A Heated Debate of Working Memory Components
Continue from last post. Seemingly, DEsposito et al.s study is not very convincing. A few years later, another group of researchers reported another fMRI data (Veltman, Rombots and Dolan, 2003), which were interpreted in a different view from DEsposito et als.
This time the data were collected using a block-design and two somehow similar but different tasks. The first one is n-back (letter, instead of number) task. The authors considered this one as a task testing "maintenance" component of WM plus "manipulation" component. And the second task is Sternberg task, which involved only "maintenance" WM . What is that? Sternberg? According to the authors, its a task that presented a letter string, which could be any length between 2 to 7. So maybe like this "tu", this "ruv", or this "rhkwifj".
The participants job was to memorise the presented string within 10 seconds. They had 8 seconds to view the string and 2 second pause?. I suppose that meant the participants viewed for 8 seconds following a 2-second blank screen. Then what happened next? 15 single letters appeared on the screen, one after another one for 1 second. Between subsequent letter was 2.5 second blank screen I suppose (2.5 SOA). The goal for a participant was to recognise if the letter has ever appeared in the first presented letter string (answer Y/N by pressing two right-hand keys). So I guess the participant needed to press one time for each individual letter.
This report mainly concerned the task effect, WM load x task interaction on the dependent measures, which includes fMRI measured signals, RT and Performance (# of item - # of error) / (# of item) x 100%
Take home message? Their conclusion is more like an interpretation, instead of sticking closely with the data. They concluded no anatomical segregation between WM "maintenance" and "manipulation" component, but function differences exits. The function difference seems to rely on the readers willingness to accept that the two tasks test different things. The reason was that because both task, contrast along the WM load, almost elicited very similar areas b dlPFC, l vlPFC, b parietal; contrast between task types and two different time period in Stenrberg task during encoding and responding, b dlPFC. However, DEsposito et al.s study found very similar activation as well. Their argument was the different signal strengths measured from dlPFC and vlPFC elicited by the two types of task. The question is then why Veltman, Rombouts and Dolan did not examine the signal strength measured by the fMRI? They simply argued that two type of tasks elicited very similar activation regions, so no anatomical difference in terms of WM manipulation and WM maintenance. It seems that there were some problems there. This may be about the methodology of fMRI study?
Veltman, D. J., Rombouts, S. A. R. B., and Dolan, R. J. (2003). Maintenance versus manipulation in verbal working memory revisited: an fMRI study. NeuroImage, 18, 247-256.