Johannes Bill

Postdoc with @jdrugowitsch and @gershbrain, interested in inference and learning in biological and artificial brains.

Mastodon: @[email protected]

[Johannes Bill @BillScientific]

How does the brain discover motion relations from a volatile visual stream? And how does discovered structure shape #MotionPerception? Find out more in our work with @gershbrain and @jdrugowitsch, now published in @NatureComms: https://t.co/vQJvMvmKCZ.

[Johannes Bill @BillScientific]

I am excited to share our new #tweeprint with @gershbrain and @jdrugowitsch studying how the brain discovers the motion structure of visual scenes online from a volatile retinal stream: https://t.co/lnyNEeLmnT 1/

[Lucy Lai, PhD @drlucylai]

hello twitterverse! here's my first, first-authored preprint with the wonderful @AnnHuang42 and @gershbrain! in this paper, we used our theory of policy compression to better understand action chunking 🧵…(1/n) https://t.co/x7xNgTiYHA

[Jan Drugowitsch @jdrugowitsch]

If you are at #cosyne2022, please come check out our lab’s diverse set of posters and talks, conveniently spaced across days for easily digestible consumption. Here a quick overview. 1/4

[Jan Drugowitsch @jdrugowitsch]

Postdoc opening alert!

Interested in the neural computations underlying decision-making and navigation? Enjoying close theory-experimental collaborations? Come join our group @ @harvardmed! See details at https://t.co/agnrKg5Lzw or ask me at #cosyne2022.

Please RT!

[Anna Kutschireiter @ankutschi]

#tweeprint time! Uncertainty is a feature, not a bug - and maybe bugs can have uncertainty, too! @MelBasnak, @RachelIreneWils, @jdrugowitsch and I provide a Bayesian perspective on the heading-direction system in the fruit fly Drosophila: https://t.co/3eaw0reEfa

[Jan Drugowitsch @jdrugowitsch]

Now available online at @NeuroCellPress: Dr. Emma Krause's excellent work on the structure of spatial trajectories encoded in awake replay. See below for a thread on the pre-print version. Get the Neuron version with added analyses and controls at https://t.co/VGRm1XHnWK.

[Johannes Bill @BillScientific]

Furthermore, the algorithm affords a neural network model that shares properties with motion-sensitive cortical areas MT and MSTd and motivates a novel class of neuroscience experiments to study latent structure representations in the brain. 5/5

[Johannes Bill @BillScientific]

I am excited to share our new #tweeprint with @gershbrain and @jdrugowitsch studying how the brain discovers the motion structure of visual scenes online from a volatile retinal stream: https://t.co/lnyNEeLmnT 1/

[Johannes Bill @BillScientific]

We derive an algorithm that decomposes motion in a scene online and explains human percepts for a wide set of stimuli, from classical psychophysics experiments to illusory motion displays, such as motion direction repulsion. 4/

[Anna Kutschireiter @ankutschi]

Bigger, better and with even more dimensions! Check out our updated pre-print on probabilistic path integration, the Circular Kalman Filter and more! https://t.co/dF7Wm3LbeZ #tweeprint

[Jan Drugowitsch @jdrugowitsch]

#tweeprint time! Check out the new work of Emma Krause and myself on showing that almost all awake hippocampal sharp-wave ripples (SWRs) appear to encode trajectories with momentum through the environment: https://t.co/rHMM5xGIoy 1/

[Anna Kutschireiter @ankutschi]

#tweeprint time! Luke Rast, @jdrugowitsch and I formalized a probabilistic theory of angular path integration in our latest pre-print, and present to you the Circular Kalman Filter: https://t.co/eUj2a3InlN 1/

[Johannes Bill @BillScientific]

This extends our previous work in Bill et al., 2020 (https://t.co/cKO9wEVVW1) by treating the underlying structure itself as a latent random variable that has to be inferred. 5/5

[Johannes Bill @BillScientific]

It is my pleasure to announce the publication of our paper “Human visual motion perception shows hallmarks of Bayesian structural inference” with @sichao_yang, @jdrugowitsch and @gershbrain. Find it at: https://t.co/nCVZBS5ibC 1/

[Johannes Bill @BillScientific]

We found that many facets of human structure perception, incl. perceptual error patterns, were quantitatively explained by a model of Bayesian structural inference—especially, when object motion was ambiguous, or hierarchically nested within other moving reference frames. 4/

[Johannes Bill @BillScientific]

@ShahabBakht @gershbrain @sichao_yang @jdrugowitsch Thank you, @ShahabBakht! There was no significant performance difference between the first and the second repetition of unique trials. However, we should look into how the likelihood under the Bayesian model evolves.