ReviewWhich “neural activity” do you mean? fMRI, MEG, oscillations and neurotransmitters
Introduction
There is no doubt that functional MRI, using the endogenous BOLD contrast, has become an incredibly popular and useful tool for neuroscience that has created a remarkable body of work in just 20 years. Given the indirect, and largely unknown, coupling of the BOLD signal to the underlying neural substrate, its usefulness is even more remarkable.
The popularity of BOLD-fMRI is at least partly driven by its surprising spatial specificity. Early in the history of the technique, it soon became clear that BOLD had exquisite spatial resolution, allowing us to generate high-resolution maps of the borders between human visual areas (Engel et al., 1994, Sereno et al., 1995) in an individual. The fact that these human retinotopic maps revealed exactly the structures and organisation we expected to see from animal neurophysiology studies was a major step forward for the field. In addition, this amazing spatial specificity of the brain's haemodynamics appears to allow us to map structures right down to the columnar level of the visual cortex (Yacoub et al., 2008). Almost magically, our ability to extract spatial information may go beyond the fundamental resolution limit of the images, as small biases in the response properties of cells in each voxel may allow us to decode what information the brain is representing/processing (Kamitani and Tong, 2005).
In my opinion, although I may be biased, fMRI has been most successful in studies of human visual cortex—precisely because these studies are designed, and their results interpreted, with direct reference to previous animal neurophysiology studies. BOLD-fMRI studies can, of course, be well designed and executed without reference to previous neurophysiological research and can reveal subtle distinctions between experimental paradigms and participant groups, but the interpretation of any finding should be necessarily limited—it should always be remembered that BOLD is a measure of haemodynamic changes in the brain and these are critically dependent on the nature of the coupling between neurons and haemodynamics. Presumably, the BOLD response is related to the energy demands of modulating various aspects of neural function, including action potentials, neurotransmitter cycling and excitatory and inhibitory post-synaptic potentials, but this still a subject of much active investigation and debate (Attwell and Iadecola, 2002, Attwell and Laughlin, 2001, Mangia et al., 2009, Shulman and Rothman, 1998). In addition, it is known that hemodynamic coupling changes across the brain, across individuals, when challenged with drugs such as caffeine, with age, with disease and with subtle changes in respiration. Many of these effects can be controlled for with appropriate physiological monitoring and calibration (Iannetti and Wise, 2007).
Given all of the above, it is a shame that so many recent fMRI-BOLD studies insist on describing their measured effects as “neural activity”. Of course, we hope these effects are in some sense correlated with neural function, but we can't be sure that this is true in all cases—this is why I emphasised the link with previous animal neurophysiological work in the visual domain as it gives at least indirect evidence that our measured BOLD-fMRI findings truly reflect neural function.
As many people have pointed out, and as I emphasise in this article, the very phrase neural activity is in itself a rather poorly specified and ultimately meaningless term. In most people's minds the term is probably a surrogate for the firing of action potentials. However, within the cortex there are multiple neural signals, at different oscillatory frequencies, that might all contribute to the metabolic demand that then drives the BOLD signal. Furthermore, it's not clear which of these neural signatures are most relevant to each aspect of perception and cognition. This complexity is outlined in Fig. 1. However, all is not lost—we have several tools at our disposal that allow us to investigate which aspects of neural function contribute to the BOLD response and, with appropriate links to behavioural paradigms, which signal is most relevant to each function.
