←
Victoria Bosch152 posts
Victoria Bosch avatar

Victoria Bosch

@__init_self

neuromantic - ML and cognitive computational neuroscience - PhD candidate at Kietzmann Lab, Osnabrück University. 🦋

◷ Joined October 2021▧ 14 Photos and videos
Victoria Bosch avatar
Victoria Bosch@__init_self·3 Nov 2025

Introducing CorText: a framework that fuses brain data directly into a large language model, allowing for interactive neural readout using natural language. tl;dr: you can now chat with a brain scan 🧠💬 1/n

Post attachment
Mark Histed avatar
⟳ Victoria Bosch retweeted
Mark Histed@HistedLab·Oct 17

I have a lot of respect for @nickcammarata and Goodfire, but I will add some thoughts from brains/neuroscience. First, polysemanticity and distributed representations are ~widely~ seen in biological brains. (In some ways this is what I have spent my career on.) 1/

interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works already-clean.goodfire.com/

Akshay K. Jagadish avatar
⟳ Victoria Bosch retweeted
Akshay K. Jagadish@akjagadish·Oct 7

Delighted to share that this work is now accepted to @NeuroCellPress's Neuroview! 🎉 Key changes: [1] considering wider variety of simulators, behavioral, neural and task-specific cognitive models [2] highlighting the use of neural and other physiological measurements for calibration and adjudication. [3] expanding on the weaknesses of current methods (e.g. bias towards WEIRD population in dataset used to derive foundation models) For more details, check the full article ⬇️

Post attachment

1/ 🚨 New opinion piece: "Can we automatize scientific discovery in the cognitive sciences?" 👥 in collaboration with @milenamr7, @WitteKristin, @marvinmathony, @marcel_binz & @cpilab We lay out a vision for a fully automated, in-silico science of the mind, where modern AI systems run every stage of the scientific discovery cycle in cognitive science 🧵 #CognitiveModeling #AutomatedDiscovery #LLMs #AI4Science

James Whittington avatar
⟳ Victoria Bosch retweeted
James Whittington@jcrwhittington·Aug 12

We’re excited to announce DiG-bench, a new benchmark for discovery! Over the last few weeks we’ve been testing frontier AI models on our novel discovery games and seeing how they score. Each game is a text-based environment, so they probe discovery capabilities in the natural domain of language models, rather than requiring additional, potentially confounding, visual understanding. TL;DR frontier models have improved a lot over the last few months. But they are still stumped by some surprisingly simple problems, even in their native text domain. With @cocosci_lab (@Princeton) @MITCoCoSci (@MIT) @SchmidhuberAI (@KAUST_News) @misovalko (@Inria) @tri_dao (@PrincetonCS) @RMBattleday @zebkDotCom @FraserGreenlee @akaijsa @ClareMaguire @TimMuller1 @kubicek_ales @physicscat0x7d @SukritSumant @thoughtchannel_ (1/5)

Post attachment
Nikolaus Kriegeskorte avatar
⟳ Victoria Bosch retweeted
Nikolaus Kriegeskorte@KriegeskorteLab·Aug 28

To learn how brains compute, we need to experimentally adjudicate among competing computational hypotheses. How do we do this in the age of complex neural network models? New review paper: “Making models disagree to learn how brains compute” with @TalGolanNeuro, @SchuettHeiko

Tal Golan avatar
⟳ Victoria Bosch retweeted
Tal Golan@TalGolanNeuro·Aug 28

How can we design experiments that make models disagree? One section of our new @NatRevNeurosci Review with @KriegeskorteLab and @SchuettHeiko examines studies that used stimulus sets designed to elicit distinct predictions from competing models. Full text link at the end. 1/14

Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

Our article "The Umwelt Representation Hypothesis: rethinking Universality" is out now in @TrendsCognSci , with my amazing colleagues: @RowanSommers @AdrienDoerig @TimKietzmann Many thanks to the journal and reviewers for their thoughtful feedback! cell.com/trends/cognitive-sc…

Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

Our proposal: To understand representational alignment between artificial neural networks and brains, don't look for the *one world model*, but use the toolbox of neuroconnectionism to map the space of ecological constraints that lead to systematic (mis)alignment instead.

Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

The URH: representations are shaped by what a system can sense, act on, and cares about: its Umwelt. Alignment is due to overlap of ecological constraints, not convergence to a veridical world model. Universality can't explain both similarity and systematic differences.

Post attachment
Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

A bee can see UV patterns on flowers. Dogs have a better sense of smell, but are dichromatic. Some people see The Dress as black/blue, others white/gold... Are these differences "world modeling failures"? We propose an alternative: the Umwelt Representation Hypothesis (URH).

Post attachment
Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

Universality assumes a globally optimal world model that all systems strive for. As a consequence, misalignment is due to failure of world modelling; similarity is universal, differences are idiosyncratic noise (Anna Karenina scenario).

Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

This led to the proposal of Universality: different information-processing systems (ANNs and brains) all converge upon the same representational structures, because they learn a shared model of reality. A prime example of this is the Platonic Representation Hypothesis.

Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

Background: Recent studies find a striking alignment between a variety of ANNs, despite differences in architecture, objective, and even across language and vision (e.g. Huh et al. 2024). This finding expands to neural representations too (see e.g. Chen & Bonner, 2025)!

Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

Check out our latest paper: "The Umwelt Representation Hypothesis: rethinking Universality". Tl;dr: we argue that representational alignment emerges from overlap in ecological constraints, not convergence to one privileged world model. cell.com/trends/cognitive-sc…

Victoria Bosch avatar
Victoria Bosch@__init_self·Aug 28

Are brains and artificial neural networks converging onto universal representations? There is a seductive idea making the rounds in NeuroAI / ML: train systems well enough, and they'll converge on the same representation of reality (i.e. unique world model). We have Thoughts™