cogpsychE1

cognition

the mental act of knowing

goldstein definition of cognition

the mind creates and controls mental functions such as perception, attention, memory, emotions, language, deciding, thinking, and reasoning

the mind is a system that creates representations of the world so that we can act within it to achieve our goals

Marvin Chun definition of cognition / cog psych

cog psych seeks to understand how we come to know about the world and how we use that knowledge to make our way through it

the stroop task

involves seeing color words printed in a certain color. Task is to name the color that the word is printed in, not what the word says

measures our ability for cognitive control

reconfiguring information processing to achieve novel tasks

Pareidolia

the psychological phenomenon of seeing meaningful patterns, especially faces, in random or ambiguous stimuli, such as clouds, toast, or the moon

Muller-Lyer illusion

lines appear to be different lengths despite being the same length. 

explanation: interference of depth

Plato: is the mind a blank slate?

arriving at the right answer through socratic method after initially being wrong implies there was knowledge existing, person just needed to be guided to it

ancient greeks were NOT the first cog psychologists bc they were philosophers not scientists. they wanted to understand the nature of knowledge

Donders

first person to quantify cognition through reaction time experiments

donders reaction time experiments

stimulus —> mental response —> behavioral response

light flashes —> percieve light —> press button

• time from stimulus to behavioral response is the reaction time

left light flashes —> perceive left light AND decide which button to push —> press left button

• again time from stimulus to behav response is reaction time, but this time with an added cognition step

take difference of two reaction times to isolate cognitive processing time

wilhelm Wundt and structuralism

“the father of psychology”

structuralism: 

• cognition is a combination of basic elements (sensations)

• we find them through analytic introspection (observe own thoughts)

• what are the core elements/processes/sensations you are experiencing

downsides of structuralism: subjective and unreliable, elements not visible

Behaviorism

did not like introspection (structuralism) bc not reliable and elements are invisible

focus on:

• observable and measurable behaviors

stimulus-outcome associations: present stimuli, see what happens, measure outcome

believed there is no need to study the mind

John Watson (behaviorist)

little albert experiment:

unconditioned stimulus = loud sound

neutral stimulus = seeing bunny, no fear

conditioned stimulus = seeing bunny

conditioned response = fear upon seeing bunny, despite no loud sound

• EXPOSURE to neutral stimulus

• LEARNING: pair stimulus with loud sounds (unconditioned stim)

• CONDITIONING: be aversed by that object bc of loud sound (bunny scary now even tho no loud sound)

john watson believed that u can take a healthy baby and condition it to become anything u want using stimulus outcome associations

B.F Skinner (behaviorist)

operant conditioning:

• behavior followed by reward = increased behav

• behavior followed by punishment = decreased behavior

Edward Tolman — return of cogniton

experiments on rats, see how they respond to rewards. but NOT in a behaviorist way

Initial exposure:

• let rat run around to learn maze layout

learning phase:

• put food reward in one arm of maze every time, rat learns to run to get the food reward

test: 

• put rat in different arm of maze to start, which way will it go?

—> behaviorists would say operant conditioning would lead the rat to continue to travel in the same path it has been traveling in

BUT

rats actually knew where reward was and changed its plans towards this goal

—> rat was able to make decisions in new situation it had never seen before, so we must think about the mind and internal processes we cant observe

Edward Tolman —- Introducing Cognitive Maps

• internal model of the world maintained in our mind

• not directly observable, but is inferred

  • break with pure behaviorism!

the cognitive revolution - mind as an information processor

• began w introduction of the digital computer

• computer as analogue for the mind

  • input —> many computations/processing —> representation (storage)

broadbent model of mind as an information processor:

• inputs —> filter —> detector —> to memory

example of a box and arrow model of the mind

modal model of memory: sensory input —> sensory memory (unattended info is lost) —> short term mem (unrehearsed info is lost) ←retrieval→ long term memory (some info may be lost over time)

3 types of information processors

serial: one computation at a time

parallel: multiple processes at the same time, governed within the system

distributed: multiple, self-acting processes

serial information processor

can only perform one computation at a time

• time to do computation = number of steps it takes to do computation

burger restaurant example:

