Yarden Cohen

I am a postdoctoral research associate in Timothy Gardner's lab at Boston University. Our lab studies songbirds as a model for the neural mechanisms underlying learning and maintaining complex stable behavior - the song. In working with canaries, a species capable of producing complex and highly variable songs, I recently developed a deep neural network tool for segmenting these complex data into its basic elements - syllables - and labeling them automatically. This approach replaced the laborious manual song annotation task, traditionally taking months of work by several people, and, together with our lab's open source microscopy techniques, allowed studying the neural coding of song elements and sequence in these complex singers.

Prior to working with songbirds I investigated motor learning and developed a theoretical and experimental perturbation-based approach to studying sequential skill acquisition (publication in preparation). Apart from motor learning, I worked on classification learning, a cognitive ubiquitous skill in which subjects are required to overcome the computationally complex task of adopting new behavioral policies. I built a feature-based modeling framework of human learning that allowed both successfully predicting behavior as well as guiding it by online model fitting and tailoring a subject-specific training set. To investigate the neural correlates of classification learning behavior, I then conducted electrophysiological recordings in non-human primates and discovered diverse and brain-region-specific dynamics that correlates to both behavior and internal learning-related representations of visual features and categories (publication in preparation).