How are Memory and Attention related in Working Memory? Elke Lange, Christian Starzynski, Ralf...

62
How are Memory and Attention related in Working Memory? Elke Lange, Christian Starzynski, Ralf Engbert University of Potsdam

Transcript of How are Memory and Attention related in Working Memory? Elke Lange, Christian Starzynski, Ralf...

How are Memory and Attention related in Working Memory?Elke Lange, Christian Starzynski,

Ralf EngbertUniversity of Potsdam

Memory and Attention

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

• Current models of Working Memory include a strong attentional component (e.g., Focus of attention in the WM model of Cowan, 2005, or Oberauer, 2002)

• Attention serves as a component for a) keeping information available b) selecting information from either the

environment (=encoding into memory) or from memory (=memory access)

c) manipulation of memory representations

• Current models of Working Memory include a strong attentional component (e.g., Focus of attention in the WM model of Cowan, 2005, or Oberauer, 2002)

• Attention serves as a component for a) keeping information available b) selecting information from either the

environment (=encoding into memory) or from memory (=memory access)

c) manipulation of memory representations

Memory Encoding

• Targets need to be selected• Usually: Selection of targets out

of non-target information• Non-targets = distractors to some

extend• Investigating selection processes

by manipulate and control distraction by irrelevant information

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Why: Distraction?

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Distraction is dangerous

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Background music instead of narcotic drugs

Distraction is helpful

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Distraction can hinder and help

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Classic Paradigm

74612835

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Why are irrelevant stimuli disturbing?

1. Similarity-based interference

2. Problem of limited attentional capacity

relevant irrelevant

relevant

relevant

irrelevant

e.g., Baddeley, 1986; Neath, 2000

e.g. Page & Norris, 2003

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

The working-memory model of Cowan (1995)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

The working-memory model of Cowan (1995)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

Activated part of

long-term memory

The working-memory model of Cowan (1995)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

Activated part of

long-term memory

The working-memory model of Cowan (1995)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

Activated part of

long-term memory

The working-memory model of Cowan (1995)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

Activated part of

long-term memory

Focus of

attention

The working-memory model of Cowan (1995)

Focus of attention:• Limited to four units

(Cowan, 2001)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

Activated part oflong-term memory

Focus ofattention

Central Executive

The working-memory model of Cowan (1995)

Focus of attention:• Limited to four units

(Cowan, 2001)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

Activated part oflong-term memory

Focus ofattention

Central Executive

voluntarilyFocus of attention:• Limited to four units

(Cowan, 2001)• Voluntary selection

The working-memory model of Cowan (1995)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Long-term memory

Activated part oflong-term memory

Focus ofattention

Central Executive

voluntarilyFocus of attention:• Limited to four units

(Cowan, 2001)• Voluntary selection• Automatic recruitment:

The working-memory model of Cowan (1995)

a) Change in physical properties (e.g. loud noise, sudden movement)

b) Personal relevance (e.g. own name)

automatically

Short sensory storeElke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

automatically

Central Executive

voluntarilyFocus of attention:• Limited to four units

(Cowan, 2001)• Voluntary selection• Automatic recruitment:

The working-memory model of Cowan (1995)

a) Change in physical properties (e.g. loud noise, sudden movement)

b) Personal relevance (e.g. own name)

Short sensory storeElke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

automatically

Central Executive

voluntarilyFocus of attention:• Limited to four units

(Cowan, 2001)• Voluntary selection• Automatic recruitment:

The working-memory model of Cowan (1995)

a) Change in physical properties (e.g. loud noise, sudden movement)

b) Personal relevance (e.g. own name)

Short sensory storeElke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Focus of attention:• Limited to four units

(Cowan, 2001)• Voluntary selection• Automatic recruitment:

The working-memory model of Cowan (1995)

a) Change in physical properties (e.g. loud noise, sudden movement)

b) Personal relevance (e.g. own name)

Central Executive

Short sensory storeElke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Predictions of the Cowan Model

change in physical properties

items in the focus loose their activation benefit

leading to memory impairment for these items

deviation of the focus

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

• Relevant task: verbal or spatial serial recall

• Irrelevant stimuli (synchronized with the relevant stimuli):– Auditory: Tones – Visual: Color stripes beside the relevant

frame

Experiment 1: Auditory vs. Visual

distractor

5

17

38

Tone distractor Color distractor

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Results: Experiment 1

40

50

60

70

80

90

verbal spatial%

co

rre

ct

silencerepetitionchange

Irrelevant colors

Verbal Spatial

40

50

60

70

80

90

verbal spatial

% c

orr

ec

t

silencerepetitionchange

Irrelevant tones

Verbal Spatial

♬?

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

• Domain-specific effect of distraction?– Distraction effect in the verbal task with

tone change– No distraction effect in the spatial task

with color change• Possible problems with the choice of the

irrelevant visual stimuli: – Maybe similarity plays a role in the visuo-

spatial domain (contingent capture: Folk, Remington, & Johnston, 1992)?

– Irrelevant stimuli outside of the visual relevant region might be not distractive (Eriksen & Eriksen, 1974; Awh & Pashler, 2000) ?

