Early vs Late Selection Theories

Question 1

INTRO

Answer 1

• Attention
• Bottleneck theories
• Early vs Late theories
• Dichotic listening tasks

Question 2

What is attention

Answer 2

The taking possession by mind, in a clear & vivid form

Implies withdrawal from some things in order to effectively deal with others

Need attention for selection (limited elaboration capacity) & action

Question 3

What are bottleneck theories

Answer 3

WW2- needed to understand how pilots act on multiple signals in different channels

Bottleneck- filter that blocks irrelevant info

Different theories to explain how info passing through bottleneck is selected

Question 4

Early vs late selection theories

Answer 4

Early- info filtered at early levels of processing

Late- info filtered at late levels of processing

Question 5

What are dichotic listening tasks

Answer 5

Pp asked to attend to 1 of 2 messages simultaneously presented

Shadowing- asked to repeat 1 of 2 messages

Virtually no memory of unattended message

Only physical attributes detected (no semantic)

Question 6

EARLY SELECTION THEORIES

Answer 6

• Broadbent’s
• Evaluation
• Triesman’s
• Evaluation
• Early selection evaluation

Question 7

Broadbent’s filter theory (1958)

Answer 7

1) Input (both ears)
2) Sensory register (physical attributes)
3) Filter (blocks unattended info, detects semantic info in attended)
4) STM (processes info for response)

Question 8

Broadbent’s filter theory (1958)- stores & channels

Answer 8

Bottleneck is at short-term store stage

1) Short-term store (S)- retains all incoming physical info
2) P system (limited capacity channel)- items selected by filter passed to here, operates in series
3) Can go back to S system for rehearsal or LTM

Question 9

Problems with Broadbent’s theory (1958)

Answer 9

• Moray (1958)- pp reported own name when presented in unattended ear- suggests some unattended stimuli are processed
• Gray & Wedderburn (1960)- pp reported unattended info when it’s meaning matched attended info- suggests stimuli not selected based in physical characteristics but according to meaning, semantic properties processed too
• Triesman (1960)- pp reported unattended info when it was logically linked to attended info- suggests during shadowing tasks, info of unattended ear still processed at semantic level & pp can recognise there’s a relationship between unattended & attended ear info

Question 10

Triesman’s Attenuation model (1960)

Answer 10

Bottleneck location more flexible

Filter not all or nothing block as pp can sometimes recall unattended channel words

Filter reduces or attenuates strength of unattended channel

Stimuli analysed through hierarchy- first based on physical cues then grammatical structure & meaning

If insufficient processing capacity, stimulus not processed at top of hierarchy

Expectations important (top-down)

Question 11

Problems with Triesman’s theory

Answer 11

• Nature of attenuation process has never been precisely specified
• Doesn’t explain semantic analysis works
• Incomplete theory
• Corteen & Dunn (1973)- pp showed GSR when words conditioned with electric shock or associated words presented in channel
• Suggests unattended channel’s words processed at semantic level, semantic generalisation & selective process in attention comes after meaning of words accessed

Question 12

Early selection evaluation

Answer 12

• New pp unable to shadow successfully- due to unfamiliarity with task rather than inability of attentional system
• Researchers can never be sure that pp haven’t switched attention to unattended channel
• Memory for unattended channel may depend on familiarity, importance or similarity to the attended channel
• There’s implicit memory for unattended channel even when no explicit memory

Question 13

LATE SELECTION THEORIES

Answer 13

• Deutsch’s late selection theory
• Evaluation
• Norman’s pertinence model
• Evaluation

Question 14

Deutsch’s late selection theory (1963)

Answer 14

All stimuli fully analysed

Most important/relevant determine response

Bottleneck placed nearer end of processing system

Question 15

Deutsch’s Evaluation

Answer 15

Triesman & Riley (1969)- targets presented in attended channel detected more than unattended channel targets- supports Triesman’s

Question 16

Norman’s pertinence model (1969)

Answer 16

All sensory inputs analysed (automatically & unconsciously before conscious)

All sensory inputs have access to semantic memory

Pertinence values assigned to inputs according to ongoing cognition activities

Selection comes after semantic memory & pertinence- based on summation of activation in semantic memory

Most messages rejected leaving most pertinent to be deeply processed

Question 17

Lewis (1970)

Answer 17

• Time taken to shadow attended channel greater where semantically related words presented in unattended channel
• Even if pp did not remember unattended channel words

Question 18

Norman’s Evaluation

Answer 18

• Even if pertinence controlled for, pp more likely to detect info in attended channel & less likely in unattended channel, pp find it difficult to recall semantic content
• Model is uneconomical- all info analysed for meaning even if not relevant to task- would require large processing capacity

Question 19

COMPARISON

Answer 19

• Comparison between Norman’s & Triesman’s

- Early vs Late Evaluation

Question 20

Norman’s vs Triesman’s models

Answer 20

Similar- some inputs more likely to be selected due to interaction between priming & input

Different- in Triesman’s, attenuator at beginning of system before LTM, in Norman’s, selection after parallel access to semantic memory

Question 21

Early vs Late Evaluation

Answer 21

• Kahneman & Treisman (1984)- different paradigms may require different attentional mechanisms
• Filtering task- pp have to select right stimuli, accuracy, accuracy of response measured, large response set, high memory load
• Early selection- limited processing of unattended stimuli
• Selective task- pp has to be fast as possible, reaction time measured, small response set, low memory load
• Late selection- speeds response to expected target
• 2 experimental paradigms are v different, demand made in info-processing system, unlikely that same type of processing required in both