Steven Phillips and Kazuhisa Niki
Neuroscience Research Institute, National Institute of Advanced Industrial Science and Technology (AIST)
Working memory is affected by items stored and the relations between them. However separating these factors has been difficult, because increased items usually accompanies increased associations/relations. In a Sternberg paired recognition paradigm, the number of items and maximum number of associates were varied independently by three list conditions: 1. AB,CD: four/one; 2. AB,CD,EF: six/one; and 3. AB,AD,CB: four/two, respectively, using Japanese characters in Experiments 1 (ideograms - kanji) and 2 (phonograms - hiragana); digits in Experiment 3; and shapes generated from Fourier descriptors in Experiment 4. Across all materials, right temporo-parietal and middle frontal gyral activity was found with increased associates, but not items during study (Figure). Because right parietal cortex has been implicated in bilateral shifts of spatial attention (Losier and Klein, 2001), we suggest that encoding of overlapping pairs in the two ssociates condition is enhanced by repeatedly attending to component items. The effects were weaker (by cluster size and Z scores) and more temporal for digits and hiragana where pairs were less likely to be novel. There were also fewer error and shorter response times for digits and hiragana. For digits, in particular, it suggests chunking: re-representing complex information to reduce memory load. Digit pairs appear less confusable when interpreted as single double-digit numbers. Support for chunking comes from observations that right parietal cortex is involved with shifts of attention, but not binding in visual feature tasks (Ashbridge et al, 1999), this role is reduced with training (Walsh et al, 1999), and neurons in monkey anterior temporal cortex were selective to conjunctions of shapes after training (Sakai and Miyashita, 1991).
Six Japanese university students for each material type were tested . Subjects were given 60 trials, consisting of an encoding phase of two or three item pairs followed by a probe pair. Subjects must determine whether the probe pair appeared in the list. We used a 3 (List) x 2 (Probe) x 10 (Trial) design.
Scanning was performed on a 3.0-T MRI Scanner (GE 3T Signa) with EPI capability. 18 axial slices (5.5 mm thick, interleaved) were set to cover the entire brain. A T2* weighted gradient echo EPI was employed. The imaging parameters were TR=2 sec, TE=32 ms, FA = 70 degrees, FOV=20x20 (64x64 mesh). Images were pre-processed (timeslice adjusted, realigned, normalized and smoothed) by SPM99. Data were estimated to establish by a fixed model in which there were 3 block (List) types for the encode phase, and 3 (List) x 2 (Probe) + 1 (error) event types for the probe phase. Block types were modelled by a box-car convolved with the canonical hemodynamic response function. Event types were modelled by the canonical hemodynamic response function. Contrasts were set at p < .05, corrected for multiple comparisons.
References:
Ashbridge, E. et al. (1999). Neuropsychologia, 37, 999-1004.
Losier, B. J. W. and Klein, R. M. (2001). Neuroscience and Biobehavioral Reviews, 25, 1-13.
Sakai and Miyashita, (1991). Nature, 354, 152-155.
Walsh, V. et al. (1999). Neuropsychologia, 37, p245-251.
|
Pcorr |
Z |
Voxels |
x |
y |
z |
Region |
BA |
|
0.000 |
Inf |
109 |
40 |
-47 |
41 |
Inferior Parietal Lobule |
40 |
|
0.000 |
6.82 |
72 |
32 |
11 |
55 |
Middle Frontal Gyrus |
6 |
|
0.003 |
5.66 |
9 |
40 |
-70 |
29 |
Middle Temporal Gyrus |
39 |
|
Table 1. Conjunction contrast (encoding phase) of b4-u6 lists for kanji and shape, indicating corrected p value, Z score, xyz-coordinates (Talairach), region and Brodmann area of the peak cluster voxel, and number of cluster voxels exceeding threshold (p<.05 corrected). |
|||||||