bcs

process flow

distal→proximal→percept→recognition

dorsal stream

• where

• spatial location, movement

ventral stream

• what

• object identification

distal stimulus

real object in the world

proximal stimulus

object processed through visual cortex

percept

object interpreted through temporal cortex

gestalt principles

• holistic processing: process the input in its entirety

• principle of simplicity

proximity

group nearby items together

similarity

group similar items together

continuation

prefer smooth, continuous forms

closure

fill in gaps to perceive a complete object

common fate

group items moving in the same direction

feature analysis

• single object

• Break objects into distinctive features or geons (3D components).

visual search

• search latency positively correlated with the similarity between the target and the distractors

search latency

time needed to find the target

event recognition

• analyze features in a scene

• scene→attending to features→categorization of the event

categorical perception

• Perception is shaped by categories we know.

• category boundaries can be shifted through new experiences

prototype

match the input with a pre-stored “prototype” (representative of the category)

exemplar

match the input with each stored instance in memory

posner and keele

• training

  • present one distorted dot pattern at a time (Each pattern is derived from one of the four original prototypes) (prototype is not shown to participants in this phase)

  • classification task

    • seeing a distorted dot pattern (target stimulus)

    • compare it to 4 options

    • judge if the target stimulus belongs to any of those 4 categories

• testing

  • three types of target stimuli: new distortions (not seen before), old distortions, original prototype (not seen before)

    • classification task: same as in training

key points for p and k study

• same task in both training and testing

• feedback given during training

• no feedback during testing

• prototype formed during training

implications of p and k study

• formation or reconstruction of a prototype by seeing/hearing many variants (distorted dot patterns, faces or words)

• normalization across variants to reconstruct the prototype

exemplar matching

• a large number of stored exemplars

• forming a prototype by generalizing across exemplars

• each heard/seen input stored as a memory trace

issues with exemplar model

• memory capacity: how much can be stored?

• novel objects: new stimulus never encountered before

• determination of recognition threshold: very detailed or common features?

prototype matching

• compare the input with the category prototype

• efficiency of processing

low categorization accuracy

distinction between a and b is vague

high categorization accuracy

distinction between a and b is clear

ABX task

you see three filled circles. judge if the third one is similar to the first or the second one

ABX task to prototypes and exemplars

• comparison between exemplars within the category is not always easy

  • still possible to detect

• comparison between exemplars near the category boundary also not so easy

• comparison between prototypes is easier

• easier to distinguish between different categories than within the same one

bottom-up

• driven by sensory input 

• prototype

• exemplar

• feature analysis

top-down

• prior familiarity about the input

• expectations

• context effect

  • word superiority effect

    • recognizing letters is faster when they’re in a real word rather than isolated or in a random string

  • contextual effect

selective attention

• focus on a very limited events/ objects/ tasks

• for efficient processing

filter theory

• irrelevant information is filtered out

• limited capacity to process information

• overflow info is filtered through the bottleneck

• all unattended messages filtered out

dichotic listening task

• participants repeat what they heard (from either ear)- “shadowing”

• most people can repeat the attended message from one ear with very few errors

• if the unattended message sounds weight, some people can notice the difference

evidence against filter theory

• cocktail party affect

  • when you hear a fire alarm in a loud concert while talking to your friends, you still notice the alarm

  • pashler

    • only 33% heard their name is not informed about the possibility of hearing their name before the experiment

• switch ears

  • message played to one (Attended) ear switched to another (unattended) ear

  • treisman

    • people repeated a few words from the nattended ear right after switch

    • people not aware of the switch and their own repetition of the words from the unattended ear

    • attention can also be influenced by the nature of the message itself- the meaning of the message

    • another evidence for top-down process

when do we notice something from the unattended ear in dichotic listening task

• in english but with an “oddball” (reversed speech)

