The present fMRI study aimed to compare the WM HRF to GM HRFs. As predicted, while the WM HRF general features and temporal characteristics were similar, the response amplitude was reduced relative to the GM HRFs. While this drop appears to be consistent with known reductions in the ratio of CBF
] and CBV
] between WM and GM, it may not reflect a direct relationship between the two factors. Rather the result represents the initial step to evaluate the WM HRF, with more work needed to fully characterize the factors that contribute to response reductions.
Interestingly though, the WM HRF could be detected successfully in the raw time course data even with a lower number of trials. While the increased response variability limited the sensitivity to amplitude differences, the overall trend and temporal characteristics were consistent with the more powerful FIR results. Detecting the raw WM HRF is important because it demonstrated that, similar to GM fMRI activation, WM fMRI activation is observable even within the raw time course data. Given that the analysis focused on ROIs with the strongest activation, the variance is likely also large in this instance. It is likely that increasing both the number of trials and the extent of activation included in the analysis would improve the ability to characterize differences between WM and GM HRFs.
Both the WM and GM activation results closely replicate prior findings on WM fMRI activation
]. Similar to these results, it was possible to demonstrate WM activation at both the group level and in the majority of individuals (87.5%). As shown in the group activation maps, most participants showed activation in the isthmus of the corpus callosum. Mazerolle and colleagues showed that white matter fibre tracts connect from the parietal lobes through the isthmus
]. This network supports the integration of high-level sensory information, such as that required by interhemispheric word and face transfer in the Sperry task
It is noteworthy that our results differ from those of Yarkoni et al.
]. In addition to a reduction in the response amplitude of the WM HRF, this study reported a delay in WM HRF peak latency. The major difference being that Yarkoni et al.
] did not restrict the analysis to activation localized first using the GM HRF. Taken together, the overall results suggest that an amplitude reduction should be expected, but more work is needed to fully characterize the response timing.
The current results should be interpreted with at least two major caveats. First, in order to identify WM activation, we first used the basic HRF function derived from GM activation. While this approach was necessitated as an initial exploratory step, the GM approximation may not have been sensitive to all WM activation. Therefore, the current results may improve sensitivity to WM activation through response feature optimization, but they do not address whether other WM HRFs exist that do not match the GM HRF. For this, a model-free analysis, like independent components analysis, would provide important additional insight
]. Second, both the temporal resolution and the regional variance in WM HRF time course data remain key factors for future investigations. Future studies should increase the temporal resolution to allow for better characterization of the WM HRF (using reduced but focused spatial coverage). Also, the vasculature supplying the corpus callosum differs from that of other WM tissue; consequently influencing the ability to generalize the current findings.