Post-stimulus fMRI and EEG responses: Evidence for a neuronal origin hypothesised to be inhibitory

Abstract

Post-stimulus undershoots, negative responses following cessation of stimulation, are widely observed in functional magnetic resonance (fMRI) blood oxygenation level dependent (BOLD) data. However, the debate surrounding whether the origin of this response phase is neuronal or vascular, and whether it provides functionally relevant information, that is additional to what is contained in the primary response, means that undershoots are widely overlooked. We simultaneously recorded electroencephalography (EEG), BOLD and cerebral blood-flow (CBF) [obtained from arterial spin labelled (ASL) fMRI] fMRI responses to hemifield checkerboard stimulation to test the potential neural origin of the fMRI post-stimulus undershoot. The post-stimulus BOLD and CBF signal amplitudes in both contralateral and ipsilateral visual cortex depended on the post-stimulus power of the occipital 8-13Hz (alpha) EEG neuronal activity, such that trials with highest EEG power showed largest fMRI undershoots in contralateral visual cortex. This correlation in post-stimulus EEG-fMRI responses was not predicted by the primary response amplitude. In the contralateral visual cortex we observed a decrease in both cerebral rate of oxygen metabolism (CMRO₂) and CBF during the post-stimulus phase. In addition, the coupling ratio (n) between CMRO₂ and CBF was significantly lower during the positive contralateral primary response phase compared with the post-stimulus phase and we propose that this reflects an altered balance of excitatory and inhibitory neuronal activity. Together our data provide strong evidence that the post-stimulus phase of the BOLD response has a neural origin which reflects, at least partially, an uncoupling of the neuronal responses driving the primary and post-stimulus responses, explaining the uncoupling of the signals measured in the two response phases. We suggest our results are consistent with inhibitory processes driving the post-stimulus EEG and fMRI responses. We therefore propose that new methods are required to model the post-stimulus and primary responses independently, enabling separate investigation of response phases in cognitive function and neurological disease.

Keywords: Alpha; Event-related synchronisation; Oxygen metabolism; Rebound; Undershoot.

Figure 1.

Average EEG alpha power (A), BOLD (B&C) and CBF (D&E) timecourses for the flicker (green) and static (purple) visual stimuli trials. Contralateral EEG, BOLD and CBF responses (A, B &D) and ipsilateral BOLD and CBF responses (C&E) are shown. EEG measures are taken from the Hilbert envelop of the individual subject virtual electrode (VE) timecourses and averaged over trials and subjects. BOLD and CBF timecourses are taken from the individual subject ROIs (defined from BOLD T-stat maps [see Analysis section]) and averaged over trials and subjects. Error bars (B-E) denote standard error over subjects. ** denotes significant (p<0.05, paired t-test over subjects) differences in average EEG, BOLD and CBF responses between flicker and static stimuli.

Figure 2.

BOLD (A&B) and CBF (C&D) haemodynamic responses to the visual flicker stimulus; sorted according to quartiles of PERS (10.5–20s) alpha power. Contralateral BOLD and CBF responses (A&C) and ipsilateral BOLD and CBF responses (B&D) are shown. Grey bars show significant (p<0.05, repeated measures ANOVA) differences in haemodynamic response amplitude between lower (black), median (blue) and upper (red) quartiles. Error bars indicate SEM.

Figure 3.

BOLD and CBF response ratios. For each subject BOLD and CBF primary and post-stimulus responses from contralateral V1 were normalised to the primary response in that region. Mean values (magenta = primary response phase; green = post-stimulus response phase) over subjects with error bars showing standard error are plotted. The grey shaded region identifies BOLD and CBF response ratios which theoretically have the same coupling ratio as the primary positive response (n₀=0.7). This area was derived using Eq. 1 for a range of model parameters: α= −0.2 – 0.2 & β = 0.9–1.5.