Summary Exp. 1

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

• Conditions: Repetition or location change• Hypotheses:

Distraction effect spatial task

No distraction effect verbal task

7

Experiment 2: Distraction by

irrelevant objects?• Relevant task: verbal or spatial• Irrelevant stimuli: visual-spatial

– Object-like– High perceptual similarity to the relevant stimuli– Inside the relevant visual-spatial region– Change has a spatial dimension

demo

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

40

50

60

70

80

90

verbal spatial

% c

orr

ec

t

repetition

change

Results: Experiment 2

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

automatically

Central Executive

Short sensory memory

voluntary

Focus of

attention

Criterion for distraction:Change in physical properties?

Not any changes but a)specific changesb)domain-specific effects

Distraction of verbal task by tone change

Distraction of spatial task by object location change

Cowan Model

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Thresholds: Effected by Task Set

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Time

Strength ofDiscrepancy Signal

Threshold A3

Threshold A2

t1 t2 t3

Threshold A1

Modified Model

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

automatically

Central Executive

Short sensory memory

voluntary

Focus of

attention

Modified Model

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

automatically

Central Executive

Short sensory memory

voluntary

Focus of

attention

Modified Model

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

automatically

Central Executive

Short sensory memory

voluntary

Focus of

attention

Threshold

Modified Model

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

automatically

Central Executive

Short sensory memory

voluntary

Focus of

attention

Threshold

Modulated by voluntary control

Experiment 3: Contingent Capture

• What is the role of similarity?• Visual Search: Target has to be

selected top-down, location-changing distractor captures attention bottom-up

• The more similar a distractor is, the more likely he is attended to

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

4

9

2

70

3

Memory Task: Serial Recall

Memory Task: Serial Recall

Verbal Task: • Digit identity• 8 items

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

1 2 3

4 5 6

7 8 90

Spatial Task: • Digit location• 5 items

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

4

Memory Task with Distractors

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

9

Memory Task with Distractors

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Memory Task with Distractors

Distractors (Exp.3)

2

4

0

3

79

Repetition of distractor position

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

2

4

0

3

79Change of distractor position

36 trials

9 trials

5K

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

No feature overlap in color/shape

Feature overlap in color/shape

Feature overlap in color/shape

Feature overlap in color/shape

Different category (object)

Different category (object)

Same category (verbal)

Same category (verbal-numeric)

Salient Not salient

Not salient

Not salient

5K

Distractors (Exp.3)

Repetition versus Change Trials

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

K 5

Repetition versus Change (collapsed across

tasks)

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

5K

Results Experiment 3

• Distraction of verbal task by object location change, if object is verbal

• Effect of distractor similarity ➜ Top-down modulated attentional

capture ➜ Contingent capture

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Experiment 4• Target needs to be selected from

non target items (visual search)

• Target needs to be encoded into and retrieved from memory

➜Can we dissociate effects due to attentional selection and memory processes?

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Record of Eye Movements

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

EL 1000 from SR ResearchSample Rate 1000 Hz

Why eye movements?• Usually the gaze indicates where

attention and information processing is located

• Information that is more difficult is fixated longer

• Information that captures attention captures the gaze

• Eye movements as indicators for attentional capture and information processing

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Monitoring

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

x

y

Time x y1 534 487

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Categorization of eye movements

Vel

ocity

xV

eloc

ity y

Time / ms

saccade

saccade

Fixational eye movements

Fixational eye movements

Eye Movements

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Verbal

Example Trial: Verbal Task

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Distractors

• No overlap of visual features (color/shape)• Categorically different• Salient

5• High feature overlap (color/shape)• Categorically identical• Highly similar distractor

2

4

0

3

79

Repetition of distractor position

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

2

4

0

3

79Change of distractor position

40 trials

8 trials

Timing of Item Cycle

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

0 200

7

off

off

1000

delay

1200

end of cycle

Change Locked Positions

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

ChangeChange - 1

Change + 1

Control - 1Control

Control + 15

2

9

3

41 Repetition of

distractor position

2

4

0

3

79Change of distractor position

Memory & Attenion: verbal

Distractor Rel. Item

ControlChange

Distractor changesposition

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Memory & Attention: verbal

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

5

ControlChange

Distractor Rel. Item

Verbal Task Error Types

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

1 2 3

4 5 6

7 8 90

1 2 3

4 5 6

7 8 90

5

8 correct items

Digit 5

Alternative item

➜ Order errors

➜ Item error: distractor intrusion

➜ Item error: control intrusion

Verbal Task Error Proportions

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

5

Control Change

Alternative IntrusionIntrusion of Digit 5Order errors

Verbal Task: Proportion of Answers

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

5

Summary Exp. 4• Change of distractor position affects saccade

rate attentional capture• BUT:

– Attentional capture does not necessarily affect memory performance (verbal task, triangle, change locked position)

– Attentional capture can affect memory for events prior to capture (verbal task, digit 5, pre-change position)

– Attentional capture can be controlled for relatively fast (verbal task, triangle, change+1 position)

– Control for attentional capture varies for different distractors

• Dissociation of visual (overt) attention and memory processes

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Summary

• A location-changing object captures the gaze (=attention) in both tasks

• But the distraction effect is top-down modulated and contingent on task properties

• If attention is directed to verbal properties verbal distractors are effective

• If attention is directed to spatial properties spatial information is distractive

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam

Thanks for your attention!

Elke B. Lange - Christian Starzynski - Ralf Engbert University of Potsdam