Martin Hebart
@martin_hebart
Followers
5K
Following
4K
Media
291
Statuses
3K
Proud dad, Prof. of Computational Cognitive Neuroscience, author of The Decoding Toolbox, founder of https://t.co/hWZCF7XuMU @ martinhebart.bsky. social
JLU Giessen/Max Planck Leipzig
Joined June 2015
I'm thrilled to announce the THINGS initiative: An initiative of researchers around the world collecting and sharing large-scale behavioral and neuroscience data for object recognition and understanding, using the same image dataset. https://t.co/wYEw1ssSfl
things-initiative.org
Large-scale behavioral and neuroscience data for object recognition and understanding.
12
91
337
Just as I want a plane to always work, I want AGI to work reliably and to generalize beyond benchmarks to the real world. So, while CHC theory is useful in humans, and using insights from psychology is the right direction, I'd be careful defining AGI through human psychometrics.
0
0
4
Introducing - Why Solana? The SOL thesis for Wall Street ↓
68
96
402
Now let's apply the authors' logic to the concept of flight. The Wright Brothers' plan flew for 12 seconds but it took decades to achieve reliable flight! With this analogy, would you rely on a plane like that of the Wright Brothers? I wouldn't (unless I was up for an adventure)!
1
0
1
The factors of the model are derived using factor analysis, a method that finds directions where different people vary a lot (like "reasoning"). But the catch is, this excludes items that most people are good at! So, the "AGI" could perform terribly on important human abilities!
1
0
1
Cattell-Horn-Carroll theory of intelligence is not a theory. It provides a description - a taxonomy - of how one can think of human intelligence. It also does not map neatly onto a neuroscientific or neuropsychological basis. Since it is descriptive, it is not testable.
1
0
1
God doesn't accept sidlers. But, He welcomes lost, helpless sinners. Are you a sidler or a sinner?
1
6
109
The idea of using insights from psychology research to define AGI are laudable, but this approach is wrong at so many levels, and adding a leading expert in psychometrics and an AI critic doesn't overcome the fundamental issues with this paper. My 2ct on what is wrong with it: 🧵
The term “AGI” is currently a vague, moving goalpost. To ground the discussion, we propose a comprehensive, testable definition of AGI. Using it, we can quantify progress: GPT-4 (2023) was 27% of the way to AGI. GPT-5 (2025) is 58%. Here’s how we define and measure it: 🧵
1
0
13
2day Prof Martin Hebart @martin_hebart @jlugiessen on "Beyond category: Revealing core representational axes of natural objects in behavior, brains, and deep neural networks" https://t.co/nSfBhaOeGB
@worldwideneuro
@ubarilan
@GondaBrain
@goodmanfaculty
#BIUVisionScienceSeminar /2
1
1
2
P.S.: The list of authors obviously reflects only a small and not representative part of the ever-growing community, and I hope that nobody feels left out!
0
0
2
Was really fun playing a (small) part in this and enjoyed the discussions in the revisions!
🧵🎉 Our mega-paper is finally published in TMLR! We're "Getting Aligned on Representational Alignment" - the degree to which internal representations of different (biological & artificial) information processing systems agree. 🧠🤖🔬🔍 #CognitiveScience #Neuroscience #AI
1
1
17
After two years of engagement with Apple, American Family Association successfully convinced the tech giant to strengthen child protection across its platforms. Apple will now block explicit content in iMessage for ALL minors and remove adult-rated app listings for kids in the
1
9
26
Predictive coding has guided neuroscience for years but it does not account for the neuronal data. We review how patterns of feedback during spatial and temporal predictions are better captured by a family of opposing theories, collectively termed BELIEF. https://t.co/QU22KK7h7y
1
21
82
They tell you neural nets are non-linear. What does "linear" even mean?! Linearity is only defined given two vector spaces, X → Y. What if we could find a different pair of spaces where NNs ARE linear? 🤯 We do it and use it for many apps, such as one-step diffusion! 🧵
22
63
560
🧠 New preprint: we show that model-guided microstimulation can steer monkey visual behavior. Paper: https://t.co/ZJm8xHJFXm 🧵
arxiv.org
Brain stimulation is a powerful tool for understanding cortical function and holds promise for therapeutic interventions in neuropsychiatric disorders. Initial visual prosthetics apply electric...
1
15
41
In NeuroAI there's been an assumption of a tradeoff between strictess and predictivity for brain-model mappings -- with stricter methods better able to identify true mechanisms, and maximizing predictivity requiring too much flexibility. We show this is false. Basically, it
1/x Our new method, the Inter-Animal Transform Class (IATC), is a principled way to compare neural network models to the brain. It's the first to ensure both accurate brain activity predictions and specific identification of neural mechanisms. Preprint: https://t.co/hPqo5PrZoc
2
7
50
Your mortgage, reimagined as a mission. Get cleared to close — the Total way.
0
0
7
Excited to share new work with @H_Lee_Masson, Ericka Wodka, @SHMostofsky and Leyla Isik! We investigated how simultaneous vision and language signals are combined in the brain using naturalistic+controlled fMRI. Read the paper here: https://t.co/vvmJOcjI4s 1/n
1
16
62
In a study, now out in Attention Perception & Psychophysics (by @Psychonomic_Soc), @ananyapassi and I have some cool insights about how parts combine in sound-shape associations (in the famous Bouba-Kiki effect). Read on to find out more! 1/12
1
7
29
Our article is out in Annual Review of Vision Science: “Visual Image Reconstruction from Brain Activity via Latent Representation” We trace the path from early brain decoding to modern NeuroAI, highlight progress & pitfalls, and discuss future directions
annualreviews.org
Visual image reconstruction, the decoding of perceptual content from brain activity into images, has advanced significantly with the integration of deep neural networks (DNNs) and generative models....
0
4
22
We’re hiring in my dept. (IEM @ Ben-Gurion Univ.)! Tenure-track faculty opening in Robotics, Data/ML, Info Systems, Stats, Production Systems, or Human Factors. https://t.co/5Us9JWQdB6 Please repost 🙏
1
8
7
Invideo just dropped Trends. It's like presets & templates to quickly generate videos for your ideas & brands for the Halloween. 8 wild examples: 1. Paranormal break-in CCTV & step by step
23
9
83
In this short letter we explain why binding problems occur in the brain and why deep neural networks need to cope with them. We respond to Scholte and de Haan (TICS 2025), who previously claimed the opposite.
2
8
35
I am pleased to share that together with the brilliant team of Ofer Lipman, @Doronfried1179 and @OrYacov from Reichman University and @ShanyGrossman from MPI for Human Development and Hamburg University, our study was just published in Nat Commun 🔗
1
3
26
Yay! Really happy that PhD student Malin Styrnal in our lab won both the Best Student Poster Award *and* the Poster of the Day Award at #ECVP2025 for her presentation "The similarity of similarity tasks: Comparing eight different measures of similarity"! (alas no photo of her!)
0
3
36
🧵 Can a purely feedforward network — with no recurrence or lateral connections — capture human-like perceptual organization? 🤯 Yes! Especially for contour integration, and we pinpoint the key inductive biases. New paper in @PLOSCompBiol with @talia_konkle & @grez72! 1/24
2
20
117
