Caveat lector: This blog is where I try out new ideas. I will often be wrong, but that's the point.

Home | Personal | Entertainment | Professional | Publications | Blog

Search Archive


Voytek Frontiers in Human Neuroscience paper: "Shifts in gamma phase-amplitude coupling frequency from theta to alpha over posterior cortex during visual tasks"

This post is about my latest paper published in the (open access!) journal Frontiers in Human Neuroscience. This paper was actually an invited submission (even though it was invited, it was still peer-reviewed!) It was part of a special topic issue, "Origins and consequences of rhythmic cortical activity".

The Frontiers journals are, simply put, amazing. For those of you unfamiliar with the peer-review process, Penny Arcade sums up the experience pretty nicely (satirized here).

In all seriousness, here's how it works:

1. As a reviewer you often see the names of every author of the paper you're reviewing, but you never get to know who your reviewers are. The counter argument against changing this and making the system double-blind is often that, "people can often tell who wrote a paper anyway based upon the content, methods, etc."

To deal with this, the Frontiers journals start off single-blind, but after all the reviews are completed the reviewers are unblinded and everyone knows who everyone is. This has lead to substantially nicer, more helpful reviews, in my opinion. All reviewer names are published along with the paper which means that the reviewers are held somewhat publicly responsible as well.

2. Speed and nature of communication. Reviewers often take weeks or months to review a paper. And then the author takes several weeks to respond. And then the reviewers go back and review the responses, etc. This can lead to a several-months long review process. Again, the Frontiers journals have addressed this nicely. After the first round of reviews the editor initiates an interactive online forum where the editor, reviewers, and authors can interact at a more rapid pace.

3. Editors and "novelty". Often papers get rejected before ever being reviewed because an editor deems the paper to be not "novel" enough to warrant publication in their journal. This tends to be a problem for the "high impact" journals. The Public Library of Science has addressed this by introducing PLoS ONE, which publishes nearly any paper deemed scientifically and methodologically sound, regardless of "novelty".

So the review process was actually quite interesting and quick. It's nice to see some publishers embracing technology a bit and allowing for rapid, forum-style communication between the authors and reviewers.

As for the paper itself, the idea grew out of a pretty simple follow-up based on an awesome paper by my friend, colleague, groomsman, and co-author, Ryan Canolty. In 2006, Ryan published a paper in Science: "High gamma power is phase-locked to theta oscillations in human neocortex". As the title implies, they found that oscillations in the human neocortex form nested rhythms across frequency bands. They showed that the phase (how "peak-like" or "trough-like" the sinusoid is) of low frequency "theta band" activity (4-8Hz) modulates the amplitude of high frequency "gamma band" activity (80-150Hz).

Bradley Voytek Frontiers in Human Neuroscience 2010 Shifts in gamma phase-amplitude coupling frequency from theta to alpha over posterior cortex during visual tasks

More simply: when the theta wave is at its lowest point, the trough, power in the gamma band is highest. You can see a toy example of this in the image above.

Great! But why do we care?

First, gamma band power correlates with single-unit (neuronal) spiking activity correlates with fMRI BOLD signal. That is, all of these different signals that we measure might be a way at getting at neuronal activity more directly.

Second, oscillations in low frequency rhythms are probably reflecting (sub-threshold) changes in the extracellular membrane potential. For neurons to "fire" an action potential, ion channels in the cells themselves must open to allow ions (and thus charge) to flow.

Third, low frequency oscillations may help coordinate long-distance communication between brain regions by "shaping" which neuron groups are more likely to respond to a stimulus by biasing the statistical probability of action potentials occurring. Thus, these nested brain rhythms might reflect a mechanism of connecting single-unit activity with huge brain networks, and may reflect the way that the brain "works".

That was really dense. Let's unpack that.

We don't know how different brain areas communicate to give rise to cognition. There's a complicated code that we don't understand. This nested oscillations idea might connect the really low-level physiology of the brain with high-level cognition that requires communication between a lot of brain regions. And it ties it all nicely together into a cool communication system where different low frequencies could act as "switches" to bias information flow between brain regions.

I've talked about oscillations here before: in my post about the paper "Endogenous Electric Fields May Guide Neocortical Network Activity", in my post on neuroimaging, "What can we measure using neuroimaging techniques?", and in the post about my paper in the Journal of Cognitive Neuroscience, "Hemicraniectomy: A new model for human electrophysiology with high spatio-temporal resolution".

In this paper we recorded data from two human patients with implanted subdural electrodes. This technique—known alternately as "electrocorticography" (ECoG), "intracranial EEG" (iEEG), or "intra-cranial electrophysiology" (ICE)—is a surgical procedure done as a treatment for (usually) epilepsy. I've talked about this stuff before (see the above links); it's a staple of my research. (For a more detailed explanation as to why someone might get electrodes surgically implanted into their brains, check out this part of one of my talks).

The first step was to recreate Ryan's findings that gamma amplitude couples to theta phase.

Bradley Voytek Frontiers in Human Neuroscience 2010 Shifts in gamma phase-amplitude coupling frequency from theta to alpha over posterior cortex during visual tasks. Theta gamma phase-amplitude cross-frequency coupling

When we recreate the conditions of Ryan's experiments (auditory tasks, frontal electrodes) we see really nice theta/gamma coupling, as can be seen in the image above. When the subjects are performing non-visual tasks, theta/gamma coupling is strong across most electrodes. The more red the electrode, the strong the theta/gamma coupling. In the comodulogram (colorful thingy on the left) you can see the average theta wave in the specific highlighted electrode. You can also see the red stripes above it that occur during the trough of the theta. The more red those stripes, the higher the gamma amplitude. Nice coupling.

Bradley Voytek Frontiers in Human Neuroscience 2010 Shifts in gamma phase-amplitude coupling frequency from theta to alpha over posterior cortex during visual tasks. Alpha gamma phase-amplitude cross-frequency coupling

Now, in contrast, when the subjects perform visual tasks, we see that at electrodes over the posterior (visual) parts of the brain begin to exhibit coupling between gamma power and a different low-frequency band; alpha. Alpha is a "visual" brain rhythm that is strongly modulated by visual attention. When subjects are visually engaged, we find that phase-amplitude coupling over the posterior cortex shifts to an alpha/gamma pairing.

This paper is the first time anyone has shown that the phase frequency in phase-amplitude coupling is selectively modulated by behavioral state.

Good times!

This work was financially supported by the (sadly defunct) American Psychological Association Diversity Program in Neuroscience grant 5-T32-MH18882 (to B.V.) and the National Institute of Neurological Disorders and Stroke grants NS21135, NS21135-22S1, and PO40813 (to B.V. and R.T.K.).

Voytek B, Canolty RT, Shestyuk A, Crone N, Parvizi J, and Knight RT (2010). Shifts in gamma phase-amplitude coupling frequency from theta to alpha over posterior cortex during visual tasks. Front Hum Neurosci.


  1. Can I ask some questions about this paper via an email?

  2. not sure if this is already known fact, but could this theta-gamma coupling be related to the presence of low-freq bands during SWS and the known role of sleep in cortical plasticity. Could the slow waves in sleep be a mechanism for altering these high level global brain networks by biases which areas communicate more/less efficiently?
    great read, thanks so much for this

  3. AdamK: Interesting you should bring that up... I've been looking into how theta/gamma coupling modulated over time. Hopefully you'll hear something more on this blog about it in the future!