CCN 2018 Conference presentation: Learned context dependent categorical perception in a songbird

Posted on Tue 19 March 2019 in birdsong

In September I published a short conference paper at CCN 2018, which is a new (2017) conference that is focused on the interface between computational neuroscience, cognitive science, and arfficial intelligence.

My conference paper was selected for an oral presentation, which is embedded below. The topic of the presentation is part of my PhD thesis, and explores the role of context in categorization, perception, and decision making. We use a novel set of tools that I developed (now located at github.com/timsainb/AVGN), where latent representations (from variational autoencoders) of birdsong are linearly sampled from to produce smoothly varying stimuli for a category learning task. In this task, we were successfully able to bias categorical perception in songbirds based upon contextual information, and are currently in the process of studying the neural mechanisms underlying that contextual bias in perception.

My presentation at CCN 2018

I also co-authored a second paper at the same conference with the lead author Marvin Thielk, that goes more into depth about the perceptual and neural representations of acoustic categories in European starlings. Other videos and papers from the conference are available on the CCN website as well!