Award details

Non-invasive laminar electrophysiology in humans

Reference	BB/M009645/1
Principal Investigator / Supervisor	Professor Gareth Barnes
Co-Investigators / Co-Supervisors	Professor Sven Bestmann
Institution	University College London
Department	Institute of Neurology
Funding type	Research
Value (£)	345,964
Status	Completed
Type	Research Grant
Start date	01/01/2015
End date	30/09/2018
Duration	45 months

Abstract

Here we wish to study how prediction and prediction error signals propagate through the cortex. We bring to bear two streams of evidence which show that these signals can be separated both anatomically and temporally. Anatomically we know that feedback (or prediction) and feedforward (or prediction error) signals originate in the infra and supra-granular layers respectively. Temporally, we know that feedforward spectra have predominant power in the gamma band (>30Hz) whereas feedback signals have dynamics constrained to lower (<30Hz) frequency bands. We have now developed recording paradigms and precise spatial models that allow us to test whether activity most likely derives from supra-granular or infra-granular layers using non-invasive MEG recordings. Therefore for the first time we are able to make distinction between feedforward and feedback signals in the human brain both spectrally and anatomically. Using a small set of spatially and temporally distinguishable tactile and visual stimuli, within a paradigm specifically developed to study prediction error, we aim to show that we can partition feedforward and feedback signals to expected and unexpected stimuli in both frequency and space. We then aim to show how these signals can be decoded to directly read-out the stimulus expected in a particular cortical area. As MEG gives us whole-brain recordings this allows us to study feedback and feedfoward signals at multiple levels of the hierarchy simultaneously, potentially extending as far back as the thalamus.

Summary

Through experience we learn what to expect from the world around us. We become familiar with particular sensory information and we use this previous experience to make predictions about what we expect to see or touch. When the sensory information is not as we expected, this information (or prediction error) is fed forward to correct future predictions. For example it may be that you have had the impression that your stationary train is leaving the station simply because another train moves alongside you. This is an example of visual information making a prediction about the state of the world- which in this case happens to be a prediction error. We know from the anatomy of the cortex that the pathways that carry this feedback (predictions) and feedforward (prediction errors) information intertwine in parallel streams which interconnect brain regions that process very low level sensory information through multiple intermediate levels right up to those brain regions in which we make decisions about what to do. Interestingly, these pathways have distinct origins with feedforward and feedback pathways originating in the upper and lower cortical layers respectively (separated by around 3-4mm). Besides being distinguishable anatomically, these feedback and feedforward streams operate within distinct frequency ranges, the feedback signals changing more slowly (about 10-20 times a second) than the feedforward (about 30-60 times a second). At present the only way that we can look at these feedforward and feedback signals as they pass through the brain is through implanting micro-electrode arrays in the brains of animals. This is because the majority of human brain scanners either can see the layers but can only watch how they change over many seconds (functional Magnetic Resonance Imaging); or they distinguish the feedback and feedfoward signals in time but cannot resolve where they are coming from (electroencephalography or EEG). This grant builds on recent technological developments in magnetoencephalography (MEG) in which we measure magnetic fields outside the head produced by electrical currents flowing in the human brain. MEG, like EEG, can distinguish between these feedforward and feedback signals in time and frequency; importantly we have recently shown that it is also possible to distinguish between cortical layers using MEG. This is made possible because we have very precise models of where the different cortical layers lie with respect to our magnetic field measuring system (MEG). In this grant we put these two things together and expect to show that we can non-invasively disentangle feedback from feedforward information in both frequency (feedback low frequency, feedforward high frequency) and space (feedforward and feedback origins in upper and lower cortical layers respectively). This is a completely safe and non-invasive technique we can use in humans. Importantly, it will allow us to study how this feedforward and feedback information propagates across multiple areas of the human brain simultaneously - something that cannot even be done in invasive animal studies.This will not only help us understand how the brain works, but will help us understand what happens when these feedback and feedforward streams become compromised in conditions such as Parkinson's disease of schizophrenia.

Impact Summary

Who will benefit? Companies and scientists conducting animal experiments can employ the methods for the replacement, reduction and refinement in animal research and for animal welfare. Currently, work in animal models provides insights into how treatments (e.g. pharmacological, and non-invasive brain stimulation) for neurological conditions (e.g. stroke, Parkinson's disease, Alzheimer's disease, multiple sclerosis) work at a molecular or cellular level. Work in humans provides insights into the behavioural effects of these treatments. However, there is a gap between mechanistic understanding and behavioural consequences which is proving a barrier to the development of treatments. The methodological platform proposed here will provide the appropriate intermediate level of description with which to bridge this gap. For example: The imbalance between predictions and prediction error coding in the cortex is thought to underlie a number of clinical conditions such as Parkinson's disease and schizophrenia. This work will allow key mechanistic insights into prediction error coding thus providing novel routes to effective treatments. Longitudinal studies of mesoscopic intracortical interactions in human patient populations will become tractable, something that would be particularly useful in examining changes in excitatory-inhibitory balance after stroke (a novel therapeutic target in modification of potential for plasticity). The possibility of reliable non-invasive brain images will at first allow more direct targeting of neurosurgery for epilepsy influencing where electrode grids for direct cortical recording might be placed and eventually possibly eliminate them altogether. This dataset would be unprecedented internationally and open up a new tier of methodological developments exploiting whole brain human imaging at a laminar level. MEG and EEG scanner manufacturers will benefit from the increased economic value of efficient scanning methods. If we could begin tounderstand how feedback and feedforward circuits operate in the human brain, it would have a dramatic impact on computer software and hardware development fostering new generations of computing systems modelled on cortical architecture. All methodological developments at the WTCN are propagated to the community through the freely available open source software package SPM supported by Barnes and colleagues. The software is GUI based and intended for use by non-mathematical users with clinical or neuroscientific expertise. That is, these users will not need to understand how the algorithms work in order to benefit from the improved estimates of brain activity. Barnes co-organises a regular SPM course for M/EEG. The on-line material for this course (and companion fMRI course) has around 25000 visits annually. In addition there is an active mailing list with over 4000 subscribers. There are also close links with all other U.K. MEG groups and a one day training course attached to the annual conference. To inform the different user groups and beneficiaries, a public web-site will be set up containing information about the goals of the project and the latest developments. It will also allow the users to contact and interact with the team of investigators. Furthermore, the investigators plan to organize a symposium at Biomag 2016 (our major MEG conference) to further disseminate and exchange ideas. Barnes has organized and presented at such symposia before. Impact activity deliverables and milestones Public engagement via web-site (hits; feedback forms) Release of the software into SPM. Feedback from researchers and user groups. Successful workshop merging clinical, technological and scientific aspects; and reaching the relevant user groups (attendance rate, questionnaire for attendants). Future application for technology transfer funding based on project results, for example to develop head-casts in collaboration with MEG system manufacturers.

Committee	Research Committee A (Animal disease, health and welfare)
Research Topics	Neuroscience and Behaviour
Research Priority	X – Research Priority information not available
Research Initiative	X - not in an Initiative
Funding Scheme	X – not Funded via a specific Funding Scheme

I accept the terms and conditions of use (opens in new window)

export PDF file

back to list new search