Predicting animals' responses to stimuli
In the summer of 2011, I took a course called Methods in Computational Neuroscience at the Marine Biological Laboratories in Woods Hole, MA. This was a fantastic course where graduate students, post-docs and young faculty came together to learn about neural coding, information theory, dynamical systems in neuroscience, learning theories, etc. The small fisherman's village provided a wonderful retreat.
In the last part of the course, we worked on individual projects under the guidance of faculty mentors. I decided to work on a topic related to my PhD thesis: how do animals sleep? To characterize sleep and wake behavior, we use the model organism of the larval zebrafish. The level to which animals respond to their environment is indicative of their arousal state. Therefore, we strive to describe responses to external stimuli and build predictive models.
I analyzed the behavior of hundreds of zebrafish larvae responding to acousto-mechanical stimuli. I found that movement dynamics prior to the stimulus were predictive of response probability. In addition, I was able to do a basic classification of responses based on movement dynamics. This was a short project that started to get me thinking about building a more comprehensive model of arousal behavior. My faculty mentor extraordinaire was Elad Scheidman from the Weizmann Institute of Science in Israel.