CGSC 170 exam 3 study guide- Jasinska PDF

Preview

Summary

summary of lecture notes using both slides, notes taken based on professors explanations, and the textbook.
course with prof Jasinska....

Description

●

●

●

●

●

●

Disciplines in cognitive science ○ Different disciplines can study a single topic in different ways ○ Cognitive science can incorporate many disciplines: psychology, artificial intelligence, anthropology, philosophy, linguistics, neuroscience, etc ○ Some ideas are language, consciousness, memory, learning: how they are all integrated! The integration challenge ○ Cognitive science is an interdisciplinary field ○ The challenge: how do we come up with an overarching theory of cognition that shows the relationships between all these disciplines, and their common ground? ■ THERE ISN'T AN ANSWER TO THIS YET! Local integration: Reasoning and evolutionary psychology ○ Human beings aren't always good at logical reasoning… but we seem to improve in specific contexts ○ Evolutionary psychologists Cosmides and Tooby believe that this is due to having evolved context-dependent rules: special modules that help us solve certain social situations Cosmides and Tooby ○ Cassava root study: people are more likely to give logically correct answer to social exchange based Wason selection task, vs. an abstract task (like numbers and vowels) ○ If the task is switched so that you are looking for people who receive a cost without benefit, people don’t perform as well ○ Conclusion: Humans have evolved a cheater detection module that helps us specifically look for people who are receiving a benefit without a cost: in other words, cheaters! Cheater detection ○ Why might there be a cheater detection module ■ So that we can identify whether people are taking advantage of us in a social situation, vs. cooperating with us, and change our actions accordingly Modularity ○ Modularity of the mind ■ The mind has different modules, and each module is for each specific kind of function ■ This is the theory ○ Massive modularity ■ Theory that there are separate modules in the brain that function on its own and focuses solely on its own function ■ This is a hypothesis about the modularity of the mind theory ○ Fodor Modules ■ Fast ■ Mandatory ■ Cognitively impenetrable ■ Domain-specific ■ Encapsulates small amount of info ■ Examples of Fodor modules are color perception, voice recognition, sentence structure processing, face recognition, etc. ○ Massive modularity ■ Key features ● Lots of domain specific modules ○ Darwinian modules ● Each module for a function kind of function

● Rejects central processing Darwinian modules are domain-specific!!! ● Examples are kin selection, cheater detection, and mate selection ■ Only fodorian modules are cognitively impenetrable! Darwinian modules can talk to each other and are not cognitively impenetrable! ○ Fodorian modules ■ Domain specific ■ Mandatory application ■ Not much info in each module ■ Fast speed ■ Assumes central processing ○ Darwinian modules ■ Domain specific ■ Not mandatory application ■ Lots of info in each module ■ Unknown speed ■ Denies central processing Object permanence and physical reasoning ○ Traditional view- william james ■ Neonates or newborn infants do not differentiate between self and others. They are only capable of reflex actions ○ Jean Piaget ■ Babies are born with some innate capabilities such as sucking nipples which are reflex like sensory motor schemas. They gradually bootstrap these schemas into complex behaviors circular reactions ■ At the age of two, also the end of sensorimotor stage, hey understand the difference between self and others ○ Dishabituation paradigm ■ Main idea: infants look longer at events they find surprising ■ Done by ● Habituating the infant to a certain type of event and then present them with an event that differs from the previous event in particular ways ● Looking time measures are calculated by measuring the amount of time spent looking at the object ■ Helps experimenters detect the events that the infants found surprising ■ Work out what the expectations of the infants could have been ■ Violation of the expectation is known as the dishabituation paradigm ○ Baillargeon's drawbridge experiment ■ Habituated children (4.5 months old) to the 180 degrees rotation on a table ■ Interested in how infants would react when something was hidden within the drawbridges range of motion ■ Way to find out in the infants has any expectations about objects they could not directly perceive ■ Two scenarios ● In one, the drawbridge passed through the hidden object ● In the others, the drawbridge stopped at the obstructing box as expected ■ He found that the infants looked at the event where the screen passed through the box ■ Expected an object that could not be seen. Proved that object permanence ■

●

○

○

○

○

○

○

Elizabeth Spelke’s principles ■ Principle of cohesion- infants perceive even the gerrymandered object as a single individual because its surfaces are in contact ■ Principle of contact- only surfaces that are in contact can move together ■ Principle of solidity and continuity- every object moves on a single trajectory through space and time. It is impossible for objects to intersect paths. More than one object cannot be in one place at the same time Neural network modules of children’s physical learning (connectionist model) ■ Yuko munakata- object permanence in infants is practical and objects continue to exist even when they are not being directly perceived ■ This is an implicit knowledge stored in patterns in the neural connections that evolve as a function of experience ■ Patterns vary based on number of neurons firing, strength and number of connections between them and relation between individual firing rates Munakata et al. ■ Used recurrent networks ■ There were hidden units whose weights were modified ■ Feedback loops transmitted activation from the hidden units back to themselves ■ The network keeps a memory of the activation of the previous stage ■ Activation of the hidden units depend on the activation pattern of the input units and the activation pattern of the previous step Modeling the balance beam problem- how different forces interact ■ Children follow different rules at each stage to determine which side the balance beam will go down ● Stage 1- number of weights will determine ● Stage 2- distance from fulcrum will determine ● Stage 3- weight and distance, but cannot vary at the same time ● Stage 4- weight and distance used accurately Jay McClelland and E. Jenkins model- artificial neural network ■ Replicated the stages of the balanced beam model in artificial neural networks ■ Focused on variation in weights than in distance like children focused more on weight than distance ■ As training progressed the system learnt to consider distance in determining quantities like overall heaviness...