Elsevier

Magnetic Resonance Imaging

Volume 28, Issue 8, October 2010, Pages 1087-1094
Magnetic Resonance Imaging

Research article
Coupling of neural activity and fMRI-BOLD in the motion area MT

https://doi.org/10.1016/j.mri.2009.12.028Get rights and content

Abstract

The fMRI-BOLD contrast is widely used to study the neural basis of sensory perception and cognition. This signal, however, reflects neural activity only indirectly, and the detailed mechanisms of neurovascular coupling and the neurophysiological correlates of the BOLD signal remain debated. Here we investigate the coupling of BOLD and electrophysiological signals in the motion area MT of the macaque monkey by simultaneously recording both signals. Our results demonstrate that a prominent neuronal response property of area MT, so-called motion opponency, can be used to induce dissociations of BOLD and neuronal firing. During the presentation of a stimulus optimally driving the local neurons, both field potentials [local field potentials (LFPs)] and spiking activity [multi-unit activity (MUA)] correlated with the BOLD signal. When introducing the motion opponency stimulus, however, correlations of MUA with BOLD were much reduced, and LFPs were a much better predictor of the BOLD signal than MUA. In addition, for a subset of recording sites we found positive BOLD and LFP responses in the presence of decreases in MUA, regardless of the stimulus used. Together, these results demonstrate that correlations between BOLD and MUA are dependent on the particular site and stimulus paradigm, and foster the notion that the fMRI-BOLD signal reflects local dendrosomatic processing and synaptic activity rather than principal neuron spiking responses.

Introduction

Blood oxygenation level-dependent (BOLD) imaging methods are ubiquitously used to study brain function. Yet, while the fMRI-BOLD signal reflects changes in brain perfusion in response to changes in neuronal activity, the exact mechanisms behind this coupling, and the electrophysiological signals that best correlate with the BOLD signal, remain debated [1], [2], [3], [4]. BOLD activations have been reported to correlate with neuronal spiking activity [5], [6], with evoked potentials [7] as well as with synchronous subthreshold activity such as local field potentials (LFPs) [8], [9], [10], [11], [12], [13]. As a result, the BOLD signal may be sensitive not only to the strength but also to the timing of neuronal activity, and not only to spiking responses but also to local synaptic activity [1], [2], [14].

To investigate the coupling of hemodynamic and neuronal responses, we previously developed an experimental setup that allows the use of microelectrodes to directly measure electrophysiological activity in the brain of nonhuman primates inside the MR system [10], [15]. Building on such simultaneous recordings, previous studies compared the visual responses of BOLD and neural signals in primary visual cortex (V1) [10], [11]. These studies not only revealed that increases in BOLD correlate more strongly with increases in high-frequency field potentials (LFPs) than with increases in multi-unit spiking activity (MUA), but also reported cases of signal dissociation: for a subset of recording sites, positive BOLD responses around the microelectrode (and increases in LFPs) occurred in the absence of measurable increases in MUA. Together with pharmacologically induced dissociations of BOLD and MUA in the same system [16], this demonstrates that BOLD responses can occur in the absence of MUA activity and hence suggests that the BOLD signal likely reflects increases in subthreshold processing and synaptic afferents rather than the local network output [2], [14].

Here we use the motion-sensitive area of the monkey (area MT) as model system to demonstrate the ubiquity of such dissociations between visual responses in BOLD and neuronal firing. For these experiments, we exploited a property of MT neurons known as motion opponency [17], [18], [19], [20]: a neuron's response to a stimulus of the preferred motion direction is considerably reduced when a second stimulus, containing the opposite direction of motion, is superimposed onto the preferred stimulus (Fig. 1A). This response suppression is not present in V1, which provides a large proportion of synaptic inputs to MT, and reflects the sensitivity of MT neurons to global stimulus motion [17], [21]. For those sites in MT where MUA activity showed strong motion suppression, we found that BOLD responses were only moderately affected by the opponency stimulus and, hence, persisted in the absence of MUA responses. In addition, and in agreement with studies in V1, we found that a subset of recording sites show a general dissociation of BOLD and MUA regardless of stimulus, in the sense that increases in the BOLD signal were accompanied by systematic decreases in MUA.

Section snippets

Materials and methods

Data was obtained from two adult macaque monkeys (Macaca mulatta) using previously established protocols [10], [22]. All procedures were approved by the local authorities (Regierungspräsidium) and were in full compliance with the guidelines of the European Community (EUVD 86/609/EEC) for the care and use of laboratory animals.

Results

We recorded combined fMRI-BOLD and electrophysiological responses from two animals during full-field visual stimulation with drifting sine-wave gratings (Fig. 1A). The stimulus paradigm consisted of three stimuli: two unidirectional gratings moving in opposite directions, and their superposition (the motion opponency stimulus). The drift direction of these stimuli was optimized for each recording site, such that one of the unidirectional gratings was drifting in the preferred direction of local

Discussion

Our results support previous studies which found increases in the fMRI-BOLD signal in the absence of increases in principal neuron spiking [10], [12], [13], [16], [23] and studies reporting stronger correlations of the BOLD signal with field potentials than with MUA [8], [9], [11]. Field potentials reflect synchronized dendrosomatic potentials and voltage-dependent membrane oscillations [24], [25] and thus likely reflect postsynaptic consequences of presynaptic input, including afferent inputs,

Acknowledgments

We thank Jozien Goense for providing expertise in designing the MRI experiments and Mirko Lindig and Denise Ipek for technical assistance during their conduction.

References (28)

  • KayserC. et al.

    A comparison of hemodynamic and neural responses in cat visual cortex using complex stimuli

    Cereb Cortex

    (2004)
  • NiessingJ. et al.

    Hemodynamic signals correlate tightly with synchronized gamma oscillations

    Science

    (2005)
  • LogothetisN.K. et al.

    Neurophysiological investigation of the basis of the fMRI signal

    Nature

    (2001)
  • CaesarK. et al.

    Dissociation of spikes, synaptic activity, and activity-dependent increments in rat cerebellar blood flow by tonic synaptic inhibition

    Proc Natl Acad Sci U S A

    (2003)
  • Cited by (0)

    This study was supported by the Max Planck Society.

    View full text