• in english and with our own name

• continues the message from the attended ear (switch ear(

when do we ignore something from the unattended ear in dichotic listening task

• in a foreign language the subject doesn’t know

attenuation theory

• irrelevant info is tuned down

• treisman

  • unattended message not completely blocked or filtered out

  • the volume is tuned down or attenuated on the unattended ear

    • less resources allocated to that ear

    • top-down influence on attention

      • words of subjective importance

      • words signaling danger 

      • require little mental effort to recognize

filter theory on DLT

• only process and hear the attended messages

• not a valid account given the findings from DLT task

attenuation theory on DLT

• unattended message tuned down

• higher attention level for familiar/ important content

spotlight approach

• perceive everything but actively cast a spotlight on the target; things on the edge of the spotlight can still be processed

selective attention on spotlight approach

• size of the spotlight varies with the size of the objects to focus on

  • depending on the processing demands of each object/ event

• control where to direct our attention

schema approach

only taking in what is needed and leaving everything else untouched

filter + attenuation

everything perceived gets processed and then irrelevant info gets filtered or tuned out

spotlight + schema

actively select the relevant info

banich

• two types of tones: long vs short. both can be heard in either left or right ear

• target: long tone presented

EEG for selective attention

• target is attended in left ear, N1 is more negative in left than right

• target attended in right ear, N1 is more negative in right than left

arousal level for selective attention

• kahneman

• attention: allocation of resources

• attention modulated by arousal level

  • task difficulty

  • level of interest

divided attention

• dual task: allport et. al

  • shadow (repeating) and memory task

  • lots of interference in words heard compared to words seen and pictures

• executive control decides how to split our cognitive resources

automatic processing

• three criteria

  • processing occurs without intention

  • unaware of the process of noticing the target

  • not interfering with other mental activity

• requires little mental effort to process

• bottom up

controlled processing

• requires conscious effort and attention shifting

• top down

testing automatic processing

• consistent mapping

  • target and distractors are different kinds of stimuli. target never appears as distractor

  • automatic

  • display time matters

• varied mapping

  • target and distractors are some kind of stimuli. target in one trial may appear as distractor in next

  • controlled

  • all variables matter

    • longer display time→easier

    • larger frame set size→harder

    • larger memory set→harder

• variables

  • varied vs consistent mapping

  • frame size

  • frame time

  • memory set size

attention shift task 

• switch between two features of a stimulus

  • control what you need to attend to at the moment

• measure cost of switch

• set up

  • trial 1→ respond based on feature 1

  • trial 2→ respond based on feature 2 = perseveration

  • trial 2 minus trial 1 = slower time, harder time shifting attention

  • tests executive control

neural underpinnings in attentional control

• zimmermann et al

  • finger sequences

  • focus internally (imagining finger movement) or externally (button press)

• results

  • both activated primary somatosensory cortex

  • switching from internal to external activated premotor cortex

  • switching from external to internal activated a stronger somatosensory cortex

• damage in right parietal lobe

  • hemineglect: unawareness of objects in visual field contralateral to lesion site

memory processes

• encoding: process that converts input into a memory trace

• storage: holding that trace in memory

• retrieval: bringing it back when needed

iconic memory

• sperling 

• whole-report task: participants see a grid of letters for a second and recall

• partial-report task: cue tone signaled which row to recall after desplay

  • recall was better if cue was immediate

  • delayed recall= less performance

limits of iconic memory

• cue by category= recall drops

  • categorization happens after sensory storage

• cue by sound= modality specific (each sense has its own corresponding sensory memory)

• lasts >1 second

severed corpus callosum

• stimulus in left visual field→ yes recognition, no processed information

• stimulus in right visual field→ yes recognition, yes processed information

broca’s area

• originally for sequence learning; evolved for grammar and speech

• Artificial grammar learning

  • Broca’s area is activated more for ungrammatical sequences

  • helps process structured patterns

Reber AGL

• two sets of sequences

  • grammatical and random

• accuracy is higher for grammatical

petersson

• exposure to many grammatical sequences many times

• after each sequence, type it out

• 5 days of training

• testing phase: sequence classification task

priming paradigm

• not seeing prime if presented <60ms

• masked prime is unconsciously processed but still can facilitate object naming

• implications: icons erased but still can be processed without being consciously aware of i