My research focuses on a computational understanding of the neural mechanisms underlying natural intelligence in vision and language. To achieve this goal, I bridge Deep Learning, Neuroscience, and Cognitive Science, building artificial neural network models that match the brain’s neural representations in their internal processing and are aligned to human behavior in their outputs.
I completed my PhD at the MIT Brain and Cognitive Sciences department with Jim DiCarlo, following Bachelor’s and Master’s degrees in computer science at TUM, LMU, and UNA. Previous work includes research in human-like vision at Harvard, natural language processing + reinforcement learning at Salesforce, as well as several other projects in industry and two startups. My work has been recognized in the news at Science magazine, MIT News, and Scientific American.
I am currently a tenure-track assistant professor at EPFL at the Neuro-X institute, with appointments at the School of Life Sciences, and the School of Computer and Communication Sciences.

New paper: a simple untrained model of language in the brain

© 2024 EPFL

In 2021 we were surprised to find that untrained language models are already decent predictors of activity in the human language system (http://doi.org/10.1073/pnas.2105646118). Badr Alkhamissi in the lab figured out the core components underlying the alignment of untrained models: tokenization and aggregation. With these findings, we built a simple untrained network “SUMA” with state-of-the-art alignment to brain and behavioral data — this feature encoder provides representations that are then useful for efficient language modeling. Directly mapping our model onto the brain, these results characterize the human language system as a generic feature encoder that aggregates incoming sensory representations for downstream use. If you disagree we hope you consider breaking our model (soon on Brain-Score/Github).


See here for social media posts:

5 abstracts accepted to CCN

© 2024 EPFL

My group will present 5 abstracts at the Cognitive Computational Neuroscience conference at MIT in Boston this fall! The projects cover new models of vision and language, new ways to evaluate these models on their brain alignment, and ideas to make use of the best models.

Current DNNs are Unable to Integrate Visual Information Across Object Discontinuities

Topographic Deep ANN Models Predict the Perceptual Effects of Direct IT Cortical Interventions

A Simple Untrained Recurrent Attention Architecture Aligns to the Human Language Network

Inferotemporal Cortex Underlies Primate Generalization Capabilities and Brain-Aligned Models Generalize Better

Scaling Laws for Task-Optimized Models of the Primate Visual Ventral Stream

Driving and Suppressing the Human Language System

We previously found GPT (2) to be a strong model of the human language system (pnas.org/doi/10.1073/pn).

In a new paper lead by Greta Tuckute, we push on this further and test how well model-selected sentences can modulate neural activity. Turns out you can almost double/completely suppress relative to baseline.
A subtle finding in this work that I find really interesting:
Under reasonable assumptions of inter-subject noise, prediction accuracy of neural activity is ~70% as good as it could possibly be. So even with these edge-case stimuli, gpt2-xl accounts for over 2/3 of the variance in the human language system.

Faculty position at EPFL

Very excited to finally make this public: I will be joining EPFL in the summer of 2023 to establish a research group focused on brain-like models of vision and language, and potential applications of these models. My appointment has been confirmed by the ETH Rat today.
The focus of my group will be:
  1. Building better (= more behaviorally and neurally aligned) models of vision and language. To build these models, we will use neural recordings from non-human primates and humans as well as human behavioral benchmarks, from the Brain-Score platform [perspective papertechnical paper, CORnet model, VOneNet modellanguage paper].
  2. Integrating multimodal representations: I think there is a lot of power in shared invariant representations; creating models that go all the way from pixel input to downstream language representations would allow us to potentially harness shared representations between these two domains, and test if we can better map such a model onto brain hierarchy.
  3. Towards clinical translation: in my mind, one of the end goals of these brain-modeling efforts is to apply them and improve people’s lives. This could involve helping blind patients (preliminary work), or people with dyslexia.
I will be hiring at all levels, please get in touch if these topics sound interesting to you.

Wiring Up Vision paper Spotlight at ICLR 2022

Our paper on reducing the number of supervised synaptic updates in computational models of vision was accepted to ICLR as a Spotlight! https://openreview.net/forum?id=g1SzIRLQXMM

The paper improved quite a bit since the preprint I think, we especially made a stronger connection to Machine Learning by showing that our proposed techniques outperform other approaches to drastically reduce the number of parameters. We retain over 40% ImageNet top-1 performance with only ~3% of parameters relative to a fully-trained network.

