Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
45 Cards in this Set
- Front
- Back
What is cognitive science, in terms of what cognition is?
|
- Cognition is information processing, and operates on representations (symbols that have meaning)
- CS takes a scientific approach: we develop theories to explain phenomena - study of information and representation processing can be divorced from biology and physiology, where computers can make talking about representations more precise - CRUM |
|
CRUM: Computational-Representation Understanding of Mind
|
- rather than just looking at an environment to explain behavior, we should look at representations
- to learn about representations, we look towards linguistics, psychology, neuroscience |
|
What are the criteria for a theory of thought?
|
we need a mathematical language that can formally express our thoughts and link them to the world
- similarly to how calculus relates to motion in the world, and geometry to location in the world |
|
what do mathematical theories of thought do?
|
they should "represent the relationships between our thoughts"
they should be an algebra that vields valid inferences |
|
what are syllogisms and what is Aristotle's relationship to them?
|
Aristotle tried to compile together all syllogisms,
these are logical arguments where one proposition is inferred from two others |
|
What did Leibniz try to do?
|
he tried to create an algebra that yields valid inferences, a mathematical language that could formalize our thoughts
|
|
What is Boole's logic what are its limits?
|
x=1 is true, x=0 is false. it locks us into very simple conclusions and doesn't have quantifiers (existential or universal)
|
|
modus ponens truth table
|
X Y X => Y
___________ T T T T F F F T T F F T |
|
modus ponens proof
|
X => Y
X ___ Y |
|
modus tollens proof
|
X => Y
not Y __ not X |
|
how is frege's modern logic different than logics before it?
|
it uses proposition logic that takes the form proposition(object)
hairy(Rex) = "Rex is hairy" IsAGSI(Paul) = "Paul is a GSI" |
|
Entscheidungsproblem and its conclusion
|
can we make a machine that can take premises in first-order logic and indicate whether conclusions are true or false?
- most likely impossible, since it would have to know every fact in the world. this would involve some knowledge of semantics |
|
define a Turing machine
|
a computational machine that runs on a set of rules (its program), has a head that marks numbers on a long tape. the head can move along the tape as it likes, as long as it is written into the program
|
|
halting problem
|
once a Turing machine starts to compute, there's no way of knowing whether it will terminate in an answer
|
|
difference between hardware and software
|
hardware is the physical structure while software is the the program that this machine runs on
|
|
what are symbols?
|
symbols are representations that have meaning. they represent something to us. (e.g. letters in a certain language represent sounds)
|
|
symbol grounding problem
|
our minds, and especially robots' “minds”, may not know when the facts that correspond to our symbols are true (e.g. What is red?)
|
|
Descartes' assumptions of the mind and the problems and wonders these entail
|
Descartes believed that the mind and body are separate since we can deny belief of our body but not our mind (we affirm our mind by thinking).
Problems this poses: how do the senses communicate with the mind, and how does the mind cause the body to act? |
|
our mind and changing symbols
|
symbols can change as long as they have the same relationships
the world can change as long as the symbols have the same truth values (but this would lead us into a Matrix-like situation) |
|
Descartes' solution to the mind-body problem
|
humans have a pineal gland where something can be sent across, from the brain to the body. conceptually he was right, but none of his details were.
|
|
universal Turing machine
|
a Turing machine that can simulate as many other Turing machines as it wants, becoming something that seems more like a brain
|
|
What does cogsci allow us to do with minds, in terms of our level of analysis?
|
we're able to study any type of mind at a level of abstraction above its physical form.
