Seeking students — especially Canadian residents, who are less expensive, but all applicants welcome — to study the computational and neural underpinnings of rhythm perception and production! Reach out with questions or interest at [email protected].
Back home after a fantastic @rppw.bsky.social . A “home conference” is where you don’t have to explain why people should care about what you care about, and having one is such a luxury.
Big congratulations and thanks to first author Yassaman Ommi, who took her Master’s project very seriously while consistently imagining that she was training a robot monkey to dance, and to her Master's Defence committee (including John Iversen) for helpful comments.
Thus, we suspect that the development of human synchronization (and human enjoyment of synchronization) is scaffolded by dopaminergic signals that reward earliness over lateness and that reinforce the imitation of observed time intervals, perhaps as a special case of rewarding action imitation.
...and the activity also showed patterns repeating with a period of two clicks, which strongly suggested the human tendency to hear metronome clicks in groups of two.
But that's not all! When we looked at the "neural activity" in the network during synchronization, we saw loops of activity with radii dependent on tempo, much like the neural activity observed in monkeys trained to synchronize...
When the training set consisted of metronomes with jittered timing, this rule produced taps slightly ahead of the clicks, just like humans do, and reduced this lead time with additional training, just like humans with musical training.
Only the rules that penalized lateness produced synchronized tapping. And only the rule that also rewarded correct intervals produced synchronized tapping that could correct error induced by metronome timing perturbations.
We explore several possible reinforcement rules. Each gives maximal reward for taps perfectly synchronized with clicks. But some strongly penalize lateness, and some additionally reward the production of intervals of appropriate duration.
The model consists of an LSTM recurrent neural network that receives periodic “auditory” "clicks" at a range of periods, produces “tap” actions that occur after a short delay, and is rewarded for its actions according to a reinforcement scheme.
Preprint alert: how do we learn to synchronize our movement with rhythmic sounds? And why is it so much fun? We’ve created a reinforcement learning model to explore possible hypotheses, and one hypothesis came out on top! Read on! papers.ssrn.com/sol3/papers....
If you are interested, please send me a CV and cover letter at [email protected], plus emails for three references. Postdocs will start ASAP, and students will probably have to wait until fall 2026 (but never too early to reach out!)
...and could also use help perfecting rhythm cognition experiment designs and analyzing neural time series data. I am looking for candidates with experience in any combination of these competencies.
Over the next two years, I am looking for at least one postdoc and several students to work with us, based at McMaster in Hamilton ON Canada. I will mainly need help developing computational models based in systems-level neurophysiology and/or predictive processing...
We will be testing several theories based on models of reinforcement learning and action selection in the basal ganglia loop and of rhythm perception as a process of Bayesian active Inference.
Dan's group will use fast-scan cyclic voltammetry to measure dopamine levels in human basal ganglia at subsecond resolution while neurosurgery patients listen and/or tap along to rhythmic stimuli. (Yes, it's possible!)
I am honored, delighted, excited, and slightly panicked to have just received a Human Frontier Science Project grant. I will work with Dan Bang at University of Aarhus, Denmark, to explore the role of dopamine in the human capacity and compulsion to predict and move along with auditory rhythms.
Not that anybody would be looking for opportunities outside the states these days, but I'm recruiting a PhD student in auditory stuff (Developmental and/or Cognitive / citizen science etc) here in Auckland, NZ
Thanks to early readers -- @perikeller.bsky.social , Peter Vuust, @sylvienozaradan.bsky.social, Alan Wing -- and to my lab for following me down this rabbit hole! Experimental results to follow in not too long!
So what can we do with this understanding? In addition to explaining experimental results, this idea poses challenges to existing models of rhythm production and entrainment, and may be a key to understanding the role of the cerebellum and the impact of autism on sensorimotor synchronization.
