For the first time, it has been confirmed that individual neurons represent the concepts we learn, regardless of the context in which they are encountered, challenging previous beliefs.
A study led by Dr. Rodrigo Quian Quiroga, head of the Neural Mechanisms of Perception and Memory Research Group at the Hospital del Mar Research Institute, has provided the first direct evidence of how neurons in the human brain store memories independently of the context in which they are acquired.
Artificial Intelligence (AI), particularly large language models like GPT-4, has shown impressive performance on reasoning tasks. But does AI truly understand abstract concepts, or is it just mimicking patterns? A new study from the University of Amsterdam and the Santa Fe Institute reveals that while GPT models perform well on some analogy tasks, they fall short when the problems are altered, highlighting key weaknesses in AI’s reasoning capabilities. The work is published in Transactions on Machine Learning Research.
Analogical reasoning is the ability to draw a comparison between two different things based on their similarities in certain aspects. It is one of the most common methods by which human beings try to understand the world and make decisions. An example of analogical reasoning: cup is to coffee as soup is to??? (the answer being: bowl)
Large language models (LLMs) like GPT-4 perform well on various tests, including those requiring analogical reasoning. But can AI models truly engage in general, robust reasoning or do they over-rely on patterns from their training data? This study by language and AI experts Martha Lewis (Institute for Logic, Language and Computation at the University of Amsterdam) and Melanie Mitchell (Santa Fe Institute) examined whether GPT models are as flexible and robust as humans in making analogies.
A research team has identified different subtypes of white matter (WM) astrocytes, including a unique type with the ability to multiply and potentially aid in brain repair. Using single-cell RNA sequencing and spatial transcriptomics, the scientists mapped astrocyte diversity across different brain regions and species, providing the first detailed molecular profile of WM astrocytes.
The team was led by Dr. Judith Fischer-Sternjak from Helmholtz Munich and Ludwig-Maximilians-Universität (LMU) München, alongside Prof. Magdalena Götz from Helmholtz Munich, LMU and the Munich Cluster for Systems Neurology (SyNergy). The research is published in the journal Nature Neuroscience.
Unveiling white matter astrocyte diversity Astrocytes, known for their crucial role in supporting neurons and maintaining brain health, have been predominantly studied in gray matter (GM), which is involved in information processing. However, white matter astrocytes, which support long-range neural connections, remain poorly understood. This study fills a major knowledge gap by showing that WM astrocytes are not a uniform population but consist of distinct subtypes with specialized roles.
A little over a year after releasing two open Gemma AI models built from the same technology behind its Gemini AI, Google is updating the family with Gemma 3.
According to the blog post, these models are intended for use by developers creating AI applications capable of running wherever they’re needed, on anything from a phone to a workstation with support for over 35 languages, as well as the ability to analyze text, images, and short videos.
The company claims that it’s the world’s best single-accelerator model, outperforming competition from Facebook’s Llama, DeepSeek, and OpenAI for performance on a host with a single GPU, as well as optimized capabilities for running on Nvidia’s GPUs and dedicated AI hardware.
Gemma 3’s vision encoder is also upgraded, with support for high-res and non-square images, while the new ShieldGemma 2 image safety classifier is available for use to filter both image input and output for content classified as sexually explicit, dangerous, or violent.
To go deeper into those claims, you can check out the 26-page technical report.
Last year it was unclear how much interest there would be in a model like Gemma, however, the popularity of DeepSeek and others shows there is interest in AI tech with lower hardware requirements.
Sustaining growth in storage and computational needs is increasingly challenging. For over a decade, exponentially more information has been produced year after year while data storage solutions are pressed to keep up. Soon, current solutions will be unable to match new information in need of storage. Computing is on a similar trajectory, with new needs emerging in search and other domains that require more efficient systems. Innovative methods are necessary to ensure the ability to address future demands, and DNA provides an opportunity at the molecular level for ultra-dense, durable, and sustainable solutions in these areas.
In this webinar, join Microsoft researcher Karin Strauss in exploring the role of biotechnology and synthetic DNA in reaching this goal. Although we have yet to achieve scalable, general-purpose molecular computation, there are areas of IT in which a molecular approach shows growing promise. These areas include storage as well as computation.
Learn how molecules, specifically synthetic DNA, can store digital data and perform certain types of special-purpose computation, like large-scale similarity search, by leveraging tools already developed by the biotechnology industry. Starting with some background on DNA and its storage potential, you’ll explore the advantages of using DNA for this application. Then, you’ll get a closer look at an end-to-end system, including encoding, synthesizing, reading, and decoding DNA. We’ll also look at an affordable full-stack digital microfluidics platform for wet lab preparations and conclude with a discussion of future hybrid systems.
Together, you’ll explore:
■ The intersection between technology and science of DNA data storage and computation. ■ The many advantages for using DNA to store data compared with other methods. ■ A detailed walkthrough of an end-to-end DNA storage system and its stages. ■ How DNA can be used for image similarity search.
Credits: Dan V. Nicolau, Mercy Lard, Till Korten, Falco C. M. J. M. van Delft, Malin Persson, Elina Bengtsson, Alf Månsson, Stefan Diez, Heiner Linke, and Dan V. Nicolau.
PNAS. 2016. DOI: 10.1073/pnas.
Animation explaining the computation principle. In this animation, a network encoding the {1,3} SSP is explained and an example agent is shown to travel the path that encodes the subset {3}. The agent enters the network from the top left-hand corner. It first encounters a split junction where it randomly decides to turn right and exclude the corresponding number. The agent then enters another split junction, where it adds the corresponding number to the subset. The number of pass junctions following each split junction determines the actual value of the integer added to the subset at the respective split junction. The exit numbers correspond to the target sums T (potential solutions) represented by each exit. The example agent arrives at the exit #3 corresponding to the total sum of the subset {3} the agent explored. Finally, the correct results (labeled in green) and incorrect results (where no agents will arrive; labeled in magenta) for this particular set {1, 3} are explained.
Acknowledgment: Dissemination of the results of ABACUS Project, funded by the European Union’s Seventh Framework Programme for Research, Technological Development and Demonstration under Grant Agreement no. 613,044 The beneficiaries and partners in ABACUS Consortium are: Lund University, Coordinator, Molecular Sense Ltd. — UK, Linnaeus University – Sweden, Dresden University – Germany and McGill University, Canada. The main result of the Project has been