• one chef who can only make one burger at a time

• time to make 5 burgers = 5 * time to make 1 burger

parallel information processor

multiple processes at the same time, governed within the system

• time to do computation = time to do one computation

ramon y cajal staining neurons:

• showed that they werent serial, they can run many processes at once

burger example:

• burger assembly line of many chefs directed by head chef, each making one burger at a time

• time to make 5 burgers = time to make 1 burger

distributed information processor

multiple self acting processes

• independent parallel systems, nothing telling them what to do

burger example:

• 4 agents doing burgers, nobody telling them what to do. they are just doing steps until they finish their task. then they look at the environment and see if there is anythign more to do. if not, cease action

representation

an internal symbol that represents an external reality

something that stands for an experience

3 types:

• language

• network

• map

we use the representations that are driven by what is useful to us. humans are super social so faces are important —> specialized regions of brain to proces faces

to bats faces arent important —> sound more imporant

language representation

represents the information in words

network representation

in your memory, independent nodes that represent things that are related to each other through associations

Map representation

function that maps a dimension in the world to a dimension in the representation

f(dimension in world) = dimension in representation

ex scale maps weight into a number

Survival-Moving study —> evolution constrains cognition

selective pressures/problems shape the operating characteristics of our system

shown list of words and one of 3 encoding conditions:

survival condition:

• rate how relevant these words would be in a survival situation

moving situation:

• rate how useful these words would be to u if u were moving to a foreign land

pleasantness:

• rate how pleasant the words are

result —> survival encoding condition led to best memory performance

• we have more memory for survival, even tho we are more likely to move than to survive —> cognition shaped by evolution

mental rotation study —> shows how evolution constrains cognition

• see pairs of objects and figure out how to make the 2nd one eaqual to the 1st

• we solve these using rotation in dimensions that our brains have experienced

• causes problems when it requires a dimension we havent experienced

Levels of explanation— what is the right level of explanation at which to explain behavior?

information processing levels of explanation:

• computational

  • what is the problem to be solved

• algorithmic

  • what are the steps to solve it?

• implementational

  • how does the brain do it?

biological levels of explanation:

• whole behaving organism

• loves

• brain areas

• neurons

plasticity:

• can behavior control the brain? —> yes!

• ex: blind ppl learning to use echolocation

  • areas that are typically dedicated to vision now process auditory info

  • fMRI shows activation of visual cortex in response to sound

  • ** early blind had more light up than late life blind ppl

reductionism vs multi-level approach —— levels of explanation

reductionism:

• explain how bottom (neurons) explains top (behavior)

  • too simple, we know that it can go the other way around bc plasticity

multi level approach: —> better!

• interactions from top to bottom and bottom to top

the neuron doctrine

the idea that individual discrete cells transmit signals in the nervous system, rather than a continuous nerve net

hierarchy of brain organization

brain —>

cerebellum, cortex, midbrain, brainstem

—> lobes of cortex

—> areas

—> neurons

neuron

cell consisting of celll body (maintains it), dendrites (appendages to cell body that recieve information), and axons (sends info from the neuron)

brain previously thought to be big neural net, but ramon y cajan neuron staining showed that its individual neurons —> neuron doctrine

RECEPTOR NEURONS:

• stimulus → receptor → axon

• have no dendrites

• AP:

  • resting -70

  • impulse +40 peak

  • goes back to resting pot

neuronal action potentials

• dont increase in intensity based on stimulus intensity. it is an all or nothing response 

  • threshold -55 mV

  • resting -70

  • stimulus

  • threshold

  • depolarizatoin —> peal +40

  • repolarization

  • refractory period and drops below -70

  • comes back to resting state

how does neruonal AP occur?

