Sem2 Week 1
No description yet. This curated deck is ready for daily study.
Card preview (24)
Example of conditioning changing physical brain structure
London taxi drivers having a larger hippocampus
Example of numerical problem-solving without linguistic basis
Amazonian Pirahã tribe, with no linguistic expressions for numbers
Define mental representation
A description of information in the mind
Advantages of cognitive modeling
• formalisation allowing implementation • makes scope, assumptions, and repercussions explicit • uncovers poorly defined aspects • prevents atheoretic testing • allows systematic comparison and new hypothesis
Guest and martins cognitive modeling framework
Framework (concepts in world) Theory (natural language) Specification (formal) Implementation (eg. Program) Hypothesis (narrow and testable) Data (observations from world or model)
Give an example of a framework (according to Guest and Martin’s levels)
Variables in the problem Eg. In the pizza problem having orders, food, pizza area…
Give an example of a theory (according to Guest and Martin’s levels)
Natural language descriptions of causal relationships Eg. X causes Y People choose the option with the most food
Specification example (according to Guest and Martin’s levels)
Formal system Eg. Equations Food = πr²
Implementation example (according to Guest and Martin’s levels)
Programs and simulations
Hypothesis example (according to Guest and Martin’s levels)
Testable statement Eg. Participants will choose the larger pizza 85% of the time
Data example (according to Guest and Martin’s levels)
Experimental results, behavioral data…
David Marr’s three levels
Computational (problem to solve) Algorithmic (procedure to solve) Implementational (actual solution)
Significance of Marr’s levels
Shows that one level bottom up cannot automatically give another Eg. Neuron implementation doesn’t explain brain algorithm We should work top down, like in a tree structure
List all arguments supporting language as a cognitive technology. (Evidence against nativism)
Language diversity, rapid change, made to be learnable, we have good general purpose pattern finding abilities, animal word learning abilities
List all arguments opposing language as a cognitive technology (i.e., supporting innateness).
Poverty of the stimulus; Linguistic universals; Critical/sensitive period effects; Feral/deprivation evidence (Genie); Neural specialisation for language (Broca’s area and Wernicke’s area)
Cultural transmission
Rapid cultural change; e.g., Old vs Modern English ~1000 yrs
Linguistic diversity
Structural cross-language variation; e.g., SOV (English) vs SVO (Japanese) word order.
General statistical learning
Domain-general pattern learning; e.g., children track syllables frequently co-occurring.
Animal word-learning parallels
Non-human label learning; e.g., dogs mapping object names Dror et al. compared to 15yr child. Alex parrot.
Cultural evolution
Languages adapt to learners; e.g., English lost noun genders to help learners
Critical period
Age-limited acquisition window; e.g., 0-9 to achieve native fluency. JB Tannenbaum et al.
Deprivation evidence
Early exposure required; e.g., feral children deficits. Genie US 1970
Neural specialisation
Dedicated brain regions; e.g., Broca’s area damage = lost grammar, still meaning Wernicke’s = still grammar, no meaning
Arguments surrounding broca’s area specialisation
Highly biologically organised - supports innate language Counter argument: not fully specialised because music/ maths also activates it