Gustav Markkula

Realistic pedestrian-driver interaction modelling using multi-agent RL with human perceptual-motor constraints Modelling pedestrian-driver interactions is critical for understanding human road user behaviour and developing safe autonomous vehicle systems. Existing approaches often rely on rule-based logic, gam...

🚨Preprint! Yueyang Wang is bringing his human behaviour modelling out of the VR lab, showing that by constraining deep multi-agent RL with human-like perceptual and motor limitations, a very small real-world dataset is sufficient to outperform a behaviour cloning baseline arxiv.org/abs/2510.27383

It is my pleasure to share that in 2028, we will be hosting the 9th International Conference on Traffic and Transport Psychology (ICTTP9). Exciting times ahead, we can’t wait to welcome you!

@markkula.bsky.social, @hfs-its-leeds.bsky.social, @envleeds.bsky.social, @universityofleeds.bsky.social

A brain-wide map of neural activity during complex behaviour - Nature The International Brain Laboratory presents a brain-wide electrophysiological map obtained from pooling data from 12 laboratories that performed the same standardized perceptual decision-making task i...

Sad day for peope who still believe brain areas are the primary organizational units of function in the brain.

www.nature.com/articles/s41...

New publication 📢 "Testing the Validity of Multiparticipant Distributed Simulation for Understanding and Modeling Road User Interaction" led by Amir Hossein Kalantari together with Yi-Shin Lin, Ali Mohammadi, Natasha Merat OBE, and @markkula.bsky.social.

Access: ieeexplore.ieee.org/document/111...

I love this experience-sampling study on the absolute pitch of earworms! It's a really nicely done study conducted in everyday settings. The link to this paper (open access) is provided below. It is also featured in a podcast (tinyurl.com/4ydrw5zp) by the Psychonomic Society. Congrats!

Oh so it does have phd level intelligence

Between two trees: Vancouver man maps more than 2,600 places to hang a hammock He had the idea while hanging in a hammock.

When your city has an open data portal with the location of every tree, you can do this:

Writing into the Wreck I, for one, never asked to live in a collapsing empire. The feeling of what you thought to be solid ground turning to a shower of scree under your feet, the...

i wrote about what it feels like to live in the collapse of the american empire

"a flake of shale in a great upheaval of the earth"

buttondown.com/theswordandt...

I am privileged to announce the publication of the Anti-Autocracy Handbook: sks.to/autocracy 1/12

Interesting. This year is the first time ever I’ve received the student feedback ”language too complicated”. Could be the case of course, or it’s that the LLMs didn’t help them understand (in my case they actually said this) which then ”proved” it was too complicated?

Calon Lan Chessboxer · Apollo EP · Song · 2015

Came across it while trying to learn more about the band behind this great song - unfortunately seems they’re no longer active. 😢 Between chess, boxing, and acoustic Americana life gets pretty full I guess

open.spotify.com/track/46rTa2...

Chess boxing - Wikipedia

I’ve never been fully into neither chess nor boxing, but I’ve just learned of the existence of what seems like the perfect sport so now maybe I have to! Invented by Enki Bilal no less

en.m.wikipedia.org/wiki/Chess_b...

must be incredibly frustrating and disheartening to have federal funding that was promised to you for important work suddenly and arbitrarily ripped away

eg Nature Publishing is on this Uni of Leeds list: library.leeds.ac.uk/info/1406/re...

And congrats on the paper btw!

I’ve recently learned that many unis at least in the UK have sent formal ”notice of rights retention” to publishers with these kinds of embargos, effectively saying that we will publish our post-review manuscripts any time we damn please 😄

Maybe check whether your uni has done the same?

Willie Brown Middle School is a San Francisco 6-8 in the Bayview District. For the entire ‘24-‘25 school year, I have been driving the 44 O’Shaughnessy bus that stops there right after school gets out. That trip has been the busiest and most challenging part of my working day.

Ah an essential part of some good Gothenburg times. Seems like you’re having a nice stay

A hypothalamus–brainstem circuit governs the prioritization of safety over essential needs - Nature Neuroscience Animals need to adapt behavior to balance survival with fulfillment of essential needs. Krauth et al. identify neurons in the lateral hypothalamus that, when activated, prioritize survival over other ...

Excited to present our study in ‪@natneuro.nature.com‬: a deep brain pathway that triggers safety-driven movement—overriding needs like finding food or social contacts. A look into how the brain weighs deep instincts. Congrats to all and first author @nkrauth.bsky.social. shorturl.at/toYFN

Happy In Theory This is the short story of my long, 20 year search for a stable academic home. There is a lot of success and a lot of pain here, and no happy ending.

New Substack post

It's very personal: my story of a 20-year academic career, and the many challenges of theoretical and cross-disciplinary work

As I put it in the subtitle: There is a lot of success and a lot of pain here, and no happy ending

thomscottphillips.substack.com/p/happy-in-t...

Haha the reverse mullet theory of academic publishing

Great post, and can definitely be read as film noir voiceover in places

If you’re not reading Talia Lavin’s wonderful series about sandwiches, you should totally start.

And the singular haunts me, ”join our Teams team” no thanks

Hehe loving it, countering the 90s embodiment critique of 70s representationalism with 50s info theory 💪

Thanks, and agreed that all these framings are possible, and in one sense interchangeable, ie it’s all about usefulness in the end.

If you have an open and shut example showing why the info transfer framing is useful (more useful than alts?) do share - I can’t fully convince myself either way 😄

Truly good faith q: what’s the basis on which you feel this certain that the computational framing is correct?

I would tend to agree that it probably is, but I can also see the dynamical systems point, eg for a weather system we wouldn’t say it reads out the sea’s temperature as it passes over it

Agreed! But more as a final refactor than as an overall development strategy, I think

R3. Adopt the “single file” philosophy. Ideally, key workflows (e.g. model or evaluation definititions) should be defined in a single file. This has emerged as a powerful force for reproducibility in ML research (see Table 1 for examples). While this approach contradicts traditional software engineering principles of modularity and abstraction, it offers distinct advantages for research code: (1) complete understanding of an important component (e.g. how a model works) requires reading only one file, (2) minimal abstraction layers and dependencies, and (3) easy adaptation of either complete implementations or specific components. One can simply copy this single file into another codebase. Removing friction from future reuse also benefits us—as we can see from Table 1, these methods get a large number of citations.

CleanRL was a revelation for me, and I noticed that this paper (which Daphne pointed me to 😄) lauded the single file approach in general. It’s an interesting question I think - readability of a finished research product vs modularity and robustness of a codebase under dev?

arxiv.org/abs/2502.20349

Folks, always share your code. It doesn’t have to be perfect to be helpful. And if you feel that it’s still too messy or not sufficiently clean to be shared, you shouldn’t submit yet. After all, there could be mistakes in your mess.