Ryoma Hattori

Assistant Professor @UFScripps. Neuroscience / Decision making / Reinforcement Learning / Machine learning

[Nature Neuroscience @NatureNeuro]

Neural activity and synaptic plasticity in mouse orbitofrontal cortex (#OFC) mediates multiple timescales of reinforcement learning (RL) for meta-RL, which parallels a form of meta-RL in artificial intelligence

https://t.co/x5gbwCJZuA

[Ryoma Hattori @ryhattori]

If you are interested in this kind of projects, positions are still available in my new lab :) https://t.co/Tq08S4W0sd

[Ryoma Hattori @ryhattori]

How can our brain mediate RL of RL (aka meta-RL)? Our new study @NatureNeuro showed that mouse OFC mediates multiple layers of RL in parallel at distinct timescales using activity dynamics and synaptic plasticity, which resembles a meta-RL model in AI. https://t.co/MqPeeAsOkJ

[Ryoma Hattori @ryhattori]

Lab website is out! Recruiting lab members. Positions are open until filled. https://t.co/Tq08S4W0sd

[Ryoma Hattori @ryhattori]

A bit late announcement on twitter: I will move to 🏝️Florida🐊 next month to open my lab as an assistant professor @UFScripps My lab will study decision making, RL, number sense, autism using VR tasks for mice. I am recruiting postdoc/student/tech!!! Email or DM me if interested.

[Ryoma Hattori @ryhattori]

@AllenNeuroLab Thank you! Excited to move to SoFlo!

[Ryoma Hattori @ryhattori]

A bit late announcement on twitter: I will move to 🏝️Florida🐊 next month to open my lab as an assistant professor @UFScripps My lab will study decision making, RL, number sense, autism using VR tasks for mice. I am recruiting postdoc/student/tech!!! Email or DM me if interested.

[Ryoma Hattori @ryhattori]

@gmishne Thank you, Gal!

[Ryoma Hattori @ryhattori]

@AssafRamot Thanks, i will miss you!

[Ryoma Hattori @ryhattori]

@IpshitaZ Thank you!

[Ryoma Hattori @ryhattori]

@neuro_phys Thank you :)

[Blake Richards @tyrell_turing]

Friends, we need your help. An Wu, a postdoc from UCSD is missing post-Cosyne. We're trying to locate her. We're worried she may have been staying at a building that had a bad fire this week. If you have any information about where An was staying or last seen please contact me.

[Ryoma Hattori @ryhattori]

I am honored to have been selected for the SFARI Bridge to Independence Award! Looking forward to investigating the reinforcement learning in autism.

[Ryoma Hattori @ryhattori]

@poppingjun In the protocol we used #1 and #2 coverslips, but #1.5 (~170μm) also works if you prefer the thickness. All glasses are prepared from normal coverslips. No special glasses are necessary.

[Ryoma Hattori @ryhattori]

Our protocol for an ultra-large cranial window and longitudinal 2-photon calcium imaging is out @STARProtocols! I hope this protocol will be useful to many. Get more data with larger craniotomy.
https://t.co/hUIPpgaFlU

[Ryoma Hattori @ryhattori]

The code is available at https://t.co/N2KD38lwaF
I substantially improved the stability of the distortion correction over the past few months, so get the latest if you downloaded the preprint version before!

[Ryoma Hattori @ryhattori]

Our PatchWarp paper is out @CellRepMethods! PatchWarp corrects fast motion artifacts and slow image distortions in calcium imaging data. It can also register images across sessions. https://t.co/yNQC10sKFU

[Ryoma Hattori @ryhattori]

Example distortion corrections by PatchWarp for cell body and axon imaging data.

[Ann Kennedy @Antihebbiann]

My Preview of @takaki_komiyama lab's recent Neuron article is now live! An interesting look at the population dynamics underlying a persistent value representation: https://t.co/wv6hj5UUTK

[Ryoma Hattori @ryhattori]

I also thank the reviewers for their constructive and helpful feedbacks which improved the paper a lot during the revisions. If the paper is of interest to you, please also check our related Cell paper from 2 years ago: https://t.co/CcUTmgiYc8 A part of data came from this one.

[Ryoma Hattori @ryhattori]

Our new paper on the cortex-wide activity dynamics and the distributed value coding during reinforcement learning is out in Neuron! 2p imaging from 6 areas (RSC, PPC, M2, ALM, V1, S1) in 2 different tasks + 2 types of RNNs with recurrent activity-based RL + new method (dsPCA).

[Takaki Komiyama @takaki_komiyama]

1/ Our new paper is out online today in Neuron @NeuroCellPress. Congrats Ryoma @ryhattori! https://t.co/bPN41xygvV

[Ryoma Hattori @ryhattori]

dsPCA identifies targeted axes in supervised way, then identifies PC axes of the remaining activity variance in unsupervised way. Thus, you can obtain the coding axes of your targeted signals and the dominant non-targeted axes of the remaining population activity variance.