Name: Tyler Harrison

Psychology Ph.D. Dissertation Defense Presentation

Date: Wednesday, July 19, 2017

Time: 1:00pm

Location: JS Coon 148 

Advisor:

Professor Randall W. Engle, Ph.D. (Georgia Tech)

Thesis Committee Members:

Associate Professor Eric Shumacher, Ph.D. (Georgia Tech)

Associate Professor Mark Wheeler, Ph.D. (Georgia Tech)

Professor Christopher K. Hertzog, Ph.D. (Georgia Tech)

Professor David Washburn, Ph.D. (Georgia State University)

Title: N-BACK AS A MEASURE OF WORKING MEMORY CAPACITY

Abstract: 

One of the most important findings in cognitive psychology, is the relationship of working memory capacity (WMC) to a host of important cognitive activities, the manner in which WMC interacts with many different cognitive variables, and the consequences for the individual when WMC is reduced due to interventions such as sleep deprivation and psychopathology.  Many studies have been conducted in the past few decades to understand the nature of WMC, what psychological mechanisms account for the relationship between WMC and real world cognitive tasks, and the brain regions associated with individual variation in WMC.  However, one often over-looked problem is that researchers use different cognitive tasks to measure and study working memory capacity: differential studies have historically used complex span tasks to assess WMC.  However, n-back tasks are often used in neuroimaging studies because the task lends itself to the requirements of fMRI studies. 

The implicit assumption is that both types of tasks measure the same construct.  For the present study, we measured both complex span performance and n-back performance, in 328 subject, to see if they measured the same construct.  We manipulated the size of the stimulus pool for the n-back tasks to see if n-back tasks with more interference (i.e., with a smaller stimulus pool) are more strongly correlated with cognitive ability.  Additionally, we manipulated lure trials within the n-back tasks to examine whether lures that produce the most interference are more strongly correlated with complex span performance.  We found that complex span tasks and n-back tasks measure separate constructs, the number of stimuli that an n-back task uses changes its correlation to other measures of cognitive ability, and the false alarms to lures that produce the most interference are more strongly correlated with both complex span tasks and fluid intelligence but only for n-back tasks with a small stimulus pool.