Paul Linton

Great to present my work "Five Illusions Challenge Our Understanding of Visual Experience" at the European Conference on Visual Perception (#ECVP2025) Project Website + Preprint in link below 👇 @ItalianAcademy @KriegeskorteLab @ZuckermanBrain @columbiacss @ColumbiaScience

New paper out in Current Directions in Psychological Science making the case for studying generative behaviors: journals.sagepub.com/doi/10.1177/09…. (no paywall: cogtoolslab.github.io/pdf/fan_cdps_2…)

We have an opening for a Lead Research Radiographer to manage our portfolio of MRI scanners, including our 3T and 7T systems @UofGlasgow @UofGMVLS jobs.gla.ac.uk/job/lead-resea… DM me or @LarsMuckli for any informal enquiries.

Sad to miss #Cosyne2026 last minute. Our poster with Ken Miller is still available virtually, with a 2-min AI video overview + Q&A powered by @tavus , using data from @AllenInstitute. Do genetically-defined cell populations encode vision differently? toosi.github.io/research/cosyn…

1/5) 🚀 New paper accepted by ICLR 2026! “Towards interpretable visual decoding with attention to brain representations” Many thanks to my lovely co-authors and advisor @adeli_hossein @wenxuan_guo_ @LibraCheng @ethanhwang_ @KriegeskorteLab #CogCompNeuro #NeuroAI #GenAI #BCI

@andpru @ZuckermanBrain Thanks Andrew. Unfortunately not, but I think this requires Roland to give the talk (at least virtually) at Western as well!

@LintonVision @ZuckermanBrain That sounds super interesting, was it recorded by chance?

Absolutely wonderful talk by Roland Fleming on “Grasping: measuring and modeling how we use our hands to pick things up” at @ZuckermanBrain

@AkiyoshiKitaoka Amazing!! Shirt in honor of his pink illusion?

Anstis先生ｷﾀ―――(ﾟ∀ﾟ)―――― !!

@guido_maiello So sorry to hear Guido, sending lots of love to you and Sara ❤️

Dear friends, we have some difficult news. Sara’s cancer has progressed despite treatment, and we’re moving to palliative care. Thank you for standing with us through this. We love you.

Super excited that @jeremyakahn from @FortuneMagazine sat down with @DanAkarca and me to discuss our vision for how @CallosumAI can transform how we build intelligent systems across algorithms and hardware through co-evolution of the entire system level architecture!

excited to share some recent work! tldr; models trained on multi-view sensory data are the first to match human-level 3D shape perception—all zero shot, with no training on experimental data/images project page: tzler.github.io/human_multiview 1/🧠

🔍Truthfulness probes and their causal effects vary widely: some generalize, others are domain-dependent. Why? We propose Truthfulness Spectrum Hypothesis: truth directions of varying generality coexist! Probe geometry predicts generalization, and post-training reshapes it! 🧵⬇️

#VideoReason We are open-sourcing the entire VBVR stack to speed-up the arrival of video reasoning as the next fundamental paradigm of intelligence - 150+ synthetic generators - 1 million training clips - Cloud-scale data factory - Unified EvalKit - 100 rule-based evaluators - Strong baseline model Checkout at video-reason.com

Team 3D or Team No 3D?

@sethlazar Like the title of the little 2002 volume on the tutorial system, "Thanks, you taught me how to think": criticalthinking.org/files/OxCHEPS_…

One of the things I loved about Oxford as an undergraduate was just this feeling, that the tutors were never there to provide answers, but to challenge you to think more deeply about what were the right questions. I think in my entire degree I never wrote an essay to a prompt, but always followed my nose (including doing some digital humanities work [in 2001-2] almost before it was technically feasible to do so... AI would have been useful for *that*). I attended three lectures total; I didn't even have my allotted weekly meetings with tutors. I'd disappear for a couple of weeks, resurface with a paper, and talk about it for a couple of hours with a world-famous expert in the field. It was such a blast. The temptation to outsource one's thinking and to treat university as an answer machine, is a real challenge for us, now. But in another sense I think the challenge will be quite surmountable, and it points to the enduring value that universities can provide. It's surely clear to any motivated student, now, that allowing AI to write your outputs deprives you of almost everything that matters. There can *definitely* be ways in which AI can help as a research tool, and as an interlocutor, but writing is where you really wrestle with the problem, where you discover (especially in philosophy) the hidden premises in your half-articulated thoughts, which now require defence. If I'd done that as an undegrad, the tutorial would have been a walk of shame (and if my tutor outsourced grading to the AI, they'd have been equally embarrassed—though some had a higher toleration for that than others...) More generally, the central challenge that we're seeing with the ability to get AI agents to produce research outputs of many different kinds is that there's no plug in the back of our heads that allows us to just do a Neo in 'The Matrix' and instantly cognise all that information and analysis that has been generated. In 'Accelerando' the protagonist can fork his identity into an AI agent, send it off to do something, and then merge the branch back in with his principal cognition. But that isn't possible for us (whether it would be a good thing if it ever was is an interesting question). If you're in a domain where all you care about is the product (some forms of software writing, for example), then that might be fine. But anybody who wants to be a thinker is going to know that that isn't something that can be outsourced. And there will always be people who want to be thinkers (after all, Aristotle wasn't completely off the mark when he described it as one of the central elements in the good life). And society will continue to need institutions whose primary function is to teach young people how to think.

I have to admit that I haven't got my head around how this is possible yet. Seemingly very high levels of 3D consistency, generated on the fly. Amazing.

@AjdDavison This is super cool: x.com/bilawalsidhu/s…

Can't wait to watch this! Mike has been doing some absolutely fascinating work on consciousness: cell.com/current-biolog…

Tackling the big philosophical questions!

Psychophysics to understand the visual cortex: frontiersin.org/journals/syste… - another paper by a great group from Tehran whose webpage is gone now.

Grazie mille Italy in New York!