1. Kim Uittenhove, Lina Chaabi, Valérie Camos, Pierre Barrouillet
Comparing interference patterns in working memory: An insight into domain-specific vs. domain-general maintenance
Kim Uittenhove, Lina Chaabi, Valérie Camos, Pierre Barrouillet

2. Maintenance in working memory
Domain-specific stores that contain representations tied to a particular domain (e.g., visual, auditory)
Phonological loop
Visuospatial sketchpad
Domain-independent mechanisms that allow working-memory storage regardless of content or modality
Attention
Maintenance in working memory

3. Testing domain-general storage
Testing interference when simultaneously storing information from different domains
Concurrent load effects
More interference = more overlap in storage mechanisms

4. A simultaneous maintenance paradigm
Fougnie, Zughni, Godwin, and Marois (2015)
8 experiments:
Basic (Expt 1)
Uncommon sounds (Expts 2-3)
Articulatory supression (Expt 4)
Long retention interval 9s (Expt 5)
With mask (Expt 6)
Blocked presentation of load + reversed task order (Expt. 7)
Intra-domain simultaneous maintenance (Expt. 8)
Probe recognition
Spatial
Verbal

5. Sample results…

6. Intrinsic domain-specific storage ?
No interference between the simultanous maintenance of auditory and visual material
Intrinsic domain-specific storage ?
Or the consequence of the paradigm ?
The way recognition probes working-memory

7. Present study
1) The role of the paradigm: Recognition versus active recall
- Recognition could rely on familiarity effects, thus rendering it more robust and operational even with less intact WM traces
- Recall performance may be a more appropriate measure of an active WM system

8. Recognition versus recall
Concurrent load interaction effects ?

9. Recognition (N = 24) F < 1 F < 1 F = 47.23 F = 76.93
Visual set size
1 spatial location
3 spatial location
5 spatial location
Auditory set size
2 letters
6 letters
F < 1
F < 1
F = 47.23
F = 76.93
Absence of concurrent load effects confirm the data of Fougnie et al. (2015)

10. Recall (N = 24)
Visual set size
1 spatial location
3 spatial location
5 spatial location
Auditory set size
2 letters
4 letters
6 letters
F = 5.34
F = 26.55
F = 26.88
F =
Significant interference effects, especially of concurrent letter load on spatial location recall accuracy

11. Role of the paradigm
Under recognition testing, there were no significant concurrent load effects.
Under recall testing, there were significant concurrent load effects.
1) Interference was detected by using an active recall paradigm, but was invisible in recognition data.
Simultaneous maintenance of visual and auditory information causes interference, hinting at the implication of domain-general maintenance mechanisms.
2) Understand the relative size of interference effects, and the relative implication of domain- general mechanisms
Comparing interference effects

12. Between- versus within-domain interference
Letters – spatial locations
Letters-digits
Ball movements – Spatial locations
Letters - Letters K K K
Using the same stimuli in the strong semantic overlap condition permits us to evaluate the effects of having separate presentation and response modalities, on the concurrent load effects.
Between
Within
Strong semantic overlap

13. Trial letters - letters
2 auditory letters
3 visual letters

14. L

15. J

16. *

17. H

18. D

19. S

20. D X B F C H #

22. L…J…
Using the same stimuli in the strong semantic overlap condition permits us to evaluate the effects of having separate presentation and response modalities, on the concurrent load effects.

23. A comparison of interference effects
(Recall acc w/ low concurrent load – Recall acc w/ high concurrent load)/ Recall acc w/ low concurrent load
(N = 24)
Concurrent load
effects ranging from F = 5.34 to F = 26.56
(N = 24)
Concurrent load
effects ranging from F = 4.25 to F = 32.44
Between- and within domain interaction effects are of comparable magnitude…
Relative importance of domain-general storage mechanisms ?
Visual letters
(N = 18)
Concurrent load
effects ranging from F = to F = 53.35
Sequential spatial locations with AS
Ball movements
Digits
With AS, auditory letter load effects on spatial locations (19%) are 70% of auditory letter load effects on visual letters (26%)
 Relatively large contribution of domain-general mechanisms in between-domain maintenance when blocking the phonological loop
Strong semantic overlap lap leads to 26% reduction in recall accuracy. Role of presentation and response modality ?
How to put in perspective interference effects ?
Spatial locations
Auditory letters
Auditory letters with AS
Auditory letters
Spatial locations
Auditory letters
Strong semantic overlap

24. Relative importance of interference effects
Effect of concurrently maintained load on recall accuracy of the memoranda
Comparison to set size effects:
Effect of load of the modality that is being recalled
(Low load recall acc – High load recall acc) / Low load recall acc
If (Interference effects == Set size effects)  adding letters in the concurrent modality has the same effect as adding them in the modality that is being recalled
Set size effects = upper limit
The effect of maintained memoranda load on recall accuracy of the same memoranda
For example, a smaller % of auditory letters is maintained when maintaining 6 compared to 2 auditory letters

25. Interference effects vs set size effects
Similar interference and set size effects for memoranda with strong semantic overlap (visual and auditory letters)
 adding letters in the concurrent modality has the same effect as adding them in the modality that is being recalled
Strong overlap in storage mechanisms, in spite of distinct presentation and recall modality
Role of stimulus-specific interference ?
Between-domain interference effects reach 43% of set size effects
But there is also representational interference, which will be larger with letters-letters, thus enlarging set size effects
Strong semantic overlap

26. Understanding relative size of interference effects
Between-domain interference effects are relatively important (43% of set size effects)
Considerable overlap in storage mechanisms ?
Larger interference effects (= almost identical to set size effects) for the strong semantic overlap condition
Larger overlap in storage mechanisms ?
More stimulus-specific interference ?

27. What about recognition ?
Concurrent load
effects Fs < 1
(N1 & N2 = 24)
(N = 9)
Small tendency for interference effects, mostly not significant
Lack of sensitivity of the recognition paradigm
Concurrent load
effects Fs ranging from < 1 to 5.71
No statistical analyses yet
Auditory letters *
Auditory letters
Auditory letters
Ball movements
Digits *
Spatial locations
Spatial locations
Visual letters
Strong semantic overlap

28. Set size and interference effects recognition
Relatively small set size effects
The recognition paradigm is less sensitive to set size effects, and by consequence interference effects, which most often are not significant

29. Discussion Recognition paradigm seems to lack sensitivity
Smaller set site effects
Non-significant interference effects
Possibly because it can operate on less intact memory traces and rely on familiarity effects, rendering it more robust than active recall
Active recall consistently yielded interference effects
Implication of domain-general maintenance mechanisms
Extent of their implication ????