Songbirds Teach Us About Brain Areas Involved in Vocal Learning

Post by Anastasia Sares

What's the science?

Vocal learning is at the core of human linguistic and musical abilities, allowing us to imitate sounds produced by others and use them for communication. Only a few other species are capable of vocal learning, and songbirds are one such species. This makes the songbird an excellent model organism to help scientists characterize vocal learning circuitry in the brain. A number of brain regions have been identified as crucial to this process— a central one is Area X, within the basal ganglia (structures responsible for learning and initiating behavior). Other notable regions in the vocal learning circuit are involved in receiving auditory inputs (AIV), coordinating motor output (the RA), and processing motivation and reward (VTA). However, we still don’t have a full picture of how this ensemble functions and how different brain regions might be involved. This week in Neuron, Ruidong Chen and colleagues showed that another area in the basal ganglia of birds, the ventral pallidum (VP) was an important part of the vocal learning system.

How did they do it?

The authors combined data from anatomical tracing, electrical stimulation, lesions, and response to distorted auditory feedback to demonstrate the importance of the ventral pallidum.

To trace the connections of neurons between different regions, they used two methods. The first was retrograde tracing with viruses: in this technique, special viral proteins are modified to be fluorescent and then injected in the target region. They naturally climb backwards from the end of a neuron and cause the region of origin to light up. The second method is antidromic spiking: it’s a similar concept but with electric signals. Stimulating the target of a neuron causes electrical signals to move backwards up it, and these backwards-moving charges can be recorded at the region of the neuron’s origin. Once they mapped out the system, they tested how the VP reacted to singing and vocal errors. Again, the authors employed a two-pronged approach. First, they performed a surgery to disrupt the function of the VP (lesion) and observed its consequences on the bird’s song development. Second, they implanted recording electrodes in the VP and put birds into an enclosed environment with speakers that would play back a distorted version of the bird’s song at specific points while the bird was singing. Finally, they also played the bird’s own song back when it wasn’t producing any song, which should only activate audition-related areas.

What did they find?

The authors found anatomical and functional evidence supporting the idea of a loop in the songbird vocal learning system incorporating the ventral pallidum (Area X→VP→VTA→Area X). The VP also received inputs from a variety of vocal learning areas. Disrupting the VP in juvenile birds resulted in abnormal song learning, indicating that it was a necessary part of the learning network. Some neurons in the VP were related to auditory information in general, as they fired during song performance and during a song played back to them later. Other neurons seemed to be calculating and responding to singing errors. During singing, these neurons responded to differences between distorted sounds and undistorted sounds, but they did not respond to songs or movement in general. Signals from these error-detecting neurons were the ones that  left the VP and traveled to their next stop (the VTA).

What's the impact?

This research adds the VP in the middle of a complex neural network controlling vocal learning, helping to map out its relationship with other already-established areas. Though the VP is usually thought of as a region processing emotion, reward, and motivation, the authors contend that it can act as an internal “critic,” helping the birds to continuously refine their songs. Since the basal ganglia are fairly well preserved across species, studying these circuits will help us understand what is going on in human vocal learning as well. Further research into these systems may help us to understand internally-driven learning processes more generally.

Chen, Ruidong et al. Songbird Ventral Pallidum Sends Diverse Performance Error Signals to Dopaminergic Midbrain. Neuron (2019). Access the original scientific publication here.

The Brain Represents Others as a Summation of Their Mental States

Post by Flora Moujaes

What's the science?

To be successful in life it helps to take people’s individual dispositions into account. For example, a good teacher will tailor their teaching strategy to the temperament of the student, encouraging shy students to participant more while challenging the assumptions of overconfident students. However, we still don’t have a clear picture of how people tailor their actions to the idiosyncrasies of specific individuals. Traditionally, it has been assumed that people may do this by representing others using traits: the unchanging characteristics that define someone, such as trustworthiness or intelligence. More recently, it has been suggested that people represent others using summed states: feelings people experience on a moment-to-moment basis, such as happiness or shame. There are three main reasons that representing others using states may be advantageous (1) states are easier to observe than traits, as they can be seen in the moment rather than having to get to know someone over a long period of time, (2) states, even if unrelated to a person’s general disposition, are an independently useful for predicting behaviour, and (3) by summing someone’s mental states over time, one can infer long-term characteristics or traits.

This week in Nature Communications, Thornton and colleagues demonstrate in an fMRI study that we represent other people as the sum of their moment-to-moment states, as our neural representations of other people are composed of combinations of representations of the mental states those people are perceived to frequently experience.

How did they do it?

To explore the hypothesis that the brain represents others according to a sum of their states, the authors began by establishing the pattern of brain activity associated with 60 different celebrities, from Shakespeare to Snoop Dog, based on the states people associated with each celebrity. To do this they first conducted an fMRI experiment to establish what patterns of activity are elicited when the brain thinks about each mental state (e.g. patience). They then conducted an experiment to determine which mental states were associated with each celebrity (e.g. is Snoop Dog patient?). Finally, they combined the data from both studies to come up with a pattern of brain activity for each celebrity based on the set of states associated with them.

They then tested whether the pattern of brain activity they created for each celebrity based on the states associated with them, reflected the brain activity seen when people thought about the celebrities. In order to do this, they conducted an additional fMRI experiment where they asked people questions about each celebrity (e.g., how much would Snoop Dog like to learn karate?) to get them to think about the celebrity. They then correlated the artificial (based on the states associated with them) and actual patterns of brain activity elicited by thinking about these famous people.