- one level of analysis: physical. hard to understand how our brain might work on this level - another level: mathematical. study our thought by formalizing it into a mathematical language |
|
Plato vs. Locke, nativism vs. empiricism
|
Plato: we remember facts from before our birth. What we see are shadows of an ideal world that are close enough for us to see it as this world
Locke: blank slate. we learn new facts about the world through experiences and perception |
|
induction vs. deduction
|
induction: inferring truths from our own experience (so our conclusions could be wrong)
deduction: forming new truths based on other truths |
|
complex vs. simple ideas
|
complex ideas are formed from simple ones. a unicorn doesn't exist, yet we can still imagine it. it's simply a horse with a rod on its head.
|
|
why might we worry about a science of the mind (Wundt's approach to psychology)
|
too introspective, relies on personal subjectivity for answers and experiments
|
|
Ebbinghaus' approach to psychology (less subjective)
|
tested his memorization every few days, collected data and made conclusions
|
|
definition of behaviorism and the assumption it makes
|
the environment explains behavior.
assumption: there's no substantial difference between humans and animals, so we can study animals if we want answers |
|
classical vs. operant conditioning
|
classical: associating a certain cue with another cue, eliciting a response
operant: using consequence to change behavior - punishment and reinforcement |
|
radical behaviorism
|
disregards mental states and representations, and mental processes. we can understand behavior solely from the environment.
|
|
Saussure and arbitrariness of signs
|
no inherent direct connection between language symbols and meaning, the signifier (sound/written form) and signified (corresponding concept)
|
|
synchronic and diachronic linguistics
|
synchronic: the study of a language at this point in time
diachronic: the study of the evolution and history of a language |
|
langue and parole
|
langue: abstract rules of a language
parole: how these rules are used in speech (rules of chess : chess strategy) |
|
phonology, morphology, syntax semantics
|
phonology: sounds of a language, combinations of sounds
morphology: how words turn into other words (tense, suffixes, prefixes), a morpheme is the smallest unit of meaning syntax: how words are combined to form sentences (arrangement, order) semantics: meaning |
|
how behaviorism applies to language
|
language is verbal behavior
problem: language is much too complicated. we can't predict language from environment and this doesn't explain how humans can understand novel sentences |
|
mathematical structure of language (Shannon)
|
language is just associations between words and letters. we can use Markov models to predict the next word or letter based on the last one.
problem: phrases in sentences can have arbitrarily long dependencies on other phrases in the sentences |
|
Logic Theorist (Simon and Newell)
|
- solved mathematical proofs using human heuristics (problem-solving strategies) when necessary
- used means-ends analysis: took whichever step made it closer to the end - |
|
new ideas from Cognitive Revolution (Newell, Chomsky, Miller)
|
- we can make computer programs that resemble human thought, and we can base these programs on human thought
- we can explore the rules that characterize language and thought through behavior - we can make inferences about the constraints on mental representations from behavior |
|
how did Chomsky counter the validity of Markov models?
|
colorless green ideas sleep furiously. this has never been seen before, and neither word has associations with the one before it, yet it is still grammatically correct.
|
|
Miller's "Magical Number 7"
|
we can remember up to seven chunks in our short-term memory.
this is an example of looking at behavior to make inferences about human thought. this doesn't explain thought however. - how can we measure capacity of the mind? - can it be used as a system of information processing? |
|
how can we study representations? (trick question...)
|
this is what cognitive science is! we use psychology, linguistics, neuroscience, etc.
the mapping from environment to behavior is complicated--representations simplify this mapping |
|
factors contributing to the defeat of behaviorism / Chomsky's argument
|
- Cognitive Rev presented ideas about using representations to study the mind. there's a complex process our minds go through between the environment and our behavior, and through representations, we can formalize it.
Chomsky: definitions are very hazy. "environment" refers to a ton of stimuli, so how can we predict behavior from the environment if we don't know what stimulus someone is responding to? - learning is usually not done through operant conditioning reinforcement, but rather, observation -"We cannot predict verbal behavior in terms of the stimuli in the speaker's environment, since we do not know what the current stimuli are until he responds" |
|
differences between Boole, propositional, and first-order logic
|
Boole is based on true/false, propositional can link an abject and a quality, and first-order can generalize about many objects and their qualities
|
|
first-order logic: which connective with which quantifier?
|
universal quantifier: =>
existential: ^ |
|
logic tips
|
- with each quantifier, we must define x and y in the sentence. Man(x) ^ Woman(y)
|