• there is synapse wher receiving cell body/dendrites and axon meet

• AP arrives, vesicles in axon containing neurotransmitter released into synaptic cleft of synapse, neurotransmitters bind to dendrites of rec cell, changing electrical charge of rec cell

Inhibitory synapse

decreases the charge of the post-synaptic cell, reduces likelihood that will overcome threshold for AP

GABA and GABA receptor

• Cl- rushes in —> negative charge of cell

excitatory synapse

increases the charge of post-synaptic cell, increases likelihood that will overcome AP threshold

glutamate and NMDA and AMPA receptors

• lets in Na+ and Ca2+

brain area

a patch of cortex that has one particular function

visual cortex and its areas

• visual cortex is contralateral = right visual field is processed on left side of brain and vice versa

• early areas of the VC do simple things —— Hubel and Wiesel cat V1 studies

Hubel and Wiesel Cat studies about V1

• record from specific neuron in V1, show it lines with differnt orientations andlocations

neuron in cat V1 encodes edges that 

1. have a particular orientation

2. appear in a particular location of the visual field

—> early VC is very specific and encodes things in a topographical representation

aka a RETINOTOPIC MAP

f(closeness to preferred location, closeness to preferred angle) = firing rate of neuron increases

lobes

occipital (visual)

temporal (auditory, objects, memory (hippocampus is behind it)

parietal (attention, orientation)

frontal (thinking, reasoning, decision making, etc)

grey vs white matter

grey matter = cell bodies, dendrites, axons, axon terminals 

• outside parts of brain

white matter:

• long--range axons, myelin

gyri and sulci

gyri = protrusions

sulci = valleys

create wrinkles in brain cortex

gives more surface area for cortex AND relative distance between areas is closer, so faster communication

major regions of the brain

cerebellum = we dont give a shi ab this

brain stem

cerebral cortex

using isotopic fractionating = turn brain into soup (homog mix of cells bodies) , stain, then count nuclei —> tells us how many neurons in a region

• V1 = 280 million

• cortex = 16 billion

• cerebellum = 69 billion

• whole brain = 86 billion

cerebellum has mad neurons, yet we ignore it in cog psych i guesss?

neural representations

how neurons represent the world

3 ways of coding:

1. specificity coding

2. population coding

3. sparse coding

specificity coding (neural representations)

one to one representation (one neuron for each thing ever)

cons: not enough neurons for this, cant think ab similarities/associations

population coding (neural representataions)

all neurons involved in the representation, but it is a distrbution

cons: number of possible representations is limited

sparse coding (neural representations)

set of neurons that represent certain things, others are quiet. a distribtuion, but not every single neuron is involved in this distribution

how are recordings taken directly from the cortex

electrocorticography

array/grid of electrodes on the cortex

can try to find neurons that respond to specific things

WARNING: not every cog fn or represtnation can be located to one particular brain area

most of the time, cognition is driven by distributed or parallel process across entire brain

distributed neural networks

looking at resting state data, we can see which regions covary with each other

we can also look at this when ppl are doing stuff

distributed neural functions

• somato-motor, visual, saleince, executive control, dorsal attention

• default mode network

can we infer causality?

Hubel and Wiesel experiment showed that area V1 represents edges and oreintations: TRUE

hubel and wiesel experiment showed that area V1 causes perception ofedges and orientations?: FALSE

• just bc one region correlates w that, doesnt mean that it caused its perception

• there might b a downstream region that causes perception

motion perception case study to make causal inferences

• area MT of VC, selective to motion in the contralateral field

• isolate area MT by having person look at field of moving dots w different levels of coherence

first step: find neuron that cares ab one particular direction

• show 100% coherent motion

• find neuron that responds strongly to one direction of 100% coherent motion

• record neuron while varying direction and coherence —> rules out that it responds strongly to everything

results:

• neuron spikes the same regardless of coherence when motion is in opposite direction

• neuron spikes more with increasing coherence for preferred direcional orientation

so: is it involved in perception tho? —> stimulate this cell while monkey responds to the movement in the cells preferred direction

• bias towards that direction

• with stimulation, monkey is more likely to report motion in the preferred direction, even if there isnt motion in that direction

Haemodynamic response (neuroimaging)

when neurons fire, they require energy (glucose/oxygen) and consume it

record oxygen in blood

• stimulus, slow esponse, initial dip in oxygen (using it up)—> then peak (brain oversupplies  that part of brain w oxygen) —> post stimulus undershoot

very slow and can be measured easily using PET or fMRI

positron emission tomography (PET)

uses radioactive tracer to measure haemodynamic response — measures blood flow/metabolism

spatial resolution = 1 cm

temporal  = 0.5 min

• radioactive tracer on oxygen

• positron hits electron —> photons

• ohotons travel in opposite directions and hit detector

• small differnce in timing reveals location

first neuroimaging of cognition in humans!

