20,000 Flanker Trials: Are the Effects Reliable, Robust, and Stable? Ken Paap 1, Sarah Wagner 1,...

Post on 14-Jan-2016

218 views 4 download

Tags:

Transcript of 20,000 Flanker Trials: Are the Effects Reliable, Robust, and Stable? Ken Paap 1, Sarah Wagner 1,...

20,000 Flanker Trials: Are the Effects Reliable, Robust, and Stable?Ken Paap1, Sarah Wagner1

, Hunter Johnson1,Morgan Bockelman1, Donish Cushing1, and Oliver Sawi1,2

1San Francisco State University, 2University of Connecticut

Executive functions (EFs) consist of a set of general-purpose control processes believed to be central to the self-regulation of thoughts and behaviors that are instrumental to accomplishing goals. EFs include components for inhibitory control, switching, monitoring, and updating.

Introduction Do Attentional Network Effects Decrease Beyond 20 Sessions?

Executive Function

Time Course of the Flanker Interference Effect:First it gets larger, then it gets smaller

The Attentional Network Test (ANT)

Test-Retest Reliability of Flanker Effect & 12 other Measures of Executive Function

Conclusion

The flanker effect has moderate test-retest reliability. As a measure of general inhibitory control, the flanker effect shows low levels of convergent

validity. The magnitude of the flanker effect increases during an initial session, shows little change

on a second day, but then continues to decrease for at least 100 sessions. Unless the flanker task is a potent training exercise for enhancing inhibitory control, the

decrease in the flanker effect reflects task-specific learning. Given that the decrease in the flanker effect involves continuous improvement on both

congruent and incongruent trials and that they highly correlate, the improvement likely involves a process common to both trial types.

Test-Retest Reliability of 4 TasksMagnitude of Flanker Effect over 3 Blocks within a Single Session (n=224)

Magnitude of Flanker Effect over 20 Sessions (n=8)

Magnitude of Flanker Effect over 2 Sessions about a Week Apart (n=79)

Session x Trial Type InteractionF(1, 78) = 7.36, p = .008 This 7 ms decrease in the magnitude of the

flanker effect is small, but statistically significant.

Block x Trial Type interactionF(1, 222) = 24.03, p < .001

The increases in the magnitude of the flanker effect are significant:Block 1 to 2, F(1, 222) = 24.03, p < .001Block 2 to 3, F(1,222) = 6.14, p = .014

Session x Trial Type InteractionF(19, 133) = 4.92, p < .001

Congruency and Flanker Effect Over 100 Sessions

Flanker Effect: Moderate Test-Retest Reliabilityand Questionable Convergent Validity

Orienting Effects Alerting Effects

Low reliability is associated with lower levels of replicability (e.g., LeBell & Paunonen, 2011)

Low convergent validity implies that the magnitude of the flanker effect is determined by task specific strategies and mechanisms rather than a general inhibitory-control ability.

Note that Salthouse reported no correlation between an arrow and letter version of the flanker task!

If incongruent trials require an additional inhibitory control process not required on congruent trials, then their very high correlation (r = .924) is surprising.

High correlations are associated with unreliable difference scores (Salthouse & Hedden, 2002) because no unique process is isolated.

Congruent and Incongruent RTs are Highly Related