Brain Rhythms: Functional Brain Networks Mediated by Oscillatory Neural Coupling

May 6, 2019

hi my name is Philip Gilley and I'm an assistant professor of audiology and the principal investigator of the neuro dynamics laboratory at the University of Colorado at Boulder this talk was first given in March 2014 at the ignite s LHS series sponsored by the Graduate Advisory Board I would certainly like to thank the board for that opportunity today I would like to share with you some of my thoughts on functional brain networks that are mediated by oscillatory neural coupling and how brain rhythms generated by these networks underlie the physiology and pathophysiology of communication understanding how the brain facilitates communication including speech language and hearing reading thinking and expressing emotion is integral to understanding human behavior and is therefore integral to shaping behavior and to successful remediation of disorders that affect communication when information is transferred across the synapse between two neurons extracellular currents are generated at the dendrites of the receiving neurons these neuronal currents then spread through the brain to the scalp where they can be monitored by a recording electrode when clusters of neurons fire synchronously in response to a stimulus the resulting change in the EEG voltage called an evoked potential can be observed as a peak in the EEG waveform sometimes whether occurring spontaneously or evoked by a stimulus clusters of neurons fire synchronously and rhythmically over time the EEG patterns observed from this activity represent a quasi periodic wave function which reflects the mass action of a neuronal cluster all told the EEG takes an oscillatory like pattern however when groups of neurons fire in such a pattern a signal is transmitted to other neuronal clusters that are connected both anatomically and functionally activity from a group of cells at one point in time that triggers activity in another group of cells at a later point in time appears as an oscillation in the EEG such that the orientation or polarity of the peaks reflect the physical orientation of the neurons that generated that response from this we can infer that oscillatory EEG might represent the functional connections between different brain regions a well-defined example of such functional connectivity can be observed in a series of responses called the auditory evoked potentials which are Peaks that appear in the EEG when hearing a sound these Peaks can be defined by their orientation their amplitude and their peak latency that is when they occur relative to the onset of a sound they can also be described by their oscillatory period or frequency of the peak when processing a sound energy transduced by the cochlea generates action potentials at the auditory nerve which leads to peak activity in the cochlear nucleus then the superior olivary complex the nucleus of the lateral lemniscus and the inferior colliculus each of these contribute two peaks in a waveform known as the auditory brainstem response which is shown here on the left from the brainstem information is then transmitted to the auditory thalamus and cortex which contribute to the middle latency and cortical responses shown here on the left as well finally higher-order cortical regions contribute to the so-called late potentials which reflect cognition and executive function taken together this pattern of sequential processing represents several hierarchies of functional brain activity there is an axial spacial hierarchy such that incoming information enters the nervous system at lower regions and higher-level processing takes place in higher brain regions neuronal circuits at the lower level of this axial hierarchy generate oscillatory activity that is relatively higher in frequency and conversely higher-level processes generate low frequency oscillations when detecting a stimulus peak changes in high frequency oscillatory activity occur earlier than changes in low frequency activity these hierarchies reflect a rather traditional account of bottom-up processing and we can further exploit these relationships to better understand the underlying brain activity in order to further explore these functional networks we recorded EEG from 64 channels on the scalp in a group of young adults during an auditory entrainment task a speech sound was repeated at a regular and steady rate while participants were asked to passively listen to the sounds in this way rhythmic oscillatory activity was entrained to the repetition rate of the sounds we then performed a high-resolution spectral temporal analysis of the EEG waveforms from each channel this separates the signal into a series of frequency specific bands that can be represented in a three dimensional time frequency plane in much the same way that speech is often represented across time and frequency in a speech spectrogram essentially this is a spectrogram of the EEG activity called a Skellig Ram or an event related spectral perturbation in this group average response we see time represented along the abscissa or x-axis and frequency along the ordinate axis magnitude is represented by color the onset of the speech sound is marked by the arrow and here we can also see a representation of the spectral temporal hierarchy of processing as shown by this thick blue line with this data we can also take the mean across either dimension to find the mean temporal envelope and the mean spectral envelope for each response these envelopes provide information about where in the time frequency plane we can filter the responses we then identified these spectral Peaks from each participant and then clustered all of the participants peak frequencies by commonly observed spectral Peaks this clustering revealed 15 common Peaks within seven frequency bands from low frequency to high frequency these bands are the Delta band the theta band the alpha band beta 1 beta 2 gamma 1 and gamma 2 at the highest end from each spectral peak an independent components analysis or ICA is used to separate these spatial distributions of that activity and its contribution to the global EEG activity these spatial maps are then used to localize the brain sources separately for each peak using a current density reconstruction algorithm to visualize these complex sources each source will be represented in six different views of the brain two views each of the left and right cortical surfaces a view of the top of the brain in the upper middle panel and a view of the front of the brain in the lower middle panel the brain activity shown here highlights the oscillatory activity of these underlying functional networks the color of activity in the brain corresponds with the color shown in the CDR spectrum in the lower right corner where the height of each bar represents the relative contribution of that frequency at a given point in time to examine the contributions of different networks we can then adjust the display filter to only show activity from a specified frequency band for example here we see the components in the Delta band now these likely act is a central entrainment mechanism for binding the active networks this Delta activity includes activity in the midbrain and limbic structures including the thalamus and then extends into regions of medial and inferior temporal cortex more specifically the bursts of Delta activity represented in the peaks of the temporal envelope the lower left portion of the figure occur when maximal activity is near the thalamus the Alpha activity seen here appears to volley in a left-to-right fashion and including through the dorsal portion of the cingulate cortex the preak unius and the auditory cortex now there are several hypotheses about function of alpha activity but many leading accounts suggest alpha as a reflection of allocating attention resources during a process here we see the peak alpha activity occurring in auditory regions of the temporal cortex the beta one activity seen here also volleys in a left-to-right fashion including in the thalamus the medial and inferior temporal lobes the auditory cortex and the parietal cortex specifically the bursts of activity between about 50 and 150 milliseconds occur in the auditory cortex the gamma1 activity which may relate to feature extraction and selective attention includes activity in the cingulate cortex the auditory cortex and the superior temporal lobes and some activity in the cerebellum here again we see the source reconstructions with all of the bands displayed the source models indicate that auditory processing is mediated by several functional brain networks that extend beyond the classical ascending auditory pathways shown earlier this includes activity in the limbic system with integrative and cognitive networks and even motor networks which include the cerebellum recall that these oscillatory frequencies are related to the sequential order of bottom-up processing here again we see the nuclei of the a sending auditory system while it is relatively straightforward to see how each of these regions can affect each other during processing less transparent is how other brain processes affect functions in these auditory regions to address this issue I conducted a meta-analysis of over 450 candidate connections between 128 distinct regions in the central nervous system and mapped these connections in an organizational structure that helps to describe the dynamical relationships between different areas in the nervous system this organization groups brain regions with common functions such as the visual system auditory and motor systems and others as shown here by this organizational map brain regions are also organized by axial hierarchy such that the head neck and body are near the bottom of the map and higher-level executive cortex is at the top finally the map is organized by five dynamical systems including transduction fats encoding which is the encoding of information related to frequency amplitude time and space association in integration functions executive functions and modulatory functions including chemical and visceral modulation together this map represents a basis for connectivity and modulation between different functional brain networks further this model of brain processing may be clinically useful for the differential diagnosis and treatment of disorders that affect different sensory systems including the auditory system for example if we can identify where in the auditory network an impairment is occurring then treatment can be targeted at that impaired process one example of this application is the diagnosis and treatment of child language learning problems in 2007 madula Sharma and her colleagues in Sydney showed that over 75% of children with language learning problems could be clinically diagnosed with more than one impairment depending on which clinical battery was used for the assessment so if we can identify specific brain processes that underlie each of these behavioural impairments then we can improve differential diagnosis and targeted treatment for this population of children we compared time frequency responses in typically developing children and in children with language learning problems those responses showed an atypical spectral temporal hierarchy in children with learning problems further when the timing of the stimuli were randomized the spectral power of the beta and alpha activity were significantly diminished in children with learning problems an analysis of the frequency coherence spectrum shows that children with listening problems had two different beta Peaks compared to their normal hearing peers which may suggest ad synchronization or a decoupling of oscillatory networks in these children in another study we compared responses in typically hearing children and in children with listening problems both with and without a clinically diagnosed auditory processing disorder responses in quiet revealed the expected spectral temporal hierarchies but with diminished beta and alpha responses when the sending and background noise interestingly when comparing the peak frequencies of each detected band children with auditory processing disorders showed a lower frequency in the beta band compared to the other groups this further corroborates the notion that beta decoupling is reflected by poor auditory processing here we show the peak frequencies of each response like beads on a string along an imaginary x-axis with low frequencies in the Delta band on the left to higher frequencies in the gamma band on the right in the case of auditory processing disorders we might think of this beta band as being knocked off the string a further analysis of these peak frequencies from all of the children revealed a very interesting trend note that we also have an imaginary y-axis here the y-axis also represents frequency but these frequencies are harmonics of the stimulus rate note that each peak frequency was very near a harmonic of the stimulus rate and also for you math geeks you might notice that these harmonics start to approximate a Lucas sequence fun stuff so taken together it is possible that different regions of the brain connects to a functional network that is a network that is actively engaged in a common goal by changing or modulating the operating frequencies to a harmonic within that network in this way functional networks become harmonically coupled this mechanism of network binding could have implications for understanding modulatory functions such as selective attention for example a harmonic network might be established for extracting information from a target talker so that each component of the harmonic network contributes to the behavioral goal that is understanding what the talker is saying if a second harmonic network is established from a competing talker then attention resources can be allocated to the network receiving the relevant information and then suppressing information from the competing talker this mechanism might also explain the observation of multiple beta Peaks in children with auditory processing disorders as if the beta oscillators were somehow unable to couple with the correct network overall this hierarchical model of processing and the model of harmonic oscillatory coupling provide fascinating new insights into dynamical brain networks and their implications for clinical application we hope that our future research further brings to light the nature of these brain rhythms thank you very much you

