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



What's in a signal? Self-paralysis for neuroscience!

It's been a while since I've had a "fun" post because I've been so busy with work stuff. There was my Japan trip, zomBcon, and I'll be giving a talk at the annual Society for Neuroscience conference next week. In between I've been writing a lot about my recent work and associated press.

Time to tone down the narcissism and approval-seeking and talk about some Crazy Science History!

I've already talked about Henry Head's self-mutilation and penis experimentation as well as Brown-Séquard sperm injections. As some of you can tell, I'm fascinated by this kind of work. So when my PhD advisor pointed out this paper to me, I knew I had to cover it here.

The paper title seems tame enough: "Nature of average evoked potentials to sound and other stimuli in man" by Reginald Bickford, James Jacobson, and D. Thane R. Cody from 1964. Sounds innocuous, but wow it's werid!

But first: context!

As I've mentioned, scalp EEG has some pretty serious limitations, even though it's my technique of choice. EEG has a long history in cognitive neuroscience but it's not without some controversy, still.

Within the EEG community, there's a now classic paper (and somewhat controversial in its own right) by Yuval-Greenberg out of the lab of my colleague Leon Deouell titled, "Transient induced gamma-band response in EEG as a manifestation of miniature saccades". As I said in my earlier post on EEG "...the EEG signal is (generally) dominated by surface cortical signals." Note the parenthetical. Generally. Turns out, we might potentially be confounding a lot of "brain" signal with what's really some other physiological signal not coming from the brain, which Yuval-Greenberg and colleagues elegantly demonstrated in their work.

By the way, all imaging techniques are susceptible to these non-brain physiological confounds, which is one reason I work with stroke patients: it certainly provides stronger causal brain/behavior evidence than imaging methods.

Anyone who's worked with EEG knows that signal artifacts from eye movements are a major potential source of noise. So is muscle activity. In fact, because I was so curious about my method of choice, I worked hard to quantify some of these artifacts in my hemicraniectomy paper.

Try this: with your finger, touch the very front top of your ear, then move your finger directly from there to your scalp. Now move straight up about 1-2 inches from there, press your finger hard against your head, and then bite down.

Feel that? That's your temporalis muscle. It's big.

Bradley Voytek

Electrical activity from the muscles can be picked up quite easily. This is a technique known as electromyography. It's pretty straightforward: your muscles contract because of electrical signals, and those signals can be detected. Similarly, using a technique called electroretinography you can record signals from your retina.

What does this mean for EEG? Well, muscle and eye activity is very strong compared to brain activity recorded from the scalp. This is why electrocorticography is so nice, and why at-home EEG devices (that shall not be named here) that get a lot of buzz for being "brain-controlled" are so fraught with issues.

It turns out that, if you're running an EEG experiment, and you get a nice big "brain" effect, what you might actually be seeing is an artifact of muscle or eye activity. Think about it this way: let's say you're running an experiment to look at the neural basis of attention. To do this you play a bunch of auditory pure tones, and 10% of the time the tone is louder and the subject is supposed to respond. When you play those louder tones, you see a big EEG response. Well, what if the loud tones cause the person to flinch or blink ever so slightly? Because of the huge amplification of the EEG signals in order to detect brain activity, these slight contractions look really big, and it would appear that you have just seen a big neural response, when in reality the person might just be flinching!

How do you disentangle these phenomena!? (What will really twist your noodle is that the flinching is also controlled by the brain!)

This is where we come to the Bickford paper (at last!)

What did Bickford do? He had his colleagues inject him with the neurotoxin curare, commonly known for its use in arrows by South American hunters.

Curare is a strong muscle relaxant. So strong, that at the doses used in the experiment, Bickford had to be placed on an artificial respirator, or else he would have asphyxiated due to the paralysis of his breathing muscles!

What they found was that the EEG responses they had been finding in response to the auditory tones were completely abolished once all of Bickford's muscles had been paralyzed. They interpreted this as demonstrating that the auditory tone responses seen in EEG in this case aren't due to brain activity, but rather due to muscle activity from slight head movements in response to the tones.

This is not to say that auditory tones don't have a neural component, of course, just that a researcher must take care of confounds when conducting experiments.

It's amazing that this kind of work is still going on. Here's a paper from an Australian group from 2008 wherein they do a very similar design, going so far as to paralyze four of their subjects! Thinking activates EMG in scalp electrical recordings. (It's unclear whether the authors were the paralyzed subjects in this experiment).

As an aside, my academic grand-father, Robert Galambos, was very involved in this kind of auditory work. His research proved that bats use hearing to echolocate, and lead to the development of the auditory brainstem response exam that's given to almost every newborn. In fact... he deserves his own post....