Friday, May 23, 2014
The concurrent validity of the N-back task as a working memory measure
Jaeggi, S.M., Buschkuehl, M., Perrig, W. J., & Meier, B. (2010). The concurrent validity of the N-back task as a working memory measure. Memory, 18, 4, 394-412. doi: http://dx.doi.org/10.1080/09658211003702171.
This article presents psychometric properties—reliability and concurrent validity—of the n-back task to assess working memory (WM) and other aspects of cognition such as executive functions (EF) and fluid intelligence (Gf). In the n-back task, the participant views or hears items in a sequence with items occasionally repeating. The participant is asked whether a current item is the same as the one presented ‘n’ back. So, in a 1-back task, the person judges if the item is the same as the one before; in a 2-back task, the person judges if the item is the same as the one before the one that was just shown (i.e., 2-items ago), etc. You can try a n-back task at this website: http://cognitivefun.net/test/4.
The n-back task is frequently used in neuroimaging studies and sometimes is considered a “pure” WM measure. However, not all of the previous evidence show strong links between n-back performance and scores on other measures commonly used to assess working memory such as simple or complex span tasks. In a simple span task, a person recalls the list of items presented such as repeating a list of digits. In a complex span, the person must manipulate the items in the list in some way (i.e., make a judgment about the item), and then recall items. Stronger associations have been found between the n-back and measures of executive functions and fluid intelligence.
In this paper, three experiments were conducted to further examine the reliability of the n-back task by measuring performance in 1-, 2-, and 3-back tasks, simple and complex span tasks, and EF and Gf. Overall, the n-task was moderately reliable based on split-half correlations comparing odd and even items (correlation range: 0.11 to 0.94). Reaction time (RT) was more reliable than accuracy, and the 2-back was more reliable than 1- and 3-back. The empirical results also confirmed previous findings of poor association between n-back and complex span tasks, but moderate correlation with simple span tasks. The test correlated poorly with EF, however it presented moderate-to-high correlations with Gf.
Although n-back and complex span tasks are considered to measure working memory, the tasks demands differ with n-back tasks requiring more continuous performance and attentional control and complex span tasks being more self-paced and self-ordered. It may be that the n-back and complex span tasks explain different parts of the variance in Gf. Caution is warranted in the use and interpretation of the n-back task.
Blogger: Alberto Filgueiras is a visiting doctoral student to the LWM lab from the Pontifical Catholic University of Rio de Janeiro, Brazil.