Midterm 1 Flashcards
What Is Cognitive Science?
•The study of cognitive processes involved in the acquisition, representation and use of human
knowledge.
•The scientific study of the mind, the brain, and intelligent behavior, whether in humans, animals,
machines or the abstract.
Introspection (until 19th century)
process of self-reflection
- could only be done by carefully trained individuals
- allowed for ability to observe conscious experience
Experimental psychology (1879-)
- Wundt
- discipline of psychology separates from philosophy
- Trained ”observers”
- Measured reaction times & word associations
- Rigidly controlled experimental procedure for describing sensations (e.g. “experimental conditions”, multiple repeated observations)
Behaviorism
- John Watson
- attempted to replace ‘subjective’ with ‘objective’ data based purely on observation of behaviour
- focused on directly observable behaviours
- people are ‘responders’ whose responses are simply products of conditioning through their experience
- learning occurs as a result of consequences of behaviour
Behaviourist Model
- Stimulus (in environment)
- Black box (can’t be studied)
- Response (behaviour)
Tolman’s Cognitive Maps
- noted that conventional behaviourists can’t explain phenomena like knowledge, thinking, planning, etc. –> don’t believe in mental phenomena
- organisms do not require stimulus-response associations only by contiguity or reward –> they selectively take in info from environment and build cognitive maps as they learn
Tolman: Latent Learning
- 3 groups of rats allowed to explore T-maze
- No reinforcement, regular reinforcement, reinforcement after 11 days
Results: Reinforcement group showed steady improvement
Reinforcement after 11 days showed sudden improvement after 11th day
No reinforcement doesn’t appear to learn much
BF Skinner’s response to Tolman
- mentalism: this way of thinking which he found flawed
- Tolman said one cannot directly observe cognitive maps they are inferred from behaviour
Minsky: Steps toward AI
computer program capable of describing structures would need:
- terms for relations (inside, left, etc.)
- a way of specifying hierarchical level at which these relations are specified
Skinner’s Verbal Behaviour
- behaviourist explanation of language development
- function: efficient way to get another individual to do something
- all other functions (communication, truth, logic, etc.) are derivatives of this primary function
- should only think of language behaviours in terms of measurable effect they have on human interaction
Noam Chomsky and the Cognitive Revolution
- behaviourism is successful in describing low-level instinctual behaviour
- BUT language can’t be solely acquired by reinforcement –> must be innate structure that helps humans acquire language
- if language was only acquired through reinforcement then it would take longer than 2 years to understand basic structure
Two observations about language learning
- poverty of stimulus: even while hearing a finite amount of sentences, can produce an infinite number
- constraints and principles cannot be learnt: children don’t know anything about grammar or syntax but can produce grammatical sentences
Aspects of a theory of language/syntax
- makes infinite use of finite means
- languages differ enormously
Central Tension: ease of language acquisition (innate?) and linguistic diversity (acquired?)
Chomsky’s Proposal
Infants are born with a Language Acquisition Device (LAD)
- helps them use input more effectively
- provides little bit of knowledge about how human languages work to get started
Criticisms of LAD
- how could it work? (black box)
- languages are so diverse that such universality is rare
- only seems to touch on syntax/structure and not semantics
What fields did Chomsky’s refutation of Skinner’s views of language revolutionize?
- cognitive psychology
- linguistics
- computer science
- analytic philosophy
Bottom up processing
start with distal stimulus
(sensory data) and build
up representation
Top-down processing
start with expectations & context to help sense/interpret incoming data stream Later stages of processing affect earlier stages (ex: 3D shape perception)
Top-Down Processing: the Single Light Source Assumption
- visual systems’ interpretation of shape is constrained by the rule
that there can only be one light source - seems to have the built-in assumption that light is
shining from above
Cigarette size: influence on perceptual judgment
- high smoking need: deprived of smoking in class before study
- low smoking need: permitted to smoke immediately before study
Results: low smoking need participants judged length of cigarette to be shorter than those is high smoking need condition
stimulation-onset asynchrony (SOA)
separation between the onsets of the two stimuli in double-stimulation paradigm
serial info processing model
- items tested one at a time
- predicts linear increase in response time as set size increases
parallel info processing model
- items tested simultaneously
- if response time increases as memory set increases, then not parallel processing
Single channel hypothesis
- first piece of info must be cleared before second can be processed
- mind can only deal with one piece of info at a time
- when several pieces are received in rapid succession, bottleneck is formed (causing slowing of decision making)
Double-stimulation paradigm
If we detect a stimulus (S1) but we are still processing
that information when a second stimulus (S1) occurs, we will be unable to process the second stimulus until we
have finished processing the first on
Sternberg Task
- give subjects short list of items to remember
- subjects decided quickly whether a new item was on the list or not
psychological refractory period (PRP)
slowing of RT2
Marr’s Approach
- Distinguished different explanatory tasks at different levels
- Gave a general theoretical framework for combining these levels
Computational Level
•Form of task analysis of a cognitive system
(a) Identify the specific information-processing
problem that the system is configured to solve (inputs, outputs)
(b) Identify general constraints upon any solution to that problem
Algorithmic Level
•Explains how the system actually performs the information-processing
task
• identifies input information and output information
• identifies algorithm for transforming input into required output
• specifies how information is represented
Physical Implementation
Level
•Physical realization of the system (e.g. the neural hardware) • Identify neural structures realizing the basic representational states to which the algorithm applies [e.g. populations of neurons] • Identify neural mechanisms that transform those representational states according to the algorithm
Which level is critical from an information-processing pov?
computational level
Marr’s analysis of the visual system
- identified two key jobs to perform
- -> provide a 3D representation of the visual environment
- -> provide object-entered rather than viewer-entered frame of reference
