Dirk Gütlin

I’m an NIH whistleblower. The scientific community cannot afford to avoid politics “Scientists are often encouraged to avoid politics. But this advice is outdated, if it was ever correct in the first place,” writes Jenna Norton, NIH whistleblower.

A heads down approach will not save science. We need to speak up, live our values, and demand integrity.

www.statnews.com/2026/03/12/n...

I personally would say your prose is fine. :)

Fortifying Robustness in Graph Neural Networks: A Loss Correction Approach to Mitigate Label Noise

ieeexplore.ieee.org/document/114...

#MLSky

Are they? How did you notice?

When ChatGPT is gone: Creativity reverts and homogeneity persists ChatGPT has been evidenced to enhance human performance in creative tasks. Yet, it is still unclear if this boosting effect sustains with and without ChatGPT. In a pre-registered seven-day lab experim...

I came across an interesting study today, it's not recent, but I'd never read it before. It found that ChatGPT boosts creative performance in the short term, but without AI this effect completely disappears. Moreover, content is becoming increasingly homogeneous,
arxiv.org/abs/2401.06816

What does it mean for culture to ‘shape’ cognition? Human culture and cognition vary widely across groups, but how exactly culture ‘shapes’ cognition remains underspecified. In this review, we outline four qualitatively different pathways by which cult...

What does it mean for culture to ‘shape’ cognition?

www.cell.com/trends/cogni...

The integrated information Φ of an integrate and fire network Author summary In this study, we explored how Integrated Information Theory (IIT), one of the leading theories of consciousness, can be applied to a small network of simulated neurons. Specifically, w...

The integrated information Φ of an integrate and fire network

journals.plos.org/ploscompbiol...

On consciousness in animals and in artificial intelligence | Journal of Neurophysiology | American Physiological Society In recent years, numerous publications have emerged claiming that animals possess consciousness. Moreover, neurophysiological literature has discussed the possible existence of consciousness in artificial intelligence (AI). Both ideas stem from analyses of the behavior of animals and AI systems. Here, I argue that it is impossible to draw conclusions about animal consciousness based solely on their behavior. Addressing the question of animal consciousness requires an understanding of its neuronal mechanisms, and the challenge lies in whether these mechanisms can be understood. Views on this issue diverge widely, ranging from Cartesian dualism, which denies any neuronal mechanisms of consciousness, to the opposing position that consciousness is one of cognitive functions whose underlying mechanisms can be understood. Here, I support the latter view. In contrast, the problem of AI consciousness belongs to the field of computer science rather than neurophysiology. Even if artificial intelligence were to acquire consciousness in the future, it would be an AI-specific form of consciousness bearing little relation to human consciousness. Consequently, insights into the mechanisms underlying AI consciousness are unlikely to advance our understanding of human consciousness.

On consciousness in animals and in artificial intelligence

journals.physiology.org/doi/abs/10.1...

#neuroskyence #MLSky

Affective Forecasting Accuracy in Everyday Life - Affective Science Affective Science - People often predict how they might feel in the future, with varying degrees of accuracy. Such affective forecasts can centre around periods of time (e.g., tomorrow, next week)...

In work recently out in Affective Science we investigated how accurately people can forecast their emotions in everyday life. Study 1 focused on forecasts for specific time periods (tomorrow, next week). Study 2 focused on forecasts for daily unpleasant events. link.springer.com/article/10.1...

Thanks for sharing our work!

For anyone interested, here's a short thread summarizing the main takeaways:

bsky.app/profile/nikl...

Compact deep neural network models of the visual cortex Nature - Parsimonious deep neural network models can be used for prediction of visual neuron responses.

DNN models of the brain are getting bigger. Are we replacing one complicated system in vivo with another in silico?

In new work, we seek the *smallest* DNN models of visual cortex, balancing prediction with parsimony.

It turns out these compact models are surprisingly small!

rdcu.be/e5H8G

Learning collision risk proactively from naturalistic driving data at scale - Nature Machine Intelligence Jiao et al. introduce a generalized safety measure for autonomous driving systems that learns collision risk from everyday driving without labels. It accurately warns in real time of crashes and near-...

Learning collision risk proactively from naturalistic driving data at scale

www.nature.com/articles/s42...

#MLSky

A neural signature of adaptive mentalization - Nature Neuroscience Successful social interaction relies on predicting others’ behavior. Buergi et al. combined computational modeling and brain imaging to show how humans adjust to changing strategies, revealing a neura...

A neural signature of adaptive mentalization

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

#neuroskyence

A reinforcement learning and sequential sampling model constrained by gaze data Author summary When people are deciding between options they have encountered in the past, their preferences are largely based on their memory of past experiences with those options. However, simply l...

A reinforcement learning and sequential sampling model constrained by gaze data

journals.plos.org/ploscompbiol...

