Factor Analyses of Executive Function Impairment Due to Brain Injury

The last few posts on executive dysfunction among clinical neuropsychiatric patients has probably made clear that both frontal damage and behavioral impairments are typically very "messy," with more than one region and more than one function frequently suffering damage. Factor analytic methods have particular promise for identifying the behavioral impairments of such patients, since these methods can remove "shared method variance" and ultimately identify the real number of underlying impairments.

In this study, Busch, McBride, Curtiss & Vanderploeg used a related technique - principal components analysis - on behavioral measures of executive function from 102 adults with traumatic brain injury.

These adults were assessed 453 days (on average) after sustaining non-penetrating head injuries resulting in traumatic brain injury. A laundry list of tasks was used, including Wicsonsin Card Sort (WCST), Stroop, backwards digit span, semantic fluency, and trail making tests, among others. The results of each subject on each test was entered into a principal components analysis which derived three factors that account for 52.7% of the variance of the original measurements.

The strongest factor received the highest loadings from semantic fluency, WCST perseveration and trail making tasks, among others. The authors intuit that this factor involves both self-generation of behavior and set-shifting. The second strongest factor received high loadings from Stroop, backwards digit span, and WCST set failures. The authors suggest this factor involves the ability to sustain attention under conditions of interference. Finally, the third factor seemed to correspond to inhibition failure, as measured by incorrect designs on the Visual Spatial Learning test and the California Verbal Learning Test.

The authors interpret these various impairments in terms of a three factor model consisting of drive or motivation (which they relate to dmPFC), sequencing (which they relate to premotor and supplementary motor cortex) and control (which they relate to dlPFC). Control is thought to contain at least two components involving social/emotional control and cognitive control.

Busch et al. conclude that the first factor (self-generation of behavior and set-shifting) is related to drive, that the second factor (sustained attention despite interference) is related to Baddeley's central executive, and that the third factor (inhibition) is related to control.

The authors note several reasons for caution in making such conclusions. First, because primarily clinical measures of EF were used, none of these measures could be truly process-pure. Second, because the entire sample was brain damaged, these results may not generalize to other populations.