Giacomo Indiveri

This paper shows alignment between LLMs and brain data is outperformed by a null model. More evidence for the argument I've been making in talks lately that we shouldn't believe any computational paper that puts less effort into null than main model.

www.biorxiv.org/content/10.1...

We're organising a "Neuromorphic hardware and algorithms" workshop at Sussex on 11th-12th of June 2026. Expect keynotes by Chiara Bartolozzi and André van Schaik as well as awesome ECR speakers. (Free) registration now open at genn-team.github.io/workshop2026/. Accepting abstracts until March 31st!

state of the LLMs: Measuring the impact of Qwen, DeepSeek, Llama, GPT-OSS, Nemotron, and all of the new entrants to the ecosystem

Impressive tipping point that happened during 2025

www.interconnects.ai/p/8-plots-th...

Some things we learned: 1. CA1 pyramidal neuron dendritic voltage dynamics during behavior are low-dimensional, well described by just two or three compartments (basal, soma, apical). This rules out models in which dendritic branches act as distinct computational compartments.

We're ending 2025 with a bang! I'm really proud to be part of the amazing team that produced this comprehensive perspective on "How #heterogeneity shapes dynamics
and computation in the #brain" (and in #neuromorphic systems, I add 😇):
authors.elsevier.com/sd/article/S...

No worries! It was just to reiterate that we should combine forces with joint projects in the same direction 😉
Hopefully we'll find some occasion in 2026...

A canonical cortical electronic circuit for neuromorphic intelligence Cortical microcircuits play a fundamental role in natural intelligence. While they inspired a wide range neural computation models and artificial intelligence algorithms, few attempts have been made t...

Dear Mackenzie, congratulations! It's a great piece! And I was happy to see how many things in common we have...
For example, see your Fig.2 with Fig.1 of doi.org/10.1101/2025...

For what concerns the general arguments, I wrote similar comments also here: www.pnas.org/doi/10.1073/... 🙂

I’m very grateful to the FMI, the tenure committee, inspiring colleagues, and all the hidden supporters who made this possible. Huge thanks to past and present group members for their curiosity and creativity. Excited for the next chapter.

Always such a vibrant NeuroAI community in Montreal !

Yet another commentary (and rant) on why following the original #neuromorphic approach, using *also* #analog circuits to emulate the physics of cortical circuits, can be fascinating:
www.pnas.org/doi/10.1073/...

In our latest D-ITET professor series interview, Prof. Laura Bégon-Lours shares insights into neuromorphic computing and ferroelectric oxides – and she explains why these technologies could transform remote devices, wearables, and smart implants.

Read the full interview: ee.ethz.ch/news-and-eve...

Excellent research on Time Modulated Arrays: Prof. Hua Wang’s Integrated Devices, Electronics, and Systems (IDEAS) Group has received the Top-10 Key Achievement Award from the Smart Networks and Services Joint Undertaking in 2025. Congratulations!

Read more: ee.ethz.ch/news-and-eve...

And it's even worst for the term "#neuromorphic" 😉

Inhibitory Plasticity Balances Excitation and Inhibition in Sensory Pathways and Memory Networks Plasticity at inhibitory synapses maintains balanced excitatory and inhibitory synaptic inputs at cortical neurons.

"Spiking Networks Hate It! Find Out the One Plasticity Trick They Don’t Want You to Know! Never stabilise models by hand again." - I woke up thinking we missed an opportunity with the title of this one. :/ www.science.org/doi/10.1126/... Also: It snowed in Vienna, 10cm white fluffies! Happy Sunday!

With my great advisors and colleagues, @achterbrain.bsky.social @zhe @danakarca.bsky.social @neural-reckoning.org, we show that if heterogeneous axonal delays (imprecise) can capture the essential temporal structure of a task, spiking networks do not need precise synaptic weights to perform well.

Scared presenter in a large conference room that might fill up with colleagues in a few minutes

Honored (and scared) of being invited to give the keynote lecture on #neuromorphic for NER at IEEE EMBS 12th Annual International Conference on #NeuralEngineering (neuro.embs.org/2025/)

Slides available here: bit.ly/3Jspfek

Oct 30, 4:30
"How does the Geometry of Brain Activity Shape Behavior?"
Valeria Fascianelli; moderator Stefano Fusi, Zuckerman Institute, Columbia.
Open seminars series; register here: tinyurl.com/379uda2z
@valeriafascianelli.bsky.social ,
@columbiauniversity.bsky.social, @stefanofusi.bsky.social

Inspiring perspective paper by Prof. Giacomo Indiveri in Neuron. In this interview, he talks about the term “neuromorphic”, the uniqueness of the Institute of Neuroinformatics and comparisons with artificial intelligence.

Read more: ee.ethz.ch/news-and-eve...

@giacomoi.bsky.social

Neuromorphic dreaming as a pathway to efficient learning in artificial agents

https://www.doi.org/10.1088/2634-4386/ae0a77

Schematic of awake-dreaming learning. (a) Awake phase: over 100 real frames, the agent and model networks interact with the Pong environment. (b) Dreaming phase: over 50 imaginary frames, the agent is disconnected from the real environment and instead interacts only with the world model. This alternation between real and simulated experience boosts sample efficiency.

#Spiking #neural #networks ( #SNN ) running on continuous-time, noisy, and highly variable computing substrates can learn reliably with #ReinforcementLearning ... Not only in real brains, but also in mixed-signal #neuromorphic hardware! 😇

Schematic of awake-dreaming learning. (a) Awake phase: over 100 real frames, the agent and model networks interact with the Pong environment. (b) Dreaming phase: over 50 imaginary frames, the agent is disconnected from the real environment and instead interacts only with the world model. This alternation between real and simulated experience boosts sample efficiency.

#Spiking #neural #networks (#SNN) running on continuous-time, noisy, and highly variable computing substrates can learn reliably with #ReinforcementLearning ... Not only in real brains, but also in mixed-signal #neuromorphic hardware! 😇

Neuromorphic dreaming […]

[Original post on fediscience.org]

#Spiking neural networks running on continuous-time, noisy, and high;y variable computing substrates can learn reliably... Not only in real brains. Also in mixed-signal #neuromorphic hardware 😇

Neuromorphic dreaming as a pathway to efficient learning in artificial agents
www.doi.org/10.1088/2634...

A thought-provoking perspective from the visionary @giacomoi.bsky.social, calling for neuromorphic computing to return to its root: fundamental neuroscience; an inspiring vision for the future of NeuroAI 🤩

What is this "neuromorphic" hype all about?
Here's my (neuro)view on it:
https://authors.elsevier.com/a/1ltRg3BtfHC-Jf

Here's the free version:
authors.elsevier.com/a/1ltRg3BtfH...

Neuromorphic is dead. Long live neuromorphic In this NeuroView, Giacomo Indiveri discusses the origins and evolution of neuromorphic engineering, reflects on recent shifts toward digital and AI-centric approaches, and advocates for a revival of ...

Curious about the meaning of "neuromorphic"?
Here's my latest NeuroView on this topic:
www.cell.com/neuron/fullt...

Exciting news! Prof. Dr. Melika Payvand is joining D-ITET as Tenure Track Assistant Professor of Autonomous Intelligent Agents. Congratulations!

Yeah robots are definitely not taking over the world.

Not the red one, at least. 🤣