PhD program alum Liberty Hamilton eavesdrops on how the human brain processes natural sounds

And same with the administrators — Kati Markowitz was our administrator for most of my grad school career. Now I realize how much she did for us on the backend (laughs). I really appreciate that now. Now I’m a faculty member, so I’m trying to figure out how we support students in these different ways. I had no idea how she did some of the stuff! And she was hilarious, which also really helped.

I was really good friends with a lot of the people at Helen Wills and I still feel like those are some of my closest friendships. That is one thing I do miss.

RH: What was your thesis about?

LH: My thesis was looking at sound representation in the auditory cortex of mice and rats. I was looking at the role of inhibitory interneurons in the encoding of simple and complex sounds. In one of my papers, we looked at recordings of extracellular potentials in the auditory cortex of mice while we stimulated parvalbumin-positive inhibitory interneurons. Then we looked at how that changed interactions between simultaneously recorded sites, and also how it changed the receptive field properties of cells.

That was a really exciting project. It was pretty early on when optogenetics was just taking off. We started that in 2008. I remember having a conversation with Shaowen about starting this project that no one in the lab was doing. He’d never tried optogenetics before, but we both agreed it was this really exciting new method, and I wanted to learn how to use it. So with the help from a lot of people from Yang Dan’s(link is external) lab and Dan Feldman’s(link is external) lab, we got that stuff up and running. Seung-Hee Lee, who was a postdoc in Yang’s lab at the time, helped me immensely in learning how to do virus injections for optogenetics. Also, the Deisseroth(link is external) lab provided some reagents and things like that. It was fun — it was this new frontier.

We ended up finding that when you stimulate these inhibitory interneurons, this increases signal to noise in the auditory cortical areas, and makes the tuning curve less noisy, so you have less background firing to random other sounds and more obvious, strong responses to pure tones that are within the receptive field.  

We also used a 16-channel probe to record from 16 sites in primary auditory cortex at the same time. It seemed that the amount of information sent from layer 4 to the more superficial layers increased with the inhibitory stimulation as well, but there was no change in within-layer dynamics. So we saw a strong feed-forward increase, and nothing within layer. This suggested that these inhibitory interneurons were mediating more of that feedforward, stimulus-driven response.

RH: Tell me about your experience as a postdoc.

LH: After my PhD I wasn’t sure exactly what I wanted to do. I thought about potentially going into industry, I thought about going into a postdoc — I was kind of on the fence for both. I ended up interviewing at a few ECoG labs, because I had interacted with a few people in Bob Knight’s(link is external) lab [who use ECoG]. I knew I was still interested in auditory neuroscience — if I were to stay in neuroscience — but I wanted the chance to work with people again.

Although the cell type manipulations I did in my dissertation were super cool, one problem with working with mice is that it’s hard to ask mice what they are thinking and to get them to do what you want them to do (laughs). I was thinking, if we could look at sound and language and speech in people, then you can more easily ask them to do a task.

I ended up having a really good conversation with Eddie Chang(link is external) at UCSF. He had been doing kind of similar work to what the Bao lab was doing in terms of looking at the encoding of sounds in the auditory cortex, but in humans instead of rodents. He and I talked about how I could use similar methods to what I’d used in my dissertation, but work with data that was from intracranial recordings in patients with epilepsy. I ended up working in his lab for three years.

We got to do some really amazing stuff with high density ECoG — so recording directly from surface of brain in patients with epilepsy. Generally, still looking at the encoding of natural sounds like speech in the temporal lobe, and in related motor areas if we were looking at speech production.

Working with these patient populations is super interesting. It’s really real. You have to talk to people who are going through something really difficult — epilepsy surgery — and as a researcher, you want to be really sensitive to that. So we felt that if [participating in the research] was something they wanted to do that could help pass the time, that could be a sort of fun way of relieving boredom in the hospital room. A lot of them are waiting to have seizures so they can be localized [in the brain], and they might be in the hospital up to a couple of weeks.

We actually made a lot of meaningful relationships with some of the people we worked with. I got to meet people I wouldn’t otherwise have been able to meet. I think collecting that data makes you realize how rare and special that is, to get to work with those types of people. So I try to be sensitive to that now in my own lab.

RH: What’s your role when working with patients?

LH: I don’t have any role in the placement of electrodes, because I’m not trained as surgeon. That’s all clinically determined. [The neurosurgeon] decides based on clinical considerations where they should put these grids or depth electrodes. If they have coverage over certain temporal areas, we’d tend to run more speech perception-type tasks, or if they have good motor coverage, maybe we’d do more speech production tasks instead.

I’ve been developing an ECoG program here at UT Austin, and I’m developing collaborations with the neurosurgeons and epileptologists at both a children’s hospital — Dell Children’s Medical Center — that has an epilepsy monitoring unit, and Dell Seton Medical Center for adult epilepsy patients.

I have been working together with the neurosurgeons and epileptologists — we’ll go to case conferences where they discuss which patients are coming in and whether or not they’ll have placement of intracranial electrodes if their epilepsy is severe enough. If there are patients that have [electrode] coverage over an area we might be able to look at for speech and language tasks, we’ll reach out to those patients and see if they are interested in participating in research during their stay.

It’s a big collaborative process between the clinicians who are making the decisions about where everything goes, and my lab where we are running specific tasks looking at perception and production of natural speech.

RH: Since ECoG is invasive, what are its advantages over other techniques to look at brain activity, such as EEG or fMRI?

LH: The huge advantage of ECoG is that you have extremely high signal to noise, and you have very high spatial and temporal resolution. With something like fMRI, it’s difficult to look at very fast processes — so things like phonetic selectivity that’s happening on the order of milliseconds is more difficult to measure using the BOLD [fMRI] signal, because that’s evolving over seconds. It’s not that you can’t do it, but with ECoG it’s much easier to see the fast fluctuations that are occurring in speech processing. Also, because the electrodes are directly on the surface of the brain, we can record signals that are more similar to extracellular invasive potentials that you could record with penetrating electrodes.

With EEG, the skull blocks out the high frequency signals that are associated with neural multiunit firing. So you can get an overall view of what’s going on but at a much wider spatial scale. [With ECoG] it’s a small signal, but with those electrodes directly on the surface of the brain, we are able to measure it and it’s a spatially localized signal.

RH: What projects are currently going on in your lab?

LH: We have a number of different projects. We’ve been building up our ECoG group at the Dell Children’s Medical Center, and something that I’m particularly interested in is the development of auditory responses in children. In my postdoc work I had looked at this particular type of response to speech, where in the temporal lobe, you see that there is this area in posterior superior temporal gyrus that specifically responds to the onsets of sentences and phrases. This is different from most of the other speech-selective areas that will respond the same, no matter where a particular speech sound is in sentence.

So if you just heard an “ah” sound you could have an “ah”-selective electrode that is always responding the same amount to “ah.” But if it’s one of these onset electrodes, you can see that it will respond to “ah” more if it occurs at the beginning of a sentence. And there are even ones that are non-selective, but just mark that something interesting has started to happen.