What’s it like to be a brain?

Written by . Filed under nostalgia, shop talk. Tagged , , , , . Bookmark the Permalink. Post a Comment. Leave a Trackback URL.

I started reading Elizabeth Costello because the premise was irresistible to me: an academic who, when invited to give talks about her literary work, gives herself over to impassioned sermons on vegetarianism. I have often had the impulse, just before giving a talk, to just chuck it and talk about something else entirely. Perhaps an extended tirade about the evils of for-profit academic publishing, or an idea for a Star Wars sequel. Really, anything but the work I’ve been invited to present.

I don’t think I realized going in just how serious Coetzee was about vegetarianism, or just how much of the book would be taken up by these sermons. I nearly quit reading several times, but kept going out of a combination of stubbornness and willful procrastination. I had reached a stage in the project I was working on — a study of brain responses related to speech sound categorization — where it was completely unclear what to do next. Thinking about the data kept taking me in circles, so I needed to try something different. I tried not thinking about the data. Archimedes taking a bath, etc. So, sitting in a park a few blocks from my office, I subjected myself to the judgment of Coetzee’s hypothetical future generations, who, he argues, would see not only the omnivores among us, but those of us who fail to stand up and stop the slaughter of innocent animals for food as little different from those complicit in genocides throughout human history.

It’s important to develop a sensitivity for the beauty and power of arguments that nonetheless fail to move you.

Coetzee forced me to deal with an uncomfortable truth: I am totally OK with the idea that pigs, for example, can be clever, loyal, sensitive and delicious. He also pushed me to defend that position, and in another fit of procrastination, I would stumble upon Steven Davis’s take on the “least harm” principle. Davis points out that, in terms of the overall amount of death and suffering per calorie, slaughtering a pasture-raised cow and eating its flesh is probably a better deal for the animal kingdom than clearing a field for soybean cultivation and keeping the plants free of pests until harvest time — to say nothing of the collateral damage incurred during the actual harvest. Every step of that process involves killing insects and at least displacing (but usually killing) rodents, no matter how carefully it is done. Just because vegans don’t eat the meat, milk, or eggs of those animals doesn’t make them any less complicit in the killing. Why draw a line between cows, chickens and pigs, whose rights must be protected, and rodents and insects, whom we can displace or slaughter with impunity when they threaten the plant-based crops we have destroyed their habitats to plant?

If I am honest with myself, I mostly like this argument because I enjoy few foods more than a good steak, generously salted, peppered and grilled on an iron skillet. But I also think it treats our relationship with nature more holistically and realistically than the alternatives. If you’re unwilling to countenance any animal death or mistreatment in the production of your food, you owe it to the rodents and insects of the world to eat only those foods that have been grown on small, unintrusive plots of land, untreated with pesticides and freely shared with the surrounding fauna. But that’s not a sustainable way of producing food in most places. Better to consider the possibility that we’re an invasive species that depends mostly on our ability to exploit, kill and otherwise inconvenience other species for our survival. Then we should ask how we can square that with having both consciousness and consciences.

What’s arresting to me in Coetzee’s arguments is how they push at the boundaries of my capacity for empathy. Because it is literally impossible to know what another organism is experiencing. You can only ever reason from analogy to your own experiences and based on some kind of observable input/response function. You can, of course, get pretty sophisticated with this. So, we know that prey animals tend to freeze under a range of conditions that we’re pretty comfortable talking about as “fear-inducing stimuli” — a stimulus previously paired with shock, smells associated with predators — without claiming to have any real understanding of what it’s like to be, say, a laboratory mouse experiencing fear.

If you are studying the neurophysiology of perception, you can’t really afford to dwell on the fact that you and I might have completely incommensurable experiences of the color red, or the taste of persimmon, or what have you, for very long. But relegating these considerations entirely to the philosophy department, having left off consideration of them in the midst of a stoned adolescent conversation on the roof of your house, is also a mistake. We have to deal with the consequences of this issue one way or another, and pretending it doesn’t exist is not a great strategy. The fact that we can’t know what another organism is experiencing means that we can’t be sure that any experimental situation, as we have designed and implemented it, is experienced by the organism under investigation in the way we intend it.

Say an intelligent alien species arrives on Earth and wants to study human intelligence. Say this species has relatively limited vision, but depends on finely tuned biosonar to navigate. Their windowless ship has an array of sensors that translate patterns of light and radio waves into signals that trained navigators can use for way-finding. Now they want to study how humans navigate a maze, so they build one in a pitch dark box. Humans turn out to be impossibly poor at this task. And yet they have observed us navigating space in their natural environment, operating vehicles, etc. How are we accomplishing this despite an apparently total lack of biosonar? It is a mystery.

In fact what’s stuck with me longest about Elizabeth Costello is the description of Kohler’s insight experiments from the perspective of Sultan the chimpanzee. It occurred to me while reading this passage that it might be helpful to think of brain regions as less like parts of ourselves than like members of a different species. Their environment and sensory apparatus — to say nothing of what they might have in place of “motivation” — are completely foreign to us, but can only be understood through the filter of our own experience.