You Might Also Like


  • Reply Michael Nguyen May 6, 2019 at 5:41 am

    Where can we see the connectivity map?

  • Reply Harish Gunasekaran May 6, 2019 at 5:41 am

    Brilliant with a clear cut illustration! Thank you, Prof Gilley!

  • Reply Thork Rynu May 6, 2019 at 5:41 am

    I'll have to watch again but how do you model lower and midbrain brain activity from EEG? Is that from correlating from other imaging methods? Nice video would love more especially on attention / filtering. Your link in the description is forbidden btw

  • Reply Nicola Rosti May 6, 2019 at 5:41 am

    Thanks for this video! It seems you've discovered how music works inside the brain…or how the brain works like a musician

  • Reply Antonia H May 6, 2019 at 5:41 am

    when you are presenting the coherence spectrum of children with learning problems (LLP) I think you make a mistake calling them children with listening problems at 13:35 don't you? Otherwise I find it a little bit confusing because in the next experiment listening problem is shortened with LP..

  • Reply Ifat Glassman May 6, 2019 at 5:41 am

    I wonder if these frequencies form a kind of constructive or destructive wave interference (if they harmonize or not, temporally). Maybe those neurons that form electrically constructive wave interference are able to "join in" with the conscious experience (while those that are left out of the harmony remain unconscious).

  • Reply Frank Sun May 6, 2019 at 5:41 am

    Help me a lot in understanding what EEG is capable of. Thanks for this awesome video!

  • Reply Fundemental Roast May 6, 2019 at 5:41 am

    This was awesome. Thank you.

  • Reply kustomweb May 6, 2019 at 5:41 am

    Best of YouTube. Bravo.

  • Reply Nick Stachowski May 6, 2019 at 5:41 am

    Still a great video. Got anything else in the works?

  • Reply Samira Khaled May 6, 2019 at 5:41 am

    Outstanding ! very concise explanation thank you !

  • Reply bahareh elahian May 6, 2019 at 5:41 am

    Thanks for sharing this great presentation. Is there any way we can access to power point (Absolutely with copy right for the author)?

  • Reply DistortedFaiths May 6, 2019 at 5:41 am

    This video is amazing. You explain complex concepts from research so eleagantly, which gives insight into what actual experimental research is being done and have wonderful techniques to understanding how cognition maps onto the brain.

  • Reply Ryan Young May 6, 2019 at 5:41 am

    What software is used to make these figures?

  • Reply Kyle Olin May 6, 2019 at 5:41 am

    Very interesting! I have very minimal knowledge of how the brain works, but found this video to be very informative. I actually got quite alot out of it. Thank you

  • Reply Cyrus Graham May 6, 2019 at 5:41 am

    9-14-2004 connecticut

  • Reply Cyrus Graham May 6, 2019 at 5:41 am


  • Reply purplebluered100 May 6, 2019 at 5:41 am


  • Reply Miguel Ibaceta May 6, 2019 at 5:41 am

    Thanks very much! beautiful work!

  • Reply Atom Nous May 6, 2019 at 5:41 am

    Does this take into account individual differences in the pattern? All brain research seems to assume that the mean score is the true pattern. What if there are two or three consistent patterns within different individuals while doing similar tasks?

  • Reply Cyrus Graham May 6, 2019 at 5:41 am

    FYI: September 14th, 2004- February 2005, evidently there is someone else name Cyrus a white male adult CYRUS .C last name beginning with the letter C looks like it might be polish, just happened to see some pamphlet that had blurred photo of some dude name Cyrus,C ( I can't pronounce the last name)

  • Reply Adam Kantorík May 6, 2019 at 5:41 am

    I don't know what I just watched, I have no idea why I watched it,.. But good video 😀

  • Leave a Reply