functional magnetic resonance imaging (fMRI)

based on magnetic properties of blood oxygen (no radioactive tracer) to measure blood oxygen level (haemodynamic response)

spatial resolution = 2 mm

temporal = 2 sec

better than PET spatially and temporally, but MAD expensive

electroencephalography (EEG)

does NOT measure haemodynamic response —- measures electrical activity of brain directly

based on electrical properteis of the brain

measures chagne in voltaeg across groups of neurons

wayyyyyyyyyyyy cheaper than fMRI

spatial resolution = 8 cm (not great…)

temporal solution = 1 ms (MAD fast)

Flanker task —- measured using EEG

left or right button press to indicate which way middle arrow points

• neutral trials = flankers dont interfere

• congruent = flankers match

• incongruent = flankers dont match

  • more errors

error related negativity tracks cognitive control:

• garph showing potential vs time after stim onset

• congruent and incongruent plots differ at 200-300 ms

• at this pt, incongruent trials have a more negative potential, congruent have more positive potential

  • this reflects how much conflict the brain is experiencing

• more incongruent = more conflict = more errors = more negative potential

neural circuits

groups of interconnected neurons

if kittens are raised in an evnironment that contains only verticals, you would predict that most of the neurons in their VC would respnd best to the bisual presentation of a 

chain link fence

solid wall

picket fence

brick wall

naturally occurring lesions

brain lesiosn due to disease and injury

causes:

• trauma

• stroke

• bleeding

• tumor

• neural disease

scotoma in visual cortex (lesions)

lesions in occipital lobe, specifically visual cortex

if damage is in left occipital lobe (calcarine sulcus) —> visual abnormality in the right visual field

how can we use lesions to understand cognition?

they can help us identify neural substrate of a particular function

but we need to be very careful 

must use double dissociations to distinguish two different processes.

single dissociation

one computational process A

behavior X that is influenced by A and behavior Y is not influenced by A

but cannot conclude that A does not affect Y at all. if keep on damaging A, will Y eventually be influenced too?

issue with single dissociation:

• differences in demand are an explanation

• perhaps one behavior is just the first to leave

double dissociation

two independent comp processes A and B

two beahviors X and Y that dpeend selectively on A and B

some lesions should only affect X and not Y

some lesions should only affect Y and not X

by showing that it is not differnces in demand (one process doesnt always leave first), show that must be two independent processes

double dissociation in language

Broca’s Aphasia (expressive aphasia)

• preserved language comprehension, but short and grammatically impoverished productions

• (know whats going on, cant say it)

Wernicke’s Aphasia (fluent aphasia):

• poor comprehension, fluent speech with relatively little content

• (dont understand language, but can easily speak a bunch of words, no meaning to the sentences tho)

most brain lesions are not well localized —> CTE study

deceased former football players

88% with neurological condition had CTE

limitations of natrurally occurring lesion studies

• subject to the whims of nature

• correlational

• not always well localized

observational studies —> correlation

naturally occurring lesions

electrical recording (electrocorticography)

neuroimaging (PET, fMRI, EEG)

interventional studies —> allow us to infer causation

surgical lesions

transcranial magnetic stimulation

transcranial direct current stimulation

transcranial magnetic stimulation (TMS)

repeated pulses of electrical stimulation for a period of time (15 min)

disrupts neural circuits for a longer period of time after that (~30 min)

self control study: apply TMS to prefrontal cortex

• small reward now or wait and large reward later

• when PFC is “lesioned” (TMSed) there is reduced preference for the larger-later option —→ more impulsive decisions made

advantages of TMS

• temporary “lesions”

• directly affects action potentials

• excellent temporal resolution

• not great spatial resolution

• can establish causality

disadvantages of TMS:

• only cortical surface can be affected

• variability in brains btwn participants hurts spatial resolution

neural map examples

motor and somatosensory cortex are topographical maps of the body sensor and motor