2022 Brain-Score Competition

We are excited to announce that submissions to the 2022 Brain-Score competition are open until February 15, 2022!

The first edition of the Brain-Score Competition proposes to evaluate computational models of primate object recognition in over 30 neuronal and behavioral benchmarks and will award $6,000 to the best submissions over three tracks: overall Brain-Score, V1, and object recognition behavior. In addition, selected participants will be invited to present their work in a Cosyne workshop which will feature some of the leading experts in vision neuroscience and computer vision.

For more information, please visit the competition website, follow Brain-Score on twitter, and join our Slack workspace! Good luck!


Our virtual ThreeDWorld is now public: www.threedworld.org

We provide a fully controllable virtual world striving to be ~photorealistic, based on the Unity engine. ThreeDWorld provides visual and audio rendering with physically realistic behavior that users can interact with through an extensive python API. Check out the code here: github.com/threedworld-mit/tdw

MIT News also wrote a great article summarizing the platform: news.mit.edu

Artificial Neural Networks Accurately Predict Language Processing in the Brain

Computational neuroscience has lately had great success at modeling perception with ANNs – but it has been unclear if this approach translates to higher cognitive systems. We made some exciting progress in modeling human language processing https://www.biorxiv.org/content/10.1101/2020.06.26.174482v1.
This work is the result of a terrific collaboration with Idan A. Blank, Greta Tuckute, Carina Kauf, Eghbal A. Hosseini, Nancy Kanwisher, Josh Tenenbaum and Ev Fedorenko.

Work by Ev Fedorenko and others has localized the language network as a set of regions that support high-level language processing (e.g. https://www.sciencedirect.com/science/article/pii/S136466131300288X) BUT the actual mechanisms underlying human language processing have remained unknown.

To evaluate model candidates of mechanisms, we use previously published human recordings: fMRI activations to short passages (Pereira et al., 2018), ECoG recordings to single words in diverse sentences (Fedorenko et al., 2016), fMRI to story fragments (Blank et al. 2014). More specifically, we present the same stimuli to models that were presented to humans and “record” model activations. We then compute a correlation score of how well the model recordings can predict human recordings with a regression fit on a subset of the stimuli.
Since we also want to figure out how close model predictions are to the internal reliability of the data, we extrapolate a ceiling of how well an “infinite number of subjects” could predict individual subjects in the data. Scores are normalized by this estimated ceiling.

So how well do models actually predict our recordings? We tested 43 diverse language models, incl. embedding, recurrent, and transformer models. Specific models (GPT2-xl) predict some of the data near perfectly, and consistently across datasets. Embeddings like GloVe do not.
The scores of models are further predicted by the task performance of models to predict the next word on the WikiText-2 language modeling dataset (evaluated as perplexity, lower is better) – but NOT by task performance on any of the GLUE benchmarks.
Since we only care about neurons because they support interesting behaviors, we tested how well models predict human reading times: specific models again do well and their success correlates with 1) their neural scores, and 2) their performance on the next-word prediction task.
We also explored the relative contributions to brain predictivity of two different aspects of model design: network architecture and training experience, ~akin to evolutionary and learning-based optimization. (see also this recent work). Intrinsic architectural properties (like size and directionality) in some models already yield representational spaces that – without any training – reliably predict brain activity. These untrained scores predict scores after training. While deep learning is mostly focused on the learning part, architecture alone works surprisingly well even on the next-word prediction task. Critically for the brain datasets, a random embedding with the same number of features as GPT2-xl does not yield reliable predictions.

Summary: 1) specific models accurately predict human language data; 2) their neural predictivity is correlated with task performance to predict the next word, 3) and with their ability to predict human reading times; 4) architecture alone already yields reasonable scores. These results suggest that predicting future inputs may shape human language processing, and they enable using ANNs as embodied hypotheses of brain mechanisms. To fuel future generations of neurally plausible models, we will soon release all our code and data.

Wiring Up Vision: Minimizing Supervised Synaptic Updates Needed to Produce a Primate Ventral Stream

Certain ANNs are surprisingly good models of primate vision, but require millions of supervised synaptic updates — this unbiological development has been the recent focus of many discussions in neuroscience. Is all this training really necessary? We approach this in new work https://www.biorxiv.org/content/10.1101/2020.06.08.140111v1.

Neuroscientists have argued for innate structure with only thin learning on top, i.e. where structure the genome dictates brain connectivity and is leveraged for rapid experience-dependent development. We took first steps at this with more brain-like neural networks.

We started from CORnet-S, the current top model on neural and behavioral benchmarks in Brain-Score.org. We first found that variants of this model which are trained for only 2% of supervised updates (epochs x images) already achieve 80% of the trained model’s score.

Even without any updates, the models’ brain predictivities are well above chance. Examining this “at-birth” synaptic connectivity and improving it with a new method “Weight Compression”, we can reach 54% without any training at all

However, to be more brain-like we require at least some training — but ideally this would not change millions of synapses requiring precise machinery to coordinate the updates. By training only critical down-sampling layers, we achieve 80% when updating only 5% of synapses.

Applying these three strategies in combination (reducing supervised epochs x images + improved at-birth connectivity + reducing synaptic updates), we achieve ~80% of a fully trained model’s brain predictivity with two orders of magnitude fewer supervised synaptic updates.

Taking a step back, we think these are first steps to model not just primate adult visual processing during inference, but also how the system is wired up from an evolutionary birth state encoded in the genome and by developmental update rules. Lots more work to do!

My development setup

Since this has been helpful to many colleagues, I thought I would put together my development setup and especially include a small how-to on setting up PyCharm with remote support.

PyCharm remote setup

I have found the remote support of PyCharm to be extremely helpful. It is only available in the professional version, but you get it for free if you’re a student.

The pros are extremely appealing: You can run everything on a remote server while the development experience feels as if it were local. PyCharm takes care of uploading changes, running the remote interpreter (even in debug mode), and displaying figures directly in the IDE.

Follow these steps to setup PyCharm in the same way:

  1. open File > Settings, navigate to Project > Python interpreter
  2. On the right side of the “Python Interpreter” line, click the gear icon and click “Add…”
  3. select “SSH Interpreter” and create a new server configuration. Enter the host and your username, click next
  4. Enter your password or private key, click Next
  5. Enter the path to the remote interpreter. I use a local miniconda installation, so for me the path is something like /braintree/home/user/miniconda3/bin/python.
  6. On the same screen, click the Folder icon to the right of “Sync folders” and input where you want your local files synced. Click Finish

That’s it! You can now program as usual in your IDE and files will automatically be transferred to the remote host. When you run/debug things, they will automatically run on the remote host. (In case it doesn’t sync, make sure your remote path is entered in Settings > Deployment > Mappings.)

Advanced tips

  • Running fig.show() with a matplotlib figure will show it directly in the IDE
  • For multiple attached projects, you can go to Settings > Build, Execution, Deployment > Deployment > (in view) Mappings to setup the local/remote path pairs
  • Make sure to use the integrated Version Control tools. If they don’t show up automatically, check that the directory is recognized under version control in Settings > Version Control
  • Use auto-formatting! PyCharm can automatically reformat your code according to a specific (or default) code style
  • Use hotkeys! They make it much quicker to accept suggestions, auto-format, run, step into etc. (I use the Eclipse keymap out of habit)
  • To install miniconda locally, I usually run wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh && bash Miniconda3-latest-Linux-x86_64.sh.


  • SurfaceBook laptop on a stand
  • 2 external 24″ monitors
  • Kinesis Freestyle2 split keyboard
  • wireless mouse
  • external webcam
  • wireless headset
  • SurfaceDock to bring it all together with minimal daily cable management



Oral presentation at NeurIPS

The work led jointly with Jonas Kubilius on “Brain-Like Object Recognition with High-Performing Shallow Recurrent ANNs” was accepted to NeurIPS as an oral! Only 36 (or 0.5%) out of the 6743 submissions were selected as an oral, so we’re very excited to present our ideas how Machine Learning and Neuroscience can interact again in the form of models of the brain.

The McGovern institute and Robin.ly have also reported about our findings.

Efficient Neural Stimulation with In-Silico Stimulation, class project

For my class project in Alan Jasanoff’s course Neurotechnology in Action, I experimented with modeling the behavioral effects of microstimulation. I used Afraz et al., Nature 2006 as a reference and started from our recently developed CORnet model (Kubilius*, Schrimpf* et al. NeurIPS 2019). By increasing the activity at parts in the model, I was able to reproduce the behavioral effects reported by Afraz et al. to a first extent. On top of that, I used the model to predict which improvements in neurotechnology hardware would best support behavioral control. I think modeling the behavioral effects of neural perturbations is an exciting direction and I’m hoping to thereby bring the models closer to clinical applications. For more details, please see the paper.

Page 1 Page 2