypes ● People working on heuristics and biases think they are formally characterizable (rules are just different from logical rules) ● Reasons based on stereotypes formally characterizable? ○ They might have to do more with semantic content rather than logical rules ○ Still, cognitive science is interested in formally characterizing these reasons ○ Other cognitive scientists lean towards explaining these reasons through mental contents, without a promise that the operations will ever be formally characterizable...