• closer in space, closer their neural representations are

retinotopic map:

• visual cortex as a map, represents the retina in space

• stained cat neurons while forced to look at moving stimulus map pattern

• pts close in visual field are close to each other on VC

auditory map:

• auditory cortex sits in small part of brain behind temporal lobe

• tonotopy

  • f(sound frequency) = space on the brain

how are neural maps formed? innate vs experience

study of musicians and cab drivers

musicians:

• professional musicians have more grey matter esp in primary motor cortex

• non-musicians have less

cab drivers:

• plasticity in the hippocampus

• cab driver exam must be able to mentally simulate path from one location to another

• found relationship between how long one has been a taxi driver and how big their hippocampus is

are these studies causational? NO

• “the experiments show that expreience causes structural changes in the brain” = FALSE

• no manipulation done —> causality cannot be established

experimental manipulation to show cat has orientation sensitive V1 neurons

• cat raised in environment where only vertical or only horizontal lines

• vertically reared

  • doesnt seem to have neurons that respond to horizontal lines

• horizontally reared

  • doesnt seem to have neurons which respond to vertiacal lines

so: visual cortex isnt intrinsically set up to do this, interacts with environment

^^ double dissociation experiment

shows experience dependent plasticity = our brains refelct our experience / our brains ARE our experience

perception

Perception: experiences resulting from stimulation of the senses

• Vision

• Hearing and touch

• Taste

• Smell

• Balance

• Temperature

• Proprioception

• Pain

• Interoception

Vision main takeaways

bottom up vs top down

object perception (bayesian inference)

gestalt grouping

why are faces special?

bottom-up visual processing

data driven

eye-to-brain

integrates basic features in the environment

top-down visual processing

based on prior experiences

uses prediction

input is interpreted using context

ex: THE TAE vs CHT CAT

interpret middle letter differently based on context of surrounding letters and overall word

evidence from neuroscience for top-down and bottom-up

though there are many connections from low to high in the visual perception wiring diagram, the great majority of these pathays involve reciprocal connections btwn areas

if there are projections from low to high area, there are usually just as many feeding from high to low

Bayesian inference is the combination of

(helmholtz) unconscious inference

and

contextual regularities

helmholtz unconscious inference

perception is a form of inference —> we see the object that is most likely causing the stimulus we observe (the likely principle)

— relates to the task of determining the object responsible for a particular image on one’s retina: the inverse projection problem

• REMEMBER: are children actually small or just really far away SNL sketch

contextual regularities — AND Oliva & Torralba “multiple personalities of a blob”

physical regularities:

• there are more vertical and horizontal edges than oblique (diagonal) edges

• Light usually comes from above

• Biological things are often roughly symmetric

• Movement starts and stops smoothly

• Objects don’t blink in and out of existence

semantic regularities

• oasters appear more often in kitchens than in bedrooms

• Pants are more likely to be worn on legs than on the head

• Glass is more likely a solid than liquid

• Multiple personalities of a blob: we use the context of the image to shape our perception of the ambiguous blob

Bayesian inference

P(H|E) = P(E|H) * P(H) / P(E)

• P(H|E) —- posterior probability of hypothesis given evidence

  • bayesian inference compares these

• P(E|H) = likelihood of evidence given hypothesis

• P(H) = prior probability (before evidence)

• P(E) = marginal probability, prob of evidence (usually can be discarded)

choose hypothesis which has highest posterior probability

on beach and see shape? is it driftwood or an umbrella?

prior is dictated by regularities

comparisons of posterior probabilities is unconscious inference

from far away, priors determine what is percieved

• once closer, visual info outweighs the priors

Gestalt Grouping

Gestalt = a shape, configuration, or structure which as an object of perception forms a specific whole or unity, incapable of expression simply in terms of its parts

