The MBB Lunch Series is free and open to the Harvard community. For lunch, please RSVP.

Capturing Animals Behaving
Ugne Klibaite
Postdoctoral Fellow, Organismic and Evolutionary Biology

Advances in computer vision and deep learning have made it possible to explore animal behavior at fine temporal and spatial scales. I will illustrate how unsupervised methods for quantifying behavior can be used to better understand behavior across several species. First, I will show how pairs of fruit flies tune as well as synchronize behavior throughout social interactions. Next, I will demonstrate how deep learning-based pose estimation can be combined with machine learning to study habituation in mice during the open field test (OFT). Finally, I will introduce a current project aimed to capture complex multi-scale phenotypes in rodent models of autism using long-term behavioral recordings. The long-term goal of this project is comprehensive phenotyping as well as developing theories of behavioral organization.

Cortical Music Selectivity Does Not Require Musical Training
Dana Boebinger
Graduate Student, Auditory Neuroscience (SHBT)

Human auditory cortex contains neural populations that respond strongly to a wide variety of music sounds, but much less strongly to sounds with similar acoustic properties or to other real-world sounds. However, it is unknown whether this selectivity for music is driven by explicit training. To answer this question, we measured fMRI responses to 192 natural sounds in 10 people with extensive musical training and 10 with almost none. Using voxel decomposition (Norman-Haignere et al., 2015) to explain voxel responses across all 20 participants in terms of a small number of components, we replicated the existence of a music-selective response component similar in tuning and anatomical distribution to our earlier report. Critically, we also estimated components separately for musicians and non-musicians and found that a music-selective component was clearly present even in individuals with almost no musical training, which was very similar to the music component found in musicians. We also found that musical genres that were less familiar to our participants (e.g., Mongolian throat singing) produced strong responses within the music component, as did drum clips with rhythm but little melody. These data replicate the finding of music selectivity, broaden its scope to include unfamiliar musical genres and rhythms, and show that it is robustly present in people with almost no musical training. Our findings demonstrate that musical training is not necessary for music selectivity to emerge in non-primary auditory cortex, raising the possibility that music-selective brain responses could be a universal property of human auditory cortex.