Electric duet

While sifting through a two-week continuous recording of electric fish in their natural habitat, I observed synchronous frequency modulations on a scale of seconds to two ten minutes between two fish. To detect these modulations, I developed a covariance-based detector, which I then used to detect these events on recordings. The following plot shows some examples of detected modulations:

To detect spatio-temporal interactions, particularly when fish chased each other, I build a cost function that peaks when (1) fish swim in the same trajectories relative to each other, (2) velocities of both fish increase, and (3) fish accelerate rapidly. By analyzing the estimated positions of fish over time, I demonstrated that those involved in frequency duets approach one another following the initiation of their “choir.” I also rendered videos depicting some of these interactions: Fish moving as data points on an electrode grid and their frequencies changing on the right-hand side.

The resulting output from this effort was displayed as a poster at the 2022 International Conference of Neuroethology (ICN). This poster can be accessed through the link provided in the GitHub repository below.