cog science exam 1

algorithmic/representational level

-how is the input represented -how is the output represented -what computational steps transform input/output

Shepard and Metzler (1971)

Mental rotation task Tasks that called for greater rotations took longer; Degree of rotation and reaction time

early selection model

Model of attention that explains selective attention by early filtering out of the unattended message. In Broadbent's early selection model, the filtering step occurs before the message is analyzed to determine its meaning.

SNARC experiment

People are faster if the correct answer is:On the right (for larger numbers)On the left (for smaller numbers)

cocktail party problem

The problem of attending to a single auditory stream in the presence of competing streams (with different acoustic and spatial properties)—for instance, attending to one person's voice in a noisy room of other voices

computation level algorithmic level implementational level

What are Marrs 3 levels of analysis in cognitive science

recognizing faces, math

What is an example if implementing marrs 3 levels?

how to understand the world in which it is situation and the interaction between the mind and world

What is the motivation for having the 3 levels?

computation

a finite number of deterministic steps to output the value of a function, given its arguments

representations of the mind

a symbol or structure that encodes information and can be manipulated with computational operations

hierarchical action planning

accomplish a task in structured chunks cognitive scientists ask: what specific aspects of a cog map are relevant, what grammar rules.. etc)

late selection model

all sources are processed into higher level representations-- source that you respond to is selected after further processing

Inputs

arguments

retina, brain, biological cells

basic understanding of the anatomy of the visual system

Rats learned to find the reward in a maze more quickly if they had previous experience with the maze, even if that previous experience did not provide a rewarding stimulus. This presents a challenge to behaviorist accounts of learning.

behaviorist vs cognitivist psychology

Cognitive science includes psychology and neuroscience, but it is not the same as psychology or neuroscience. Cognitive science asks the why/how questions.

cog sci vs neuro science & psychology

levels of representation

cognitive processing can transform different signals to the same higher level of representation of the word; bottom up processing/ top down processing

Bottom-up processing

decompose the sound sources based on sensory features (example: high pitch, quiet, loud, sound change)

top-down processing

high level features (example: word meaning) help you listen and separate sources.

• Identify an object (computational level) • Compute edges (part of algorithmic level) • Brain cells are connected together in a way that carries out an algorithm. (implementational level)

how does the concept of "levels of representation" apply to vision and how can the visual system compute?

latent learning

learning that takes place in the absence of reinforcement (reward)

place learning

rats learn to go to a specific place rather than a sequence of motor responses. use a "cognitive map" as a representation

Self-Driving Cars • Identify type and location of objects (computational task)• Shoot out lasers; Time lasers to calculate depth; algorithm compares timing of laser pulses (algorithmic/representational level) • Made of metal, lasers, etc. (implementation) vs. Humans • Identify type and location of objects (computational task) • No lasers; Use retinal disparity to figure out depth; algorithm compares difference between what each eye sees (algorithmic/representational level) • Made of biological tissues (implementation)

self-driving cars "see" vs how human do using Marrs 3 levels

Algorithm

the finite set of deterministic steps

outputs

values

figuring out which algorithm and representation humans use and how

what are basic problems presented to the visual system that have to be solved by the mind?

top-down: concept dog helps you listen for related words (wag, tail, friend) bottom-down: (sound frequencies ; bass or treble)

what are examples of top-down vs. bottom up processing

- artificial intelligence - psychology - neuroscience -philosophy - computer science involved because

what are several disciplines involved in cog sci and why are the involved?

Cognitive scientists do experiments to figure out how people represent things in the world. How thinking things think. (perceptions, think, behave)

what does cog science seek to explain

computation level

what does the system need to accomplish/what is the goal

SNARC and shepard and Metzler

what experiments can be used to investigate the form of mental representations

multiplication, square root function

what is an example of a function

certain words carry more information than others. "she takes her coffee with pepper and sugar" --> emphasize on pepper so it takes less processing power

what is an example of information theory in speech

working memory capacity is for items, not total information and in order to "fit more" info you need to recode the stimuli making it more efficient (chunking) and helps for long term memory.

what is miller insights that link information theory with cognitive processing

thinking can best be understood in terms of representational structures in the mind and computational procedures that operate on those structures

what is the central hypothesis of cognitive science?

thinking as doing and explain thoughts

what is the purpose of computational models in cognitive science

implementational level

what physical stuff does this computational task (neurons, transistors..)

we don't fully understand or know how the mind separates the input signals into different sources (auditory stream segregation)

whats the problem of source separation

Lavie's perceptual load theory

whether early or late section models are correct:___________ as a necessary condition for selective attention. under low load conditions, all of the light patterns on the screen get processed into high level representations so they are processed into letters