within subject and different across subjects. The voxel coordinates reported in the
tables are transformed (Brett et al., 2001) from MNI to Talairach space. It was found
that activated clusters spanned across several brain areas in the rotated > follow
contrasts. For these contrasts, we counted the number of significant voxels using
Talairach Daemon software (Lancaster et al. 2000). Location of peak activation for
regions having more than 5 significant voxels were identified using ROI masks based
on the the Talairach daemon with the help of WFU Pick Atlas software (Maldjian et
al. 2003). For the other contrasts, the number of significant voxels per cluster has
been reported.
Brain activation results at selected cortical and subcortical areas are overlaid on
normalized structural MRI of one of the subjects’ skull stripped using the Brain
Extraction Tool (Smith, 2002) available in the MRIcro software (Rorden & Brett,
2000). Functional overlays on transverse slices were achieved using MRIcro software.
At selected functional regions of interest, average blood oxygen-level dependent
(BOLD) signal was calculated for each of the six sessions from a 3mm spherical
volume for each subject. We performed brain-behavior correlation analysis taking
average values across subjects (Bland & Altman, 1994). Pearson correlation
coefficient (R) and its two-tailed significance level (p) were computed.
RESULTS
Behavioral Results
Two behavioral measures were calculated - the average number of sets completed per
trial in a block indicating the accuracy (Fig. 2 a) and the average set completion time
revealing the speed of performance (Fig. 2 b). Repeated measures ANOVA for the
follow condition revealed that accuracy and response times were similar across the
two experiments (visual-normal and motor-normal). Further there were no learning
related improvements in the follow condition. Thus, follow serves as a stable baseline
measure to assess the progress of performance in the test conditions.
Repeated measures ANOVA for the normal (normalm, normalv) and rotated (motor,
visual) test conditions revealed significant (p<0.0001) improvements in accuracy and
response times from the early to the late stage. Figure 2 shows comparisons between
the test conditions across the two experiments. While the normal conditions had
similar performance measures, the comparisons between the rotated settings revealed
superior performance for the motor setting than the visual setting. While subjects
attained accuracy levels similar to that of the motor setting by the late stage of visual
setting, the response times remained significantly slower in both early and late stages
of visual setting.