Gestalt principles are TOP-DOWN processing

quick and automatic

reflects regularities in the environment and the likelihood principle

proximity (gestalt)

objects tend to be grouped together if they are close to each other

closure (gestalt)

tendency to complete figures or forms as a whole, even if information is sparse

good continuation (gestalt)

tendency to parse figures with smooth, continuous lines, rather than abrupt or sharp angles

ex: this x is two straight lines, not two V shapes

good figure (gestalt)

patterns are interpreted in such a way that the resulting structure is as simple as possible

similarity (gestalt)

tendency to group objects together if they are similar

familiarity (gestalt)

tendency to group objects that are familiar or meaningful

are faces special? is there an innate hardwired face module or is it experience dependent plasticity?

arguments for innate hardwired face module

• hollow face illusion (bayesian inference, more likely to see outward facing face than hollow face)

• thatcher effect (upside down face with right side up eyes, we dont notice anything wrong)

• prosopagnosia (some ppl born w inability to recognize/detect/distinguish ffamiliar faces bc of damage to FFA / fusiform gyrus)

• stimulating the FFA using implanted electrodes (when stimulating FFA, induces action potentials, causes patient to temporarily have difficulty recognizing face)

• face selectivity in infants fMRI experiment (are we born with FFA? infants and parents in fMRI shown faces and places. in adult see specialized areas FFA and PPA. in infant, we also see that)

arguments against innate hardwired (FOR experience dependent plasticity)

• greebles experiment (trained to become expert at analyzing made up creatures. after training, increased FFA response to Greebles creatures)

• face deprivation study (raise monkeys never seeing researcher face or other monkey faces before. show them face in fMRI, no face area activation for ones that have never seen a face. Hand selective region responds tho.

  • for free-viewing behavior (images w both hands and feet showing), deprived only looked at hands. non-deprived looked all over at faces and hands

speech segmentation is

organizing the sounds of speech into individual words

TOP-DOWN processing that influences simple perception. need knowledge of ur language to know when words start and stop

maps for touch AND the two point discrimination task

somatotopic map

homonculus

two point discrimination task

• different sizes of objects with arched edge and flat edge. start large. touch thing to body area and ask person if it was the arched or flat edge. then go smaller and see what is smallest threshold they can detect

• lower the threshold, the finer the representation of that body area in the somatosensory cortex (ex, fingertips, tongue, lip, are very sensitive)

object perception in touch (quarter trick) — bayesian inference

• two far away regions on two different fingers are feeling coin, so you are more likely to believe that there must be two coins rather than one coin bc two coins are more likely to make u feel that sensation in those two areas

hearing anatomy

• air vibrations sound —> eardrum —> mechanical wave makes bones move —> fluid wave into cochlea —> hair cells vibrate for specific frequencies

  • high frequencies travel less

  • low tones travel more —further into cochlea

• —> cells in brainstem —> … —> auditory cortex “heschl’s gyrus”

auditory cortex (as a map)

tonotopic map

use fMRI and play high and low tones and see which areas light up

map bc we see certain colors in cetain areas, pitch to spatial representation

plasticity in the auditory map

play 2 tones, which is more familiar

• most ppl choose tone 2. why? it is the bleep noise tone on TV!

then play og tone in video 1 and either a sharp or flat tone in video 2

• ask which has the correct tone now

• the further away (in semitones) the wrong tone was, the more likely they got it correct

this bleep noise is not a commonly occurring frequency… we have been trained to know it (plasticity)

gestalt grouping in hearing

• good continuation: gliding tone with small gaps in between. then filled with white noise. we fill in the gaps to hear smooth continuous gliding tone

• proximity: hear a progression of notes that jumps from high and low a lot, instead of hearing jumping, we group high parts as the melody and low parts as the baseline

crossmodal perception

combining senses / interaction btwn senses

examples of crossmodal perception

• McGurk effect

• Rabbit Illusion

• Rubber Hand illusion

McGurk Effect (crossmodal perception)

vision modulates hearing

interaction between top-down and bottom-up processing

• bottom up thinks ba

• vision comes in and makes u think fa

rubber hand illusion (crossmodal perception)

researcher brushes rubber and real (concealed) hand at same time, but participant only sees the fake hand

over time, participants report thtat they can feel touch in the rubber hand. even when no stim of real hand anymore

mental chronometry

determining the amount of time needed to carry out various cognitive tasks

skeptics about mental imagery think it is an

epiphenomenon