Attention to attributes and objects in working memory

1. Attention to attributes and objects in working memory Cowan, N., Blume, C. L. and Saults, J. S. (2012). Journal of Experimental Psychology: Learning, Memory, and Cognition

2. Take home message • Multiple attributes compete for attention, but that retaining the binding between attributes is accomplished only by retaining the attributes themselves.

3. Previous research • Performance levels were similar no matter whether participants had to pay attention to one or several attributes of the stimuli at the same time. (Luck & Vogel, 1997) • Knowledge of the color and orientation of objects in an array are largely independent. (Bays, Wu, & Husain, 2011; Fougnie & Alvarez, 2011)

4. Limitation of previous research • They did not have as tools models of the number of items encoded into WM.

5. Present study • Use 4 models to reexamine the effects of attention on the retention of two common attributes of ordinary object, color and shape • Attribute binding is addressed in present study • Color and shape recall were correlated (Gajewski and Brockmole, 2006) • Binding is nothing more than independent attributes that are retained for the same object by chance (Vul and Rich, 2010)

6. Models • Model 1: from Pashler (1988)memory array  test array (1 possible change). • Model 2: “Reverse-Pashler” Modelmemory array  single, central test probe

7. Models • Model 3: from Cowan (2001)memory array  single test probe in original position • Model 4: Color-shape binding modelprobe for every trial is either identical to one of the array items or consists of features from two of the array items recombined

8. Modeling limitations • Different models are logically appropriate for different change detection procedures and task demands • 2 assumptions of all models: • The attributes values were categorically distinct • WM consists of a fixed number of slots

9. Questions for experiments • Experiment 1:How binding information is retained in WM with reference to a type of probe combined or newly combined, with the shape from one array object and color from another. • Experiment 2:whether attending to both color and shape concurrently is tantamount to attending to their binding.

10. To examine memory for attribute and binding information across a variety of attention conditions and circumstances to determine whether attributes within objects compete for attention. Experiment 1

11. stimuli • 7 colors • 7 shapes

12. Probe types • The study phase was unchanged throughout all conditions

13. Design • 3 factors: probe type (new attribute or recombined attribute)probe location (target location or center)attention condition(attend to color, shape, or both) • Counterbalance (attention  location  type) • 64 practice and 256 test • Attention: color : shape :both = 1:1:2

14. New-attribute recombined-attribute

15. Result • Position • Probe type

16. Result – items in WM • The estimate of items in WM was higher for centrally located probes than for target-location probes • New-attribute analysis:only a main effect of attention: both>single

17. Result – items in WM • Target-location analysis: • A very large advantage for new-attribute changes over attribute-recombination changes • An advantage for attention to one attribute over attention divided between two attributes

18. Discussion • It illustrate the limitations of various testing procedures • Suggests a limitation in the use of a method in which the probe is placed in the location of a target item • No evidence of a benefit from restricting search in this way in this procedure

19. Discussion • Central probe may be the best probe placement, especially when the experimental goal is to include attribute-recombination trials to be compared to new-attribute trials

20. Discussion – item and binding information • The number of objects for which individuals could retain attributes was greater than the number for which they could retain attribute bindings • This experiment can’t judge the memory strength separately for color and shape attributes

21. Discussion – effects of attention • There was an effect of dividing attention between attributes on the ability to retain any one attribute in WM. • The extra amount of information retain when only one attribute is task relevant turns out to be a relatively small but significant factor.

22. Discussion • The procedure in experiment 1 does not distinguish between effects of attention at encoding and retrieval

23. Whether the effects of dividing attention occurred during encoding or retrieval processes Whether attention to bindings draws resources away from the entry of attributes into WM Experiment 2

24. design • 3 types of attribute change:color, shape, color-shape combination • 4 types of trial blocks:attend-color, attend-shape, attend both, attend-combination

25. Color-shape combination color shape

26. design • Same color (or shape) retrieval probes • 3 levels (4 types) of attention at WM encoding: • Attention to the corresponding single-attribute • Attention to both color and shape • Attention to the specific color-shape combinations • If the effect in Exp1 occur during encoding: dividing attention between 2 attributes at encoding should reduce the number of objects with color (or shape) in WM, compared to single-attribute color (or shape) condition.

27. method • 32 practice trials attend-color: attend-shape: both : combination=1:1:2:4 • 256 test trials (32:32:64:128)

28. Result – proportion correct

29. Result – Items in WM • Attribute v.s. binding information:the number of color-shape bindings in WM was smaller than the number of colors or shape in WM. • Attribute and attention effects:2 main effects and no interaction • Colors retained > shape retained • Single attribute> both ≈ specific combination

30. Result – Items in WM • Derivation of bindings from attribute WM: • Divided attention does not impede memory for bindings any more than it impedes memory for attributes • The object for which one attribute is encoded are independent of the objects for which another attribute is encoded.

31. Result – Items in WM • WM capacity expressed in object:the result is rather similar across condition

32. Discussion • Replicate experiment 1: • The number of binding in WM was less than that of attributes, regardless of the attention condition • Effect of dividing attention between two attributes of an objects • Encoding the two attributes absorbed as much attention as encoding the binding

33. Discussion • Provide a simple model suggesting that the information available on binding trials was quite similar to what would be expected if the color and shape attributes were independently coded • Yield an estimate of the number of objects for which at least one attribute was encoded into WM • Slight but significant advantage for color over shape

34. General discussion • A fixed number of objects represented in WM, about 3 on average • Partial completion of objects in WM • Variation in relative proportion of objects having color versus shape representation, depending on the direction of attention • Knowledge of color-shape binding only to a degree

35. General discussion

36. General discussion • Color and shape can be directly associated without location as an intermediary • The capacity limit is the number of objects not the number of spatial location (Lee & Chun, 2001) • Pairs of attributes that are encoded for the same object may be mentally bound(Allen et al., 2006) • Sometimes the representation of attributes within an object is incomplete because of attentional limitation

37. Thank you for listening The end