Tag
Demis Hassabis announces that AI can now reconstruct what people are dreaming from brain scans, predicting sci-fi-like devices in the next few years.
Scientists recorded individual neurons in bilingual brains for the first time and found that the brain does not translate words using shared neurons but instead organizes each language into a geometric map of meaning with the same structure, similar to vector space isomorphism in LLMs.
Microsoft Research and collaborators introduce generative causal testing (GCT), a method that distills black-box brain prediction models into testable explanations and validates them with fMRI experiments, revealing specific brain region responses to language concepts.
This paper presents CALHippo, a framework for 3D mapping of neurons and glial cells in the human hippocampus using state-of-the-art segmentation and density estimation models.
The author describes building a spiking neural network engine that initially failed the NARMA-10 benchmark, but by applying heterogeneous wire delays from neuroscience, it achieved usable memory depth and a 15x computational efficiency advantage over continuous nets on a recognition task.
This paper reports that slow breathing can modulate brain function and influence risk-taking behavior.
A newsletter covering two main stories: the practical engineering challenges of solar geoengineering, and new research into interoception and its health implications. Also includes news of SpaceX's valuation surpassing Amazon.
Oswald Steward and colleagues won the 2026 Kavli Prize in Neuroscience for discovering that neurons can produce proteins near synapses, fundamentally changing understanding of memory and brain plasticity.
An exploration of why mapping the brain's connectome is valuable, arguing that unlike AI systems where design is in code outside weights, brains must encode all design physically, making architecture the key to understanding.
This paper proposes a framework for neocortical learning that meets criteria for computational, algorithmic, and implementational plausibility, using error-driven predictive learning via temporal derivatives and corticothalamic circuits. It suggests potential improvements over backpropagation.
A new method reveals that a single cortical neuron can classify cats vs dogs, recognize spoken words, and solve 10-bit parity, tasks previously thought to require entire networks.
This newsletter covers two major tech stories: a biotech company's first human trial of 'reprogramming' cells to reverse aging, and the growing field of interoception research. It also notes SpaceX's record-breaking IPO.
An explainer on interoception, the hidden sense of internal body signals, and how decoding these signals could transform treatments for conditions like obesity and anxiety.
This position paper argues that integrating explicit memory, analogous to human hippocampal memory, is essential for advancing LLMs toward AGI. It draws on neuroscience to propose that higher-order cognitive functions require explicit memory beyond implicit statistical learning.
This paper presents an empirical study evaluating general-purpose coding agents on a fly optogenetics data-to-discovery pipeline, finding that while agents can automate individual stages, they struggle with end-to-end tasks requiring scientific judgment and resource management.
A new analysis published in Neuron argues that current scientific methods may conflate consciousness with general information processing, raising questions about claims of consciousness in AI, animals, fetuses, and organoids.
Jeff Bezos has funded Flourish, a neuro-AI startup valued at $2.5 billion with $500 million in funding, co-founded by former Amazon executive Rob Williams and neuroscientist Thomas Reardon. The company aims to build brain-inspired AI systems called Cortex AI that can run on 50 watts or less and continuously learn, addressing key limitations of current LLMs.
Eero P. Simoncelli, a founding member of NYU's Center for Data Science, was elected to the National Academy of Sciences for his distinguished research on how brains and machines represent visual information.
A new dataset and model predict emotions from EEG data with over twice the performance of previous state-of-the-art.
A 2025 review and pilot trial show that creatine supplementation raises brain energy levels and slows cognitive decline in early Alzheimer's patients by 30%, highlighting the supplement's overlooked neurological benefits.