[BIRDS CHIRPING] JESSE GOLDBERG: So I got interested in neuroscience actually as a teenager when I was taking a chemistry class. And we did organic chemistry. And I learned that all of life is driven by chemical processes. And with the functioning of digestion or a liver or a kidney, this is rather intuitive.
But what seemed magical to me is that something so seemingly immaterial like thought or emotion, that that could also be driven by chemical processes. That really got me hooked on neuroscience.
But in medical school I learned through interacting with neurological patients and with psychiatric patients that many neurological diseases have cognitive or psychiatric parts of them. For example, Parkinson's patients have a difficulty initiating movement. But they also have difficulty initiating thought.
And someone who is psychotic, or a floridly psychotic unmedicated schizophrenic will have highly disorganized thinking, but also highly disorganized moving. And so I started to go up on a path to study motor systems as a way of getting at these somewhat more interesting cognitive aspects that got me interested in the brain in the first place.
So birdsong is in some ways the perfect motor behavior, as far as I'm concerned, because it's learned. Baby birds start off by babbling. And they have to learn to produce the stereotyped or skilled adult song.
It's also a good behavior because it's sequential. So a bird song unravels in a sequence. Chicka wa anh, chicka wa anh, chicka wa anh.
And one of the things in addition to this complex behavior that the birds learn that make songbirds such an attractive model system for neuroscientists is that they have a specialized brain circuit for singing. Here's a schematic of the songbird brain. And what we now know are these discrete nuclei that are entirely devoted to song production and learning.
And these nuclei, thanks to recent advances in our understanding of avian neuroanatomy, are now understood in the context of mammalian forebrain architecture, by which I mean we know that some of the structures in the song system are akin to our cerebral cortex. Other structures are akin to our basal ganglia or thalamus. So by studying the song system, we can learn actually about how parts of our own brains work.
So what's been so exciting about the songbird field in the last several years-- and some of our findings have contributed to this-- is that it's not just the general similarities of the forebrain architecture between mammals and songbirds that exist, but actually the specific cell types and their specific interconnections seem to be evolutionarily conserved. So for example, in mammals, you have a dopamine projection to a basal ganglia region that has these four specific striatal cell classes and two pallidal cell classes. And the structure of the circuit in mammals is very well understood. And especially in the context of Parkinson's disease, where these neurons die that result in abnormal activity patterns in these specific types of neurons.
But what we found in the songbird is that the same pathway exists. There's a dopamine projection to the basal ganglia with the same cell classes. And one of our recent findings was that these GPe and GPi specific types of pallidal neurons, we recorded them in songbirds. And then we also got data from monkeys. And we saw that they exhibit almost identical firing patterns during behavior.
And so what this hints at is that these two pathways are not only evolutionary conserved in their structure, but we've also recently found that basal ganglia lesions block song learning just like basal ganglia lesions in mammals block motor learning. And motor learning is impaired in many of these diseases, such as Parkinson's, Huntington's, and dystonia.
And now we have a similar circuit where we can analyze the behavior of these neurons to see how basically the whole thing works for learning.
So when I went and looked at what the basal ganglia are actually doing in other animals, it became clear from reading the literature that the basal ganglia are involved in learning. And it's a specific type of learning-- motor learning-- and trial and error learning.
So we think of trial and error learning as involving three key processes the brain has to carry out. First, you have to try new things when you're a beginner. Second, you have to evaluate the outcome of those new things. You have to say was that good or was that bad. And then third, you have to use that evaluation signal to train your performance so it gets better in the future.
So if we think about this in the context of learning to play tennis, for example, you set out to play. You're a beginner. And the first thing you do is you spray balls all over the court. This trial and trial variability is essential for learning. And then of course, with practice, as a function of evaluation and using that evaluation skill to improve, you can become a stereotyped expert.
Birds learn to sing much like that trial and error learning process I just discussed as a tennis example. What baby zebra finches are trying to do is copy their dad's song. And the adult song of a zebra finch is a highly stereotyped motor output that sounds a little bit scratchy, but very stereotyped like this.
So what we're looking at here is a spectrogram visualizing frequency versus time. And this adult bird has a stereotype song abc, abc, abc And this is what this male uses to attract mates.
Now a juvenile bird may hear this song. But he can't immediately produce it. He has to figure out how to do it on his own.
And the first vocalizations that a juvenile makes is highly variable, akin to vocal babbling. And these syllables, they sound different every time they sing them. And I'll just play you an example now.
So if you look at minutes of babbling, you don't hear any repetitive structure. But after a couple of weeks of practice, the bird gets a little bit better. And you can see that some syllable structure becomes repetitive.
You'll notice here there's a high-pitched note that the bird repeats several times.
So this is kind of like the intermediate phase of learning. We call it plastic song. But he's not a beginner, but he is not an expert. He's an intermediate level.
But then give them a few more weeks of practice, and he will nail his father's song. And you can see this here. [BIRDSONG] With the same three syllables abc that his dad had.
A central ingredient in trial and error learning, as I mentioned, is the generation of variability. When you're a beginner at something, you have to try a lot of new things. And we and others in the songbird field have made tremendous progress in understanding exactly how the brain can produce variability. And for example, we found that a specific part of the brain in the frontal cortical region and in the thalamus, which is part of this basal ganglia circuit, actively produce variability.
And by taking this part of the circuit out by lesioning it or inactivating it, we can make a highly variable babbling bird sound adult-like in its stereotyping. And the implication is that the variability that we exhibit as juveniles, or when we're beginners at something, is not the de facto output of an immature or unlearned motor system, but rather as actively generated. And this discovery that variability or exploration can be actively generated in the brain of songbirds really changes the way that we might think of variable behaviors in humans.
In short, one of the big challenges in neuroscience is to understand how behavior is driven by the coordinated activity of neurons inside the brain. And the problem is that there are billions of neurons in humans. And in the songbird, which is a simpler system, there are still hundreds of thousands of neurons that are producing behavior.
So the real challenge is to understand how the electrical activity in these populations of neurons produces behavior. And what we have in the songbird is a very quantifiable behavior, cause when we record the song, and a tractable circuit that drives that behavior. So we can begin to understand how electrical signals produce a learned behavior.
We've received your request
You will be notified by email when the transcript and captions are available. The process may take up to 5 business days. Please contact firstname.lastname@example.org if you have any questions about this request.
Many species of birds learn their songs. Initially, they copy their parent's songs very imperfectly, a stage called babbling. But with practice and repetition they eventually get it right. Jesse Goldberg, in the Department of Neurobiology and Behavior, is studying the neural basis of bird song learning and points out that it bears many similarities to human learning.