5 - Attention I & II: Origins & Early-Late Selection Debate

Question 1

How is selective attention limited? What is the ‘Cocktail Party Problem’ (Cherry, 1953)?

Answer 1

1. Limited in the number of stimuli we can process.
2. Attend to one at the expense of others.
3. People are limited capacity systems: don’t treat all stimuli equally.

Cocktail Party Problem:

• How are we able to pick out relevant conversation in crowded environment?
• “Picking out” conversation from the b/ground noise.
  
  • Transduction of noise to neural signal is selective.
• To study, we look at unattended messages

Question 2

From Cherry’s Dichotic Listening and Shadowing paradigm - what characteristics of the unattended message do people perceive preattentively? How did his experiment work?

Answer 2

• Pts recall only superficial physical features such as voice vs. non-voice, male vs. male speaker.

Design:

• Pts wore headphones with M1 (attended channel) and M2 (unattended channel)
• Manipulated characteristics of M2, e.g. switch to German, to female, to pure tone, reversed speech.
• Pts asked to report content or features of M2 (unattended message).

Conclusion:

Sensory, physical features are processed preattentively, but meaning requires focal attention.

Question 3

How did Cherry manipulate his experiment to show that source localisation is the key mechanism that solves the Cocktail Party Problem?

Answer 3

• Binaural presentation of M1 & M2: both ears receive both messages, same voice, different content.
• Pts took 10/20 presentations to understand contents of M1.
• Source localisation in space is an important cue for comprehending sound - phase differences key in arrival times at ear.

Question 4

What is a key criticism of Cherry?

Answer 4

His experiments confound memory and attention by relying on participants recall.

Question 5

Describe the structure of Broadbent’s (1958) Split- Filter Theory of Attention - components include short term store (STS), selective filter, limited capacity channel.

Answer 5

• Attention acts as a filter to select stimuli for further processing.
• Meaning is extracted in limited capacity channel.
• Filter precedes channel - protects from overload.
• All stimuli briefly stored in STS (WM?)
• Raw acoustic trace decays quickly if not selected.

Question 6

According to Broadbent, why does the recall of digits from his split-span task take longer when participants are asked to recall them in temporal order? Note: presentation was a dichotic digit stream, i.e. Left ear 1, Right ear 8, Left ear 7, Right ear 5, etc.

Answer 6

• Short Term Store is organised in two sections - one for each ear.
• To extract meaning from temporal recall, the attention filter needs to switch 5 times from left to right each to extract the numbers.
• Ear-by-ear recall needs 1 filter switch.
• Switches take time, STS trace delays.

Question 7

How did the ‘Dear Aunt Jane’ Split-span experiment call into question the Filter Theory?

Answer 7

• Split-span experiment with meaningful material.
• The preferred recall follows semantic context, not presentation ear.
• Filter theory would say meaning can’t transcend ear presentation.

Question 8

Moray (1959) also called into question Filter Theory by replicating study but inserting the participant’s own name into the M2 unattended channel - what did he find?

Answer 8

• Person’s own name often detected on unattended channel
• Selection based on meaning not consistent with the idea that meaning only extracted on the attended channel.

Question 9

What underlies the main difference between Early & Late Selection Theories?

Answer 9

Disagreement about location and properties of filter.

• Early Selections hold that the filter is directly after the sensory analysis and before the Semantic Analysis stage (LTM).
• Late Selectionists hold that the filter is after the semantic analysis stage (LTM). Further downstream.
• ES and LS theories agree recognition needs (a) encoding, (b) access to LTM
• LS theory: All stimuli access LTM, not sufficient for awareness
• ES theory: LTM activation = conscious awareness
• LS theory: need to pass filter for awareness

Question 10

Treisman’s (1961) Attenuation Model was similar to Broadbent’s filter, however, it differs in one way - what is the difference & what characteristics of the stimuli cause that change?

Answer 10

• Broadbent’s filter completely blocks unattended stimuli - Treisman’s partly blocks (attenuates) it - - “turning down the volume”.
• The filter is biased by the context** of the message, or the **message salience.
  
  • Highly salient stimuli (name), semantically related material (Dear Aunt Jane) gets through filter, shifts attention

Question 11

What evidence did Treisman & Geffen (1967) provide for the Attenuated Model? NB: simple Split-span experiment Shadow M1/Ignore M2).

What is the key criticism of this model?

Answer 11

Evidence:

• % of correct detections higher on the shadowed channel, but not zero on unattended channel.
• Consistent with a filter that attenuates stimuli instead of blocking them.

Criticism:

• Complexity of filter: Needs to respond to semantic context, distinguish related from unrelated stimuli – simpler alternative?
  
  • Late selection filter after LTM for access to semantic information.

Question 12

Describe Norman’s (1968) Late Selection Theory components - LTM, top-down (pertinence), bottom down (stimulus).

Answer 12

• LTM is central.
  
  • Inactive memory traces w/i LTM
• Things can be activated both top-down (pertinence)
• & bottom-up (stimulus driven).
• Things need both kinds of activation to get through the filter…
  *

Question 13

What evidence did McKay (1973) provide for late selection? (“They threw stones towards the bank” (ambiguous).

Answer 13

Shadow:

“They threw stones towards the bank” (ambiguous)

Ignore:

“ … … … … … river” or “ … … … … … money”

Recognition:

“They threw stones towards …”

(1) “…the side of the river”
(2) “… the savings and loan”

Recognition biased by previous shadowing task.

i.e. if ignoring … river - recognition was biased towards river bank.

Question 14

How did Von Wright, Anderson & Stenman (1975) use classical conditioning to provide evidence for the Late Selection Theory? (Banana + Shock).

Answer 14

• Conditioned galvanic skin response (GSR) (change in skin conductance) experiement.

1. Paired “banana” with a shock -> GSR
2. Shadowing Task:
   
   • Shadow “… … … “
   • Ignore “… … Banana … …”

• Semantic activation in the absence of attention via GSR from the unattended channel when “banana” was presented.
• Generalised to other words in category: other fruits.