What did they find?

Testing the Summed State Account: They found evidence that people did represent others as a sum of their states, as there was a correlation between the artificial patterns of brain activity created for each celebrity based on the states associated with them, and the actual brain activity of each participant when thinking about the celebrity. Summed State vs. Trait Accounts: They also compared whether participants’ neural representations of celebrities were better explained by considering the states or traits associated with a celebrity. They found that the summed states account consistently outperformed the trait alternative in explaining person representation.

For more details see Thornton’s summary on Twitter.

What's the impact?

This is the first study to show that when you think about a person, your brain may represent them as the sum of the mental states you think that they frequently experience. This suggests that people tailor their actions to the unique characteristics of individuals by observing differences in their momentary thoughts and feelings. These findings also help us understand the relationship between how we might think about people's momentary thoughts and feelings to infer their long-term traits. Thus, the trait–state divide may be narrower than commonly thought. Overall, the summed state hypothesis provides a compelling model of how the mind and brain may learn about, represent, and predict other people.

Thornton et al. The brain represents people as the mental states they habitually experience. Nature Communications (2019). Access the original scientific publication here.

Fast Learners Have High-Dimensional Representations of Neural Activity

Post by Shireen Parimoo

What's the science?

People naturally group new pieces of information into categories to differentiate between them, eliciting distinct patterns of brain activity. One measure of quantifying how information is coded in the brain is by estimating the geometric representation or dimensionality of neural responses. For example, stimuli can be coded in a psychological dimension and represented based on their size and color. Patterns of brain activity can be represented in a high dimension, meaning that the information is coded along many dimensions and that neural responses to each stimulus are likely more distinct from one another. Alternatively, low-dimensionality of neural representations would suggest that there are fewer dimensions along which new information can be coded in the brain, resulting in less distinction between stimuli. Is there a relationship between the dimensionality of neural responses and how fast people can learn new information? This week in Nature Neuroscience, Tang and colleagues used functional magnetic resonance imaging (fMRI) to examine how the dimensionality of neural representations is related to the speed with which people learn new information.

How did they do it?

Nineteen young adults completed various tasks for four days while undergoing fMRI scanning. On each day, they first completed a value-learning task, in which they learned the monetary value of twelve stimuli that ranged from $1 and $12. They were then tested on the stimulus-value associations in a value-judgment task and their response accuracy was used as a measure of learning effectiveness. In the value-judgment task, participants were shown the stimuli and had to indicate if their monetary value was in the top 50% ($7-$12) or the bottom 50% ($1-$6) of the range. In the fMRI data, dimensionality was estimated from label assortativity and separability values of the neural responses. Label assortativity provides a measure of how easily neural responses can be distinguished from each other. In general, higher assortativity is associated with greater dimensionality, but it is also possible for high dimensional representations to have low assortativity. Assortativity was calculated using a linear support vector machine, which is a machine learning technique used to classify information into categories. On the other hand, separability values provide a more direct measure of dimensionality; this was calculated by assigning binary labels to neural activity for each stimulus and using a support vector machine to classify the neural responses according to these labels. High separability values indicate that neural responses are organized in a larger number of dimensions, whereas low values indicate lower dimensionality of representations in neural responses.

The authors examined the relationship between dimensionality and how effectively participants learned the stimulus-value associations. They also conducted a virtual lesioning analysis, in which neural activity in specific brain regions was systematically excluded from the analysis to see which regions contributed to the association between separability and learning effectiveness. Finally, as high dimensionality of neural responses is associated with greater reliance on neural resources, the authors investigated whether resources are allocated efficiently through a balance between high- and low-dimensionality of representations during learning. Participants’ response accuracy was correlated with the dimensionality of representations to specific stimuli (stimulus dimension) during learning and to neural responses independent of the stimulus (embedding dimension), which represents the overall dimensionality of participants’ neural representations.

What did they find?

Response accuracy on the value-judgment task improved over the course of four days, and fast learners – participants who performed above average on the first day – had high accuracy on subsequent days. Assortativity of brain activity was higher among participants who learned fast. That is, stimulus-specific neural responses of fast learners could be distinguished from each other more easily than that of slower learners. Fast learners also had higher dimensionality of neural representations, which is another indication that their representations of the different stimuli were more distinct from one another. In fact, high performance on the first day of the experiment was correlated with high dimensionality on the fourth day, suggesting that effective learning facilitates more precise coding of information in the brain. Interestingly, there was a positive correlation between response accuracy and dimensionality in the stimulus dimension, but a negative correlation in the embedding dimension. This means that fast learners allocate neural resources efficiently by representing all stimuli in an overall low dimensional space, but within that low-dimensional embedding space, specific stimulus-value associations are represented by higher dimensional, and thus more distinct patterns of activity.

What's the impact?

This is the first study to demonstrate how the efficiency of neural responses is optimized during learning, depending on whether it is associated with learning stimulus associations or independently of specific stimuli. The approach of quantifying the dimensionality of representations in neural responses and the finding that dimensionality is correlated with behavioral performance both have important implications for our understanding of how information is represented in the brain and how that affects behavior.

Tang et al. Effective learning is accompanied by high-dimensional and efficient representations of neural activity. Nature Neuroscience (2019). Access the original scientific publication here.