The unconscious processes that support our perception of the world are no less alien for taking place somewhere in our own bodies. If anything, we are in a better position to imagine

This is a very common way to ask questions about how organisms represent the world when we can’t ask them directly (in the case of, say, other species, or pre-lingual infants) or we have reasons not to trust the responses we’d get from more direct questions (as I discussed in my previous post, regarding crafty adults in psychometric studies). Habituation — the decrease in response strength to successive presentations of the same stimulus– seems to be a basic property of neurons that is conserved across species. When habituation generalizes from one stimulus to another, it gives us some evidence that whatever is habituating — a sea slug, an infant, an ensemble of cells in our own brain — is treating the two stimuli as somehow “the same.”

In our experiments, we wanted to find regions that treated multiple different recordings of “ra” or “la” as somehow the same. So I made recording of myself producing these syllables with different tones of voice: now rising, now falling, now more enthusiastically, now less so, with a bit of vocal fry. (I did this alone in a sound-attenuated room, and still it was a bit embarrassing.) And these are what we played people over headphones while they relaxed in the MRI scanner. The hope was that we would observe a pattern of habituation in this condition, compared to the condition in which the speech categories alternated pseudorandomly, and that this would show us which brain regions treat all the different versions of “ra” and “la” as “the same.”

We found differences between the repeating and alternating conditions in the left supramarginal gyrus, which was consistent with a lot of studies of speech perception that had looked for habituation to speech sounds. But a closer look revealed that this difference could not be due to habituation. Specifically, we couldn’t find any evidence that responses actually declined as people listened to the trains of stimuli drawn from the same category. Instead, we showed that the pattern of responses in supramarginal gyrus were more or less what you would predict if you assumed that the response to every stimulus in the repeating trains was the same as the response to the first. So the difference between the repeating condition and the alternating condition was the result of an exaggerated response in the alternating condition, rather than habituation in the repeating condition, as we’d assumed.

This required us to rethink the relationship between what was happening in our task and this region’s role in speech sound categorization. We’d imagined a passive process of absorbing information from the environment, but the kind of “change detection” responses we found may be better interpreted in terms of attention. This was, if not disappointing, a bit disorienting, because we’d designed the experiments to try and take attention out of the equation, and focus on perceptual processes. And now attention had come, unbidden, into the equation.

What we found next only deepened my confusion. If this region is particularly interested in speech categories, it ought to show invariance to speaker. That is, a “ra” is a “ra” is a “ra,” independent of who is saying it. So, I made recordings of a female lab member doing a variety of renditions of these sounds, and then ran an experiment with two critical conditions: one in which people heard recordings of the same person producing different speech sounds, and one in which the same speech sound was produced by the two different people. (There was, of course, a control condition in which the same speaker repeated the same speech sound, as in the previous experiment.)

We found change detection responses in supramarginal gyrus as we had in the previous experiment, but in this case it responded to changes in who was talking. The response to phonetic change that we (and many others) had previously observed was too small to be statistically reliable. This was a bit of a crisis, because we’d designed a series of experiments around the premise that we could observe a signature for speech sound categorization in this paradigm, and that they would reflect habituation. The results were interesting, but they didn’t fit at all with the logic of the study as we’d designed it. Thus the long lunch breaks spent in the park, trying not to think about the data.

Eventually, we arrived at a way of thinking about these data that made some sense. The results could be interpreted in terms of the differences in the salience of the two types of change, and the relative surprising-ness of hearing one speaker produce different speech sounds, vs. hearing alternating syllables from different speakers. Indeed, the vast majority of our experience with speech involves a single person producing somewhat unpredictable trains of stimuli drawn from different speech sound categories. But when do we ever experience uncertainty about who will be speaking from moment to moment?

It’s amazing how easy it is to be blind, when trying to reduce a phenomenon to its component parts, to how the haphazard rearrangement of those parts must appear to the participant. I took a long detour through Coetzee’s protagonist’s imagination of Sultan the chimpanzee’s internal monologue, but I might just as well have tried to listened to the stimuli as a naive study participant. The problem was nothing so exotic as understanding an animal mind, or the phenomenology of a patch of cortex many synapses away from the outside world. The problem was that we had been focused on the supramarginal gyrus without considering its milieu — inside the head of someone who might find different aspects of the speech signal interesting. You know, like a normal person, instead of a scientist who has spent years thinking about the “lack of invariance” problem in speech perception..

Image Credit: Delicious-looking brain sandwich. A twist on a St. Louis classic (surely you’ve seen the Brains 25¢ sign outside someone’s lab? ever noticed the Arch in the background?). Photo by Holly Fann for the Riverfront Times.

The paper we eventually wrote about this work:
Zevin J.D., Yang J., Skipper J.I. & McCandliss B.D. (2010). Domain general change detection accounts for “dishabituation” effects in temporal-parietal regions in functional magnetic resonance imaging studies of speech perception., The Journal of Neuroscience, 30 1110 -1117. PMID:

Post a Comment

Your email is never published nor shared. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>


Get every new post delivered to your Inbox

Join other followers: