Intro to Cog Sci final notes (Fall 2013)

25 Pages
Unlock Document

Cognitive Science
Terry Regier

Cognitive Science Finals Notes 12/9/2013 6:29:00 PM INNATENESS  Rationalist vs Empiricist o Rationalist: knowledge comes partly from reason, some innate; Plato, Leibniz, Boole, Frege, Turing o Empiricist: neither principles nor ideas are innate, knowledge comes from experience o Aristotle = blank slate o Plato = ―The Meno‖ (‗like the sting ray we have made him numb‘) learning is remembering, slave boy remembers cutting edge geometry o Plato‘s problem: How can we know so much given so little evidence? o Poverty of Stimulus argument: limited experience not enough to account for the total of our knowledge as well as concepts like infinity (something we can understand but never actually experience); in linguistics, refers to the argument that linguistic input received by young children is in itself insufficient to explain children‘s detailed knowledge of their first language o Locke: white paper, all knowledge comes from experience  Two sources of knowledge: objects of sensation (colors, tastes, etc) and the operations of our minds (reflection, thinking, doubting)  Children acquire knowledge gradually by degrees, something not predicted by theory of innate knowledge o Leibniz (on Locke): experience is necessary but insufficient; need some innate knowledge  On necessary truths: no matter how many times something occurs (instances) one cannot establish a universal truth (apple falls to the ground 100 times but one cannot be certain it always will)  Veins in the marble o Induction: bottom-up logic, set of observations that create a general law o Deduction: top-down reasoning, conclusion follows directly from evidence given  If AB and BC then AC ALPHABET OF HUMAN THOUGHT  Syllogism: premises (statement A and statement B) that result in a conclusion different from the premises (statement C) o ie Given all As are Bs and all Bs are Cs then all As are Cs  Aristotle‘s project: catalogue valid syllogisms  Aristotle‘s achievement: determined the validity of syllogisms, categorized them, and reduced some to proofs  Leibniz: fundamentally optimistic, benevolent God, criticized in Camadie  Leibniz‘s ―Wonderful Idea‖: we need an ideal language that ‗perfectly represents the relationships between our thoughts‘ o A) the Universal Characteristic: compendium of all human knowledge; described in key underlying notions and put into symbols; reduce rules of deduction to manipulations of symbols  AI ahead of its time! o B) Language that aids clarity of thought, notation does most of the work o C) ―Let us calculate without further ado to see who is right‖  Boole: prepositional logic = starts to represent sentences in terms of classes/sets; law of thought, explains Aristotle‘s syllogisms o Evaluating truth of formulas o Foundational rule: xx=x (set A and set A is set A) o Other rules: x+(x+1) = 1 (set A taken with everything not set A is everything) and o ―If you‘re Leibniz and a cookie then you‘re dipped in chocolate‖ = L  C ^ D ―If you‘re neither Leibniz nor a cookie then you‘re not dipped in chocolate‖ = 7L ^ 7C  7D LOGIC  Frege: Begriffsschrift = concept script = first order predicate logic o Built off Boole‘s prepositional logic, step closer to Leibniz‘s dream  Instead of prepositions, uses predicates to denote properties of objects o ie hairy(rex) ^ smelly(rex) = Rex the dog is smelly and hairy o Adds upside down A for ―all‖ and backwards E for ―there exists‖  ie All penguins are black and white = AxPenguin(x)  Blackwhite(x)  ie Everybody loves somebody = Ax Ey loves (x,y)  Contains truth values; model formula is assignment of truth values to make the formula true  Russell‘s letter to Frege: Frege used sets of sets, opens possibility that a set can be a member of itself o Extraordinary sets contain themselves o Ordinary sets do not contain themselves o Problem: set O is a set of all ordinary sets (a set of sets that do not contain themselves)  It can‘t belong to itself but that would make it ordinary so it would have to belong to itself (fail) o Paradox can later be avoided  Frege did not provide a procedure that could tell if conclusion C followed from premises P INFINITY  A rationalist challenge to empiricism o ie our experience is finite but we can conceive of the infinite  Diagonal method used to create new subset from given subsets  Cantor uses diagonal method to demonstrate infinity in multiple sizes (countably vs uncountably infinite) THOUGHT AS COMPUTATION  Hilbert decision problem (Entscheidungsproblem): wants a procedure that accepts premises P and conclusion C and determines whether C follows from P  Turing used diagonal method to show that no such procedure exists (halting problem: diagonal method used to create a set that is not the halting set for any T.M.) o End of Leibniz‘s dream  Turing Machine: an abstract machine M that has o 1. An internal state of finite number of possible states Q o 2. An infinite input tape where symbols can be read & written o 3. A finite set of rules that specify what M should do as a function of what it sees on the tape  Universal Turing Machine: Turing Machine that can simulate any Turing Machine and rules are included on tape  Universal Turing machines as general-purpose computers, distinction between program and data.  The mind and the computer: different hardware, same software. THOUGHT, COMPUTATION, AND THE WORLD  Can describe world in formulas and use syntactic operations on those formulas to discover new things that are true about world  Logical formulas contain prepositions that correspond to things in the real world with assigned truth values  Symbol grounding problem: we need to know when the facts about the world corresponding to our symbols are true  View of the mind o Senses provide us with info about what facts are true of the world o We use inference to derive new facts from this knowledge o Our inferences lead us to take actions that provide us with more information (loops again)  Descartes: deceiving (doubting) senses, evil genius problem, only thing that is certain is that I am a thing that thinks; empiricist  Dualism: mind and body are separate entities and cannot explain the mind in physical terms o Mind-body problem: If separate entities, how do you receive info from sense or move an arm?  Materialism: mind can be explained in physical terms and understood through science and physics BEHAVIOR AND THE MIND  Boole about the mind: scientific method cannot be applied to the mind because there ―the general truth is seen in the particular instance, and it is not confirmed by the repetition of instances‖ versus in nature, one makes hypotheses and inferences o Mind is special, thinking about thinking enough to infer general laws o Worries for science of the mind  Thoughts are intangible  Must study something very intimate in a disinterested way  Complicated mind to be reduced to simple laws and principles o Wilhelm Wundt: Father of experimental psychology, ―complication apparatus‖ used to measure reaction time  Must press button when tone is heard: when told to focus on tone, reaction later than when told to focus on button  Apperception: the process of making an experience clear in the conscious, requires time and effort  Inferring an unobservant mental process from observed behavior  Introspection: self-reporting of experience, subjective o Hermann Ebbinghaus  Measured memory retention  List of items: forgetting curve, spacing effect (list of items better remembered when studied at spaced intervals rather than all at once)  Recorded objective measures  NO INTROSPECTION o Behaviorism (1950s): analysis of behavior rather than mind, emphasis on environment and experience, no significant difference between man and animal, empiricist  Watson: behaviorist, no introspection, ―no dividing line between man and brute‖ (extreme)  ―Give me a dozen healthy infants, well-formed, and my own specified world to bring them up in and I'll guarantee to take any one at random and train him to become any type of specialist I might select -doctor, lawyer, artist, merchant-chief and, yes, even beggar-man and thief, regardless of his talents, penchants, tendencies, abilities, vocations, and race of his ancestors.‖  Behavior observed in animals  How animals learn  Classical conditioning: Learning that one cue (conditioned stimulus) is associated with another that naturally elicits a reaction (unconditioned stimulus) o ie Ivan Pavlov dog experiment o ie Little Albert noise/rat experiment  Operant Conditioning: Learning that performing an action results in a reward (reinforcement) or punishment o ie Edward Thorndike cat in puzzle box experiment  B.F. Skinner: radical behaviorist, no mental states, ―our behavior has changed but there is no evidence that we have acquired knowledge‖  Skinner box: devices for carefully studying learning and reinforcement in animals THE COGNITIVE REVOLUTION  Tolman: latent learning, cognitive maps o Latent learning: different groups of rats ran the maze (rewarded, unrewarded, rewarded only after 10 days)  rats rewarded after 10 days learned more quickly  Had latently learned the maze and then when reinforced, could build upon that latent learning, in an apparent ―insight‖ (quick learning) o Cognitive maps: rats did not merely associate actions with rewards but rather had a ―cognitive map‖ of the maze (mental representation of the world)  Mental representations key to cognitive science  Sets apart from behavior  Newell and Simon: ―The Logic Theory Machine‖ paper presented at the Symposium on Information Theory at MIT 1956 (birthday of cognitive science) o First AI system o Proofs of mathematical facts expressed in inference rules o Heuristics inspired by human problem-solving strategies  George Miller: ―The Magical Number Seven (plus or minus two) paper at same symposium o Limits on our capacity to process information o Can hold about seven ―chunks‖ of information in memory o Hierarchal structure of chunks within chunks (phone number digits into area code + office code + extension etc)  Lashley: plans are hierarchal (steps to go to SF from here)  ^Some thought is hierarchally structured  Noam Chomsky: ―Three models of language‖ paper at same symposium o Language is hierarchally structured  Chunking (ie a noun phrase)  Results of these papers/symposium!!! o Birth of Cognitive Science o Behaviorism isn‘t sufficient (Tolman shows mental representations) nor necessary (Newell & Simon, Miller, Chomsky show rigorous science can be accomplished w/o extreme assumptions) LANGUAGE  Whitehead: mentor of Bertrand Russell (letter to Frege) and co- authored Principia Mathematica (which built on Frege and sought to ground all mathematical proofs in logic) o Dinner conversation with Skinner: explain my behavior as I say ―No black scorpion is falling upon this table‖  Mental representation necessary  Skinner believed animals/humans respond exclusively to stimuli so difficult to explain this hypothetical situation  Skinner publishes Verbal Behavior in response 23 years later, proposing an account of human language in behaviorist terms o Negatively reviewed by Chomsky  Chomsky: rationalist, language reflects knowledge, mental representations necessary, some knowledge innate o Hierarchal structure, chunks of language o We know rules about which sequences of words can be English sentences (syntax) o Chomsky‘s general rules of language similar to Leibniz, Boole, Frege rules of thought  Knowledge is rule-based so it can apply to new things o Infinitely generative  Colorless green ideas sleep furiously  Poverty of stimulus argument: the structure couldn‘t have been learned from the linguistic input that children receive o No negative evidence (very little: ―don‘t say this‖) o Therefore innate knowledge  Language Acquisition Device  Universal Grammar  Language and language learning are constrained: some logically possible hypotheses are never considered.  