This study examined two hypotheses about what drives performance on working memory span tasks: the resource-sharing hypothesis and the temporal decay hypothesis. The researchers developed a new paradigm to test these hypotheses. In one experiment, they found that equating duration while manipulating cognitive cost did not significantly change span performance across tasks. A second experiment showed that a continuous operation span task requiring sustained attention had a lower span than a reading operation span task. This supports the idea that working memory span depends more on the attentional demands of the processing task than on duration alone.

1. Temporal Decay and Resource Sharing
in Working Memory Span Measures:
A New Paradigm
Pierre Barrouillet
Valérie Camos
Sophie Bernardin
Université de Bourgogne
Université René Descartes - Paris V

2. Working Memory Span Tasks
They involve:
• Maintenance of some information
• Simultaneous treatment

3. The Counting Span Task
(Case et al. 1982)
Participants are asked to:
• Count arrays of dots
• Maintain the results
The number of arrays increases until failure to recall

4. 8

5. 10

6. 6

7. Recall
8 10 6
Counting span of 3

8. Working Memory Span Measures
• They predict performances on complex activities
• They strongly increase with age

9. Developmental increase:
Alternative hypotheses
• Resource sharing - Case et al (1982)
• Temporal decay - Towse et al (1995, 1998)

10. The resource-sharing hypothesis:
Case et al. (1982)
OS
Treatment T
P
S
STSS
Maintenance

11. The resource-sharing hypothesis:
The developmental effect
OS
Treatment
Increase of processing
efficiency
STSS
Maintenance

12. The temporal decay hypothesis:
Towse & Hitch (1995)
Duration of processing and maintenance
With the development
Processing becomes faster
Duration of maintenance decreases
Then the span increases
Probability of recall

13. The temporal decay hypothesis:
Switching between Processing and Storage
Time
P S P S P S Children
Switching
P S P S P S Adults
Duration of processing determines duration of maintenance
because individuals switch from storage to processing thus
leading to the decay of memory traces.

14. The temporal decay hypothesis:
A test
Towse, Hitch, & Hutton, 1998, 2000
Long final condition
Card 1 Card 2 Card 3 Card 4
Higher
span Duration of maintenance
predicted
Card 1 Card 2 Card 3 Card 4
Short final condition

15. Results
Operation Span
6
5
4
Short final
3
Long final
2
1
0
8 years 9 years 10 years adults
Short-final span > Long-final span
Similar results with Counting Span and Reading Span

16. The temporal decay hypothesis:
Are the evidences compelling ?
Long final condition
Card 1 Card 2 Card 3 Card 4
Duration of maintenance
Card 1 Card 2 Card 3 Card 4
Short final condition
Long final condition more costly than short final condition
WM task only begins at the end of card 1

17. A new paradigm:
Barrouillet & Camos (2001)
Towse & Hitch:
equated the cost while manipulating the duration
Barrouillet & Camos:
equated the duration while manipulating the cost

18. A new paradigm:
Barrouillet & Camos (2001)
Time
Operation span task
L Ope 1 F Ope 2 B Ope 3 T Ope 4
L ba ba F ba ba ba B ba ba ba ba ba ba T ba ba ba
Baba span task
Mean solution
time for
operation 1

19. The results in children and adults:
Barrouillet & Camos (2001)
5
4,5
4
3,5
3
2,5 Operation Span
2 Baba Span
1,5
1
0,5
0
9 years 11 years adults
No interaction with age

20. Conclusions from the first experiment
• We cannot jettison any notion of cognitive resource in
accounting for performances in working memory tasks.
• Solving problems instead of saying « ba ba » did not result in
any dramatic decrease in span.
• Individuals can switch attention from the operations to the
letters to be remembered while solving operations

21. Prediction
• Tasks that would require continuous attentional focusing should
have a detrimental effect on span
A recipe for constructing working memory span tasks:
• Complex activities (e.g., reading, problem solving) are not
needed as a processing component.
• We just need a task that captures attention.

22. A new task :
The Continuous Operation Span task
Read the letter, the number, and then perform the operations aloud
Try to remember the letters
A preview
R 4 +1 -1 +1
The red operand indicates that a new letter will appear

23. R

24. 4

25. +1

26. -1

27. +1

28. H

29. 9

30. -2

31. -1

32. Recall

33. The Reading Operation Span task
Participants perform the same task except that they just have to
read the operations and their results
R 4 +1 5 -1 4 +1 5
Same duration as the Continuous Operation Span task

34. Test of our prediction
in adults
Compare:
• Continuous Operation Span
• Reading Operation Span
• Baba Span: saying "ba" during the same duration
Prediction:
COS should be smaller than the ROS
that should be smaller than the Baba Span

35. Results
4,5
4
3,5
3
2,5
2
1,5
1
0,5
0
COS ROS Baba Span
All the differences are significant

36. Conclusion
• The time does matter but ...
• Working memory span depend strongly on the processing
component the task involves
• There is no need to use complex activities to create working
memory span tasks
• Cognitive cost = attentional demand

37. Conclusion
Cognitive Cost = Attentional Demand
(Engle et al., 1999)
To perform self-paced complex operations (7 + 9 + 8 = 23 ?)
results in higher spans than ...
To perform computer-driven memory retrievals (+/- 1 or 2)

38. Conclusion
Continuous Operation Spans are significantly lower than Operation Spans
4
3,5
3
2,5
2
1,5
1
0,5
0
COS OS

39. Take Home Message
WM Spans strongly depend on Cognitive Cost
Cognitive Cost is Attentional Demand