Section snippets
Firing rates, perception and oscillations
Until recently, when people thought about the neural signatures underpinning perception and cognition, there was an implicit assumption that the key measure is the firing rates of neurons. This view arose from the seminal observations that individual neurons in visual cortex were exquisitely tuned to fundamental properties of the visual scene, such as retinotopic location and stimulus orientation (Hubel and Wiesel, 1962). Surprisingly, individual firing rates can also demonstrate specificity to
The relationship between BOLD and oscillatory activity in the cortex
There are many experimental approaches that we can adopt to try and understand which aspects of neural function drive the haemodynamic response in BOLD-fMRI. These include simultaneous invasive electrode recordings and BOLD-fMRI in animals such as rat (Boorman et al., 2010) and macaque (Logothetis et al., 2001) and studies comparing electrode recordings in implanted human epilepsy patients with BOLD-fMRI in healthy human participants performing the same tasks (Mukamel et al., 2005). The results
MEG and fMRI
An alternative, non-invasive, experimental approach is to perform the same experiments in human using BOLD-fMRI and techniques such as EEG and MEG and see which electrophysiological signals seem to match the BOLD response. To a certain extent, this can be done using simultaneous EEG–fMRI, but due to the MR environment, it can be difficult to measure the full spectrum of oscillatory responses and true source-localisation, as opposed to simple temporal correlation with the BOLD signal, can be
Primary visual cortex gamma
In the 1980s and 90s investigations of the local-field potential in cat primary visual cortex revealed that presentation of simple static stimuli induced an oscillation in the visual gamma band (30–90 Hz) that was sustained throughout the presentation of the stimulus (Gray and Singer, 1989, Kayser et al., 2003). Note that the stimulus itself is not temporally varying—the oscillation, and its properties such as frequency and bandwidth, are intrinsic properties of the cortex itself and arise from
Individual variability: induced gamma as a biomarker of inhibition?
Another way of investigating the coupling between function, neural signals and the BOLD response is to study the variability of each parameter across participants. Individual variability and repeatability of the BOLD response have been studied several times (See McGonigle, in press) but relatively little work has been done on the variability/repeatability of cortical oscillatory signals. Recently, our group and others (Hoogenboom et al., 2006, Muthukumaraswamy et al., 2010) have shown that
Conclusion: The BOLD “inverse problem”
In this article, I've attempted to show how adopting a multi-modal imaging approach, using MEG, fMRI and MRS can start to help us understand some of the complexity underpinning the tri-partite relationship between human cognition, neural responses and the BOLD-fMRI signal. These non-invasive approaches are complementary to animal studies attempting to shed light on the same problems.
In some ways, the use of fMRI in cognitive and clinical neuroscience has raced ahead of our understanding of
Acknowledgments
In this paper I have reviewed several of the studies I have been involved in over the last 10 years or so. Obviously many people have contributed to these endeavours and my thanks go to all of my collaborators, in particular those at Cardiff, Aston and Nottingham Universities in the UK. The data shown in Fig. 2 was part of long-standing collaborations with Arjan Hillebrand, Paul Furlong, Ian Holliday and Gareth Barnes when we worked together at Aston University and Peter Morris and Matt Brookes
References (98)
- et al.
The neural basis of functional brain imaging signals
Trends Neurosci.
(2002) - et al.
Dorsolateral prefrontal γ-aminobutyric acid in men predicts individual differences in rash impulsivity
Biol. Psychiatry
(2011) - et al.
Individual differences in subconscious motor control predicted by GABA concentration in SMA
Current biology: CB
(2010) - et al.
GLM-beamformer method demonstrates stationary field, alpha ERD and gamma ERS co-localisation with fMRI BOLD response in visual cortex
Neuroimage
(2005) - et al.
Aging gracefully: compensatory brain activity in high-performing older adults
Neuroimage
(2002) - et al.
Dynamic statistical parametric mapping: combining fMRI and MEG for high-resolution imaging of cortical activity
Neuron
(2000) - et al.
Baseline GABA concentration and fMRI response
Neuroimage
(2010) - et al.
A framework for local cortical oscillation patterns
Trends Cogn Sci (Regul Ed)
(2011) - et al.
The temporal frequency tuning of human visual cortex investigated using synthetic aperture magnetometry
Neuroimage
(2004) A mechanism for cognitive dynamics: neuronal communication through neuronal coherence
Trends Cogn Sci (Regul Ed)
(2005)
The model- and the data-gamma
Neuron
The gamma cycle
Trends Neurosci.
Relating MEG measured motor cortical oscillations to resting γ-aminobutyric acid (GABA) concentration
Neuroimage
Localization of human somatosensory cortex using spatially filtered magnetoencephalography
Neurosci. Lett.