The constraints are specifically linguistic in nature – rather than stemming from more general cognitive forces. THE DISCIPLINE MATURES  Symbolic vs imagistic representations o Symbolic: language-like, prepositional; ie Frege, Chomsky, Newell & Simon o Imagistic: experience-like, associative  Shepard and Metzler: mental rotation  Kosslyn: scanning  Marr‘s 3 Level Framework o Computational theory: functional goal (generic description of problem) o Representation and algorithm: rules by which the hardware bits are working to achieve that goal (rules, laws of physics…) o Hardware implementation: physical implementation or mechanism  ie comp theory: need to fly, algorithmic: aerodynamics, implementation: feathers/steel, etc  Marr‘s Model of Vision o Sees image: 1) primal sketch 2) 2.5D sketch 3) 3D sketch o Continuum from imagistic to symbolic o Hierarchal structure in perception  Representational level THE TURING TEST AND ITS CRITICS  Theories of the mind driven by current technology o Blank slate  White paper  Ipad w/o apps  Physical symbols hypothesis (Newell & Simon) o Newell and Simon (1975): A physical symbol system has the necessary and sufficient means for general intelligent action. o ie problem-solving! o Physical Symbol Systems o 1. Symbols are physical patterns. o 2. Symbols can be combined to form complex symbol structures. o 3. Contains processes for manipulating complex symbol structures. o 4. The processes for generating and transforming complex symbol structures can themselves be represented by symbols and symbol structures. o Syntax: the identification and manipulation of such symbols based purely on their shape. o Semantics: the meaning and changes in meaning that these syntactic manipulations are meant to correspond to. o Search space: Space of possible problem solutions – must search through this space to find optimal solution. o **Intelligence as search through a symbol structure  Turing test o Alan Turing: The world's first and most influential computational scientist. o Imitation game: human or machine?  Criterion: can convince you it‘s a human  Linguistic in nature o Chinese room argument  John Searle  Turing test is an inadequate criterion  Searle (who speaks no Chinese) receives input in Chinese and through syntactic symbol manipulation based on symbol shape, produces appropriate output in Chinese  Still has no understanding of Chinese  ARGUMENTS/REPLIES  Systems reply: John doesn‘t understand Chinese but the system as a whole does  Searle: Even with rules memorized and after becoming the system, still no understanding of Chinese  Robot reply: true that room doesn‘t understand Chinese but not because it‘s symbol- based…instead it is because it has no embodied experience to link its symbols to  Searle: Even if grounded, they‘re still just meaningless symbols  **symbol grounding problem: problem of how words and thoughts become meaningful to speakers and thinkers while the problem of intentionality refers to how words and thoughts connect up in the world THE TURN TO THE BRAIN  Parts of brain: o Forebrain o Midbrain o Hindbrain  Lobes of cerebral hemisphere: o Frontal- motor control, speech, planning o Parietal- spatial, sensory integration o Temporal- objects, auditory o Occipital- vision  Two hemispheres o Some cerebral asymmetries o Right-handed people have more areas in their left hemisphere associated with language. o But most functions are in both hemispheres o Contralateral organization o Information from left visual field is processed in the right hemisphere, and vice versa.  More CP (categorical perception in right visual/left hemisphere due to language and category processing) (CP=Faster or more accurate discrimination of stimuli that straddle a boundary)  Two streams hypothesis o Ungerleider and Mishkin monkey experiments o Identification versus localization  Interpretation: parietal cortex involved in spatial recognition and receives most of the information from visual cortex in same hemisphere o So damage to right parietal cortex leads to left hemineglect o Dorsal- ―where‖ pathway o Ventral- ―what‖ pathway o Place cells in hippocampus: a neural correlate of cognitive maps.  From rats running around a triangle track spatial firing of place cells  Words in the brain: serial versus parallel processing models o Serial model of word processing  Visual input  auditory form  semantic processing (meaning)  articulation (speaking)  ie passively view words  passively listen to words  generate a related verb  speak a visually presented word o Image people’s brains while they participate in tasks that tap into these hypothesized processes  Problem: some tasks tap into more than one ―functional box‖  Solution: subtraction method‖ = subtract the image of one task from th
More Less

Related notes for COG SCI 190

Log In


Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.