Working Memory Capacity Influences Strategic Choice on a Cognitive Control Task
Lauren L. Richmond1, Thomas S. Redick2, Todd Braver3
1 Temple University, 2Indiana University‐Purdue University Columbus, 3Washington University in St. Louis
Background
Experiment 2 continued
General Methods
•WMC has been shown to influence abilities in a number of EXPERIMENT 1: EXPERIMENT 2: EXPERIMENT 3:
domains including fluid intelligence, reasoning and reading N=110
N=107
N=108
21.33 YOA (5.30) 20.56 YOA (2.27) 20.45 YOA (2.69)
comprehension (Engle, 2010)
26.61% male
55.14% male
32.40% male
•There is some suggestive evidence that people with high •Participants completed one cognitive control task (AX‐
WMC exhibit proactive control, whereas low WMC people CPT), and two complex WMC tasks (Operation Span and exhibit primarily reactive control. Symmetry Span; Unsworth, Heitz, Schrock& Engle, 2005). Order of administration was randomized.
Illustrations of proactive and reactive control. Reprinted from Braver, 2012.
R2
change
Accuracy
RT
AX
AY
BX
CX
.088***
.016
.001
.019^
.024*
.001
.062***
.059**
•AX, AY and BX data largely support the pattern of results from experiment 1. •While increased accuracy on CX trials was associated with increases in WMC, increased RTs for correct CX trials were also associated with higher WMC.
Experiment 3
•Participants were instructed as to the optimal strategy
Operation span and symmetry span task schematic. •The AX‐CPT paradigm involves attending to both a cue and •Hierarchical regressions were conducted. First, BY a target to determine the appropriate response. performance was entered into the model as a control AX‐CPT illustration; target sequence
variable. Next, WMC was entered into the model.
+
Experiment 1
A
+
+
X
Frequency
R2 change
Accuracy
RT
AX
40%
AY
10%
AX
.077***
.001
BX
10%
AY
.002
.037**
BY
40%
BX
.093***
.004
^: p < .10, *: p < .05, **: p < .01, ***: p < .001
Figure 3. Reaction time (RT) and error data for long‐interval trials (5,000 ms between cue and probe) for low and high WMC on the AX‐CPT paradigm. Reprinted from Redick & Engle, 2011.
•The higher one’s WMC, the greater likelihood of using proactive control (higher accuracy on AX and BX trials, longer RTs on AY trials). •It is not well understood how WMC might influence strategy choice. Redick and Engle’s (2011) results could be explained by two hypotheses other than reduced proactive control in low spans
•Namely, low spans may have reduced sensitivity to the predictive validity of the B‐cue (addressed by E1) OR low WMC participants may exhibit a target response bias (addressed by E2). In addition, both high and low spans showed impaired AY trial performance, which indicates the use of a proactive strategy.
Experiment 2
Frequency
AX
AY
BX
BY
Proactive
Reactive
70%
40%
10%
40%
10%
10%
10%
10%
Proactive R2
change
Accuracy
RT
Reactive R2
change
Accuracy
RT
AX
AY
BX
.107***
.008
AX
.004
.001
AY
.031^
.001
BX
.114***
.005
.002
.001
.110***
.004
•Higher WMC influences one’s ability to do well on AX trials whether the A‐cue is predictive of an X target or not (i.e. high WMC still confers an overall advantage in this task).
Conclusion
•Low WMC is associated with reduced levels of proactive control.
•WMC may not only modulate abilities in other cognitive domains by directly facilitating a given process, but may also support the optimal strategy choice and approach given task constraints and goals.
•In a task such as the AX‐CPT, high WMC can confer benefits as well as costs.
Frequency
AX
40%
AY
10%
BX
10%
BY
10%
CX
30%
We would like to thank Brynne DiMinichi for assistance with data collection
and the Redick lab for helpful comments on earlier versions of this poster. For an electronic poster reprint, please scan this code: