Across individuals, faster latencies of peak evoked neuromagnetic fields in occipital cortex were related to greater FA in areas thought to contain fibers from cortical regions that modulate early visual responses: posterior parietal cortex and frontal eye field (FEF). We interpret these relations to reflect that white matter fibers linking these ocular motor regions to early visual regions transmit information that modulates the visual response, and that the speed of transmission depends on white matter properties, such as myelination and axon diameter, that contribute to FA (Waxman 1980
). Increased myelination would correspond to increased conduction velocity in fibers from top-down regions and thus a reduced latency of the visual evoked response. While there are prior reports of relations between FA and physiological measures such as the amplitude of steady-state visual evoked potentials (Butler et al 2005
), measures of functional connectivity (Boorman et al 2007
), and task-related BOLD signal (Baird et al 2005
; Olesen et al 2003
), this is the first report of relations between FA and the timing
of evoked responses. Although it will be important to validate these findings with larger samples and well-established visual paradigms, these relations suggest that biophysical properties of white matter affect the timing of early visual responses. More importantly, this preliminary report demonstrates a non-invasive method to relate the timing information from evoked-response experiments to the biophysical properties of white matter measured with DTI using a whole brain approach.
Factors that may contribute to individual differences in white matter physiology among healthy individuals are largely unknown, but likely include gene expression (Michailov et al 2004
). Individual differences in white matter physiology, particularly myelination have been proposed to contribute to variation in information processing speed (Luciano et al 2004
) consistent with recent reports of correlations between FA and cognitive reaction time (Bucur et al 2007
; Gold et al 2007
; Madden et al 2004
; Manoach et al 2007
; Nestor et al 2007
; Tuch et al 2005
; Westerhausen et al 2006
). Here we report relations between FA and the latency of evoked neural responses, which are presumably more proximal measures of neuronal function than are behavioral measures of latency. These relations may be based on the speed of axonal transmission from frontal and parietal regions that modulate early visual responses.
Classic peaks in event-related potentials such as the P1, P170 and N200 are thought to arise from sources within the occipital, parietal, and temporal lobes due to sensory and intermediate levels of processing. The neuromagnetic responses in this study originated in the occipital lobe and may represent an equivalent of the P100 or P1m component of the visual evoked response, in our case representing a response to foveating a stimulus following a saccade. Early visual responses are usually measured in response to a stimulus appearing in foveal vision during fixation. The standard deviation of the peak latency for early visual responses is approximately 15 ms, with some dependence on the experimental design (Fahle and Bach 2006
). Variability in the latency of early visual responses has been attributed to small saccadic eye movements that occur during fixation (Gur et al 1997
; Martinez-Conde et al 2000
). Since fixational microsaccades and longer-range saccades rely on overlapping circuitry, we reasoned that longer range saccades might also give rise to variability of visual responses both within and across individuals. In the present study the latency of the evoked magnetic responses ranged from 83 to 118 ms and there was a division between participants showing short vs. long latencies (see plots in ). While a larger sample might have produced a more continuous distribution of latencies, the division in latency may also reflect that the visual responses from short and long latency participants arose from different occipital regions. The limited spatial resolution of our MEG source modeling approach does not allow us to exclude this explanation of our findings. However, if this were the case, we would not expect to see strong and specific inverse correlations between the latency of these visual responses and FA in the white matter presumed to underlie top-down ocular motor regions.
Although plausible, the explanation of top-down modulation of the timing of early visual responses rests on the presumption that the correlated areas contain fibers that originate in posterior parietal cortex and FEF and synapse in early visual areas, which is not something that we can demonstrate using these techniques. For example, we hypothesize that the right frontal region showing a significant correlation contains fibers from FEF. The putative human homologue of FEF is located in the vicinity of the precentral sulcus and gyrus (Koyama et al 2004
; Paus 1996
) with distinct regions in the superior and inferior portions (Luna et al 1998
; Simo et al 2005
). While it is possible that the region showing a correlation contains fibers from FEF, we cannot definitively pinpoint fiber origins or endpoints. The correlated region in the deep white matter of right centrum semiovale is even less regionally specific. While the centrum semiovale contains major white matter fascicles from frontal and parietal cortex, whether the region identified by the correlation analysis carries fibers from ocular motor regions is impossible to determine in the present study. While DTI-based tractography may provide suggestive evidence of fiber origins and endpoints, it requires numerous assumptions regarding crossing fibers and the location and size of seed regions. An advantage of the present technique is that it makes no such assumptions, rather it examines relations of latency to FA in the entire brain. In the present study the findings were regionally specific in that they lay in regions that could plausibly contain fibers from ocular motor regions.
Although the anatomy of white matter in humans is not well established (c.f., Schmahmann et al 2007
), extensive anatomical connections between FEF, lateral intraparietal area (LIP), and occipital areas have been documented in monkeys (Andersen et al 1990
; Blatt et al 1990
; Cavada and Goldman-Rakic 1989
). Moreover, electrical stimulation of macaque FEF neurons modulates activity in V4 neurons (Moore and Armstrong 2003
). There is also evidence of top-down modulation of occipital responses in humans. Transcranial magnetic stimulation of FEF modulates both fMRI visual responses in early human retinotopic cortex (areas V1–V4) (Ruff et al 2006
) and event-related potentials (ERPs) recorded from occipital electrodes (Taylor et al 2007
). When applied to right posterior parietal cortex during visual search, transcranial magnetic stimulation eliminates the early phase of N2pc, an ERP generated in occipital lobe (Fuggetta et al 2006
). Finally, patients with lesions of right parietal cortex show abnormal fMRI visual responses in areas V1–V4, but only under conditions of increased attentional load, suggesting a failure of top-down attentional modulation of visual responses (Vuilleumier and Driver 2007
). These findings suggest that anatomical connections exist between FEF and posterior parietal cortex and early visual regions that may modulate both early and later visual responses.
It should be emphasized that some of the voxels in the correlated clusters contained gray matter. We chose not to restrict our analyses to deep subcortical white matter as it excludes white matter close to the gray-white border that emanates from specific regions, such as subcortical U-fibers. An important advantage of conducting the analysis in the whole volume is that meaningful, regionally specific correlations in these border areas can be detected. A disadvantage of sampling close to the gray-white border in a group analysis is that it results in partial average volume effects due to voxels that either span gray and white matter or that fall entirely in gray matter for only some participants given variations in cortical anatomy. The presence of gray matter in the correlated regions does not necessarily invalidate the results. Gray matter also contains myelinated fiber tracts that could contribute to the speed of neural conduction. However, since our FA values likely varied across individuals depending on how much gray matter was included, partial volume averaging represents a potential confound in our analyses. In control analyses, we were able to exclude partial volume effects as an account of our findings. Specifically, when we measured averaged FA values in individual participants based on voxels that lay entirely in white matter, the correlations with latency remained significant.
In conclusion, we observed relations between FA in parietal and frontal white matter and the latency of the MEG visual response in occipital cortex time-locked to arrival at a saccadic goal. These preliminary results suggest that the microstructural integrity of white matter connecting top-down cortical regions to early visual areas contributes to inter-individual variability of the timing of visual evoked responses. More generally, we introduce a noninvasive method that can illuminate the contribution of white matter physiology to inter-individual variability in the latency of evoked potentials in health and in neuropathologic conditions.