Dynamic Representation of Multidimensional Object Properties in the Human Brain Our visual world consists of an immense number of unique objects and yet, we are easily able to identify, distinguish, and reason about the things we see within a few hundred milliseconds. Here, we us...

Dynamic Representation of Multidimensional Object Properties in the Human Brain

www.jneurosci.org/content/46/9...

For about a century, the US was the world's main supplier of a drug called soft propaganda TV entertainment.

This looks like some of them got high on their own supply and became delusional.

One donor accounts for 65% of Reform UK's 2025 donation income.

The party's average reported donation is £112k.

Not exactly a people's revolution.

Neural Tuning for Ordinal Processing: Convergent Patterns in Human Brains and Artificial Networks Processing ordinality, i.e., the rank of an item in a series such as 1st, 2nd, 3rd, etc., is a fundamental skill shared by humans and animals. While humans often use symbolic sequences like numbers or...

Neural Tuning for Ordinal Processing: Convergent Patterns in Human Brains and Artificial Networks

www.jneurosci.org/content/46/9...

#MLSky

Information theoretic measures of neural and behavioural coupling predict representational drift Author summary Activity in the brain represents information about the outside world and how we interact with it. Recent evidence shows that these representations slowly change day to day, while memori...

Out now in @plos.org CB: doi.org/10.1371/jour...
Neural representations of tasks change over time, even in the absence of changes in task performance. But neurons change tuning at different rates. How does a neuron's stability relate to its interactions with the population?
#compneuro #neuroskyence

High-fidelity neural speech reconstruction through an efficient acoustic-linguistic dual-pathway framework Concurrent decoding of acoustic detail and linguistic structure enables natural, intelligible speech synthesis from limited human cortical recordings, resolving a fundamental constraint in neural spee...

High-fidelity neural speech reconstruction through an efficient acoustic-linguistic dual-pathway framework

elifesciences.org/articles/109...

🧠🤖🧠📉

Multimodality as a safeguard of honesty in communication and language: from Animals to Humans

www.cambridge.org/core/journal...

📢 PhD position in the NeuroAI of Language

Why can LLMs predict brain activity so well? We're hiring a PhD student to find out -- AI interpretability meets neuroimaging
Deadline March 20
Please RT 🙏
👇
mpi.nl/career-education/vacancies/vacancy/fully-funded-4-year-phd-position-neuroai-language

I am totally pumped about this new work . "Task-trained RNNs" are a powerful and influential framework in neuroscience, but have lacked a firm theoretical footing. This work provides one, and makes direct contact with the classical theory of random RNNs:
www.biorxiv.org/content/10.6...

Dissociable dynamic effects of expectation during statistical learning Neural responses during statistical learning reveal dissociable dynamic effects of expectation, supporting the opposing process theory within trials while demonstrating contrasting effects across tria...

🚨New paper alert 🚨

Using EEG and decoding analyses we find dynamic effects of expectation during statistical learning, both within and across trials 🧠

@auksz.bsky.social @fedemar.bsky.social
Caspar Schwiedrzik

elifesciences.org/articles/103...

AI Translations Are Adding ‘Hallucinations’ to Wikipedia Articles AI translated articles swapped sources or added unsourced sentences with no explanation, while others added paragraphs sourced from completely unrelated material.

Wikipedia editors investigated how OKA was operating and found that it was mostly relying on cheap labor from contractors in the Global South, and that these contractors were instructed to copy/paste articles to popular LLMs to produce translations.

Framework for hierarchical deep reinforcement learning with conceptual embedding Deep reinforcement learning (DRL) faces challenges when the combinatorial state-action space becomes excessively large. Hierarchical DRL (HDRL) provid…

Framework for hierarchical deep reinforcement learning with conceptual embedding

www.sciencedirect.com/science/arti...

Recurrent cortical networks encode natural sensory statistics via sequence filtering The visual cortex receives a stream of high-dimensional sensory input. The role of dense local, recurrent cortical connections in shaping responses to these inputs has been unclear. Here, we show that...

Recurrent cortical networks encode natural sensory statistics via sequence filtering

www.cell.com/neuron/fullt...

#compneuro #neuroskyence

The generalized Hierarchical Gaussian Filter

elifesciences.org/reviewed-pre...

#stats #cogpsy

Orbitofrontal cortex drives predictive filtering of sensory responses - Nature Neuroscience Top-down projections from the orbitofrontal cortex carry predictive signals that grow with sound experience and suppress the auditory cortex via inhibitory circuits, revealing a predictive mechanism f...

Orbitofrontal cortex drives predictive filtering of sensory responses

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

#neuroskyence

The effects of political advertising on Facebook and Instagram before the 2020 US election - Nature Human Behaviour Allcott et al. conducted an experiment in 2020, removing political ads from the feeds of randomly selected Facebook and Instagram users. There were no statistically significant effects on political ou...

The effects of political advertising on Facebook and Instagram before the 2020 US election

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