The missing link: analogous human and primate cortical gamma oscillations
Neuroimage
GABA(A) alpha-1 subunit mediated desynchronization of elevated low frequency oscillations alleviates specific dysfunction in stroke—a case report
Clin. Neurophysiol.
Localizing human visual gamma-band activity in frequency, time and space
Neuroimage
BOLD functional MRI in disease and pharmacological studies: room for improvement?
Magn. Reson. Imaging
Cross-frequency coupling between neuronal oscillations
Trends Cogn Sci (Regul Ed)
The spatial distribution and temporal dynamics of brain regions activated during the perception of object and non-object patterns
Neuroimage
Semantic and phonological task-set priming and stimulus processing investigated using magnetoencephalography (MEG)
Neuropsychologia
Consistent and precise localization of brain activity in human primary visual cortex by MEG and fMRI
Neuroimage
An in vivo assay of synaptic function mediating human cognition
Curr. Biol.
Spatiotemporal frequency tuning of BOLD and gamma band MEG responses compared in primary visual cortex
Neuroimage
Visual gamma oscillations and evoked responses: variability, repeatability and structural MRI correlates
Neuroimage
New vistas for alpha-frequency band oscillations
Trends Neurosci.
Visual word recognition: the first half second
Neuroimage
Event-related EEG/MEG synchronization and desynchronization: basic principles
Clin. Neurophysiol.
In vivo magnetic resonance spectroscopy of GABA: A methodological review
Progress in Nuclear Magnetic Resonance Spectroscopy
Differences in gamma frequencies across visual cortex restrict their possible use in computation
Neuron
Group imaging of task-related changes in cortical synchronisation using nonparametric permutation testing
Neuroimage
Task-related changes in cortical synchronization are spatially coincident with the hemodynamic response
Neuroimage
The role of GABA in human motor learning
Curr. Biol.
Interpreting PET and fMRI measures of functional neural activity: the effects of synaptic inhibition on cortical activation in human imaging studies
Brain Res. Bull.
Signal processing in magnetoencephalography
Methods
Relating BOLD fMRI and neural oscillations through convolution and optimal linear weighting
Neuroimage
Induced Gamma activity in primary visual cortex is related to luminance and not color contrast: an MEG study
J. Vis.
Induced visual illusions and gamma oscillations in human primary visual cortex
Eur. J. Neurosci.
Spatiotemporal activity of a cortical network for processing visual motion revealed by MEG and fMRI
J. Neurophysiol.
An energy budget for signaling in the grey matter of the brain
J. Cereb. Blood Flow Metab.
On the electroencephalogram of man
Electroencephalogr. Clin. Neurophysiol.
Negative blood oxygen level dependence in the rat: a model for investigating the role of suppression in neurovascular coupling
J. Neurosci.
What determines the frequency of fast network oscillations with irregular neural discharges?
I. Synaptic dynamics and excitation-inhibition balance. Journal of Neurophysiology
Synchronization of visual responses between the cortex, lateral geniculate nucleus, and retina in the anesthetized cat
J. Neurosci.
Elevated endogenous GABA level correlates with decreased fMRI signals in the rat brain during acute inhibition of GABA transaminase
J. Neurosci. Res.
Frequency of gamma oscillations routes flow of information in the hippocampus
Nature
Interactions between posterior gamma and frontal alpha/beta oscillations during imagined actions
Front. Hum. Neurosci.
Spatial effects in the detection of gamma-aminobutyric acid: improved sensitivity at high fields using inner volume saturation
Magn. Reson. Med.
Cited by (110)
A functional account of stimulation-based aerobic glycolysis and its role in interpreting BOLD signal intensity increases in neuroimaging experiments
2023, Neuroscience and Biobehavioral ReviewsLearning without contingencies: A loss of synergy between memory and reward circuits in schizophrenia
2023, Schizophrenia ResearchA unified view on beamformers for M/EEG source reconstruction
2022, NeuroImage