This site is 100% ad supported. Please add an exception to adblock for this site.

Cognitive Science 2

Terms

undefined, object
copy deck
What is a decision problem?
given a particular input, in produces a specific output.
What are the four components of a turing machine?
1. machine table-the program
2. state-a # that represents the state
3. tape-has infiinate length, write 1 or 0
4. write head
what are do each or these numbers represent? 00.01R
1. first 0 is machine's current state
2. second 0 is what machine reads
3. third 0 is new state
4. the one is what it writes on tape
What is a universal Turing machine?
it can imitate any other turiing machine and can compute any computable functions.
What is The Turing-Church hypothesis?
if a problem can be computed by a Turing machine, it can be computed by any reasonable computer.
what makes a function computable?
1. there is a finite # of steps
2. it makes use of finite set of operations
3. comes to a hault for any valid input data
is our mind like a Turing machine?
no, because we don't have unlimited memory. we use heuristics. we use induction.
Who wrote ELIZA and what did it do?
Weizenbaum wrote ELIZA. It tricked people into thinking it was an intelligent psychologist. People thought of it as first steps in AI research.
what is strong and weak AI?
weak AI: computers can simulate real mental processes
strong:the idea that an appropriately programmed computer is a mind
do computers use syntax or semantics?
computers: syntactic engines
humans: semantic engines
what is the language of thought?
LOT is the idea that just as we form sentences syntactically, we also think syntactically as well. it proposes that we think in terms of symbols just as we speak in terms of works.
What was the change in studying AI, from GOFAI to Brooks?
GOFAI was top down and Brooks decided from bottom up.
What did Newel think brains and computers where? hint: symbol or connectionist
physical symbol systems.
what are the properties of symbol systems? what are properties of symbols?
memory, symbols, operations.
symbols can designate things in the world or the can represent operations.
in a symbol system, what does cognition boil down to?
it boils down to searching for the appropriate operatot. humans are efficient through heuristic search.
Who made GPS? What did it do?
Newell and Simon made it. it was able to seperate task-independent and task-dependent tasks. it seeks operators that will take it closer to the goal state.
what can operators of GPS do?
1.compare sybol structures
2. creat new symbol structures
3. read input and write output
4. store sybol structures to memory
what is means-ends analysis?
it is applying the operator that eliminates the most diferences between the current state and the goal state.GPS brought means-ends analysis to AI
what is the importance of GPS?
one of the first attempts to provide a general theory of problem solving behavior.
What are production system models?
what are two models of production systems?
they are models most commonly used in cognitive science today. basically they are if-else statements. ACT-R and SOAR (cognition can be modeled using single architecture)
what is an expert system? what are its components? example of one
1.knowledge base- has lots of knowledge that is gathered from experts.
2.inference engine- has ability to use encounter situation, use knowledge and make a decision. MYCIN and CYC
what are problems with expert systems?
1. they can't learn
2. they don't have intuition
What is CYC? what are its problems?
a program that attempts to do common sense. they often rely on its context.
what is the problem with symbolic AI?
brittleness: the system fails whenever it tries to do things beyond its given knowledge. and frame problem.
how do neural networks compute in comparison with computers?
slow and noisy vs. fast and error-free
plastic connections vs. fixed architecture
what is a perceptron? what type of logic does it use? what types of problems can it be used for?
simple neural network. uses Boolean logic. AND and OR.
with regards to perceptrons, which problems are linearly seperable? which aren't?
linear- AND and OR
not-XOR
What is a PDP model?
a more complex neural network.
what are two rules for learning connectivity?
Hebb rules-increase connectivity when two fire together?
Delta rule: change weights in proportion to the amount of error b/t desired and actual output.
what makes PDP models better than perceptrons?
they have multiple layers that can do more complex things?
what are different methods of machine learning?
unsupervised. reinforcement, supervised
what are examples under unsupervised learning?
Hebbian learning-repetition of neurons firing.
ocular dominance-deprivation of one eye, reduces representation in cortex
what two modesl represents the reinforcement learning?
actor/critic model
TD Gammon
example of supervised learning?
NetTalk
what are the differences between connectionist and symolic models?
connectionist models do not explicitly store symbols, but store info in connection weights.
what has been used more often over the past 20 years?
connectionist approach
what is the hard problem of consciousness?
how our consciousness relates to our subjective experience
what is the difference between implicit and explicit distinctoin?
explicit are memory you are aware of in the past
implicit - the memory of how to ride a bicycle
what is the difference between declarative a nondeclarative memory?
declarative-knowing events -"knowing that..."
nondeclarative-"knowing how"
eg skill learning, repetition priming
Explain double dissociation between memory systems?
amnesic patients-good repetion priming, bad recognition memory.
occipital lesion-bad repetition priming, good recognition memory
what is subliminal perception?
people can be afftected by things they are not even consciously aware of
what is subliminal priming?
when you see bread, you are faster to say butter.
is subliminal priming real?
there is evidence that people feel it is real but people are aware of some queue, just don't know what it is.
is backwards speech something we can determine?
we can determine gender, different voices. can't tell how many words. can't tell if it is sensible or not
so what? what do we actually know about unconscious perception?
what about subconscious propagnosia?
the effeects are small and limited.
even if we cannot recognize faces, we have agalvanic skin response (GSR) that rises when we recognize someone.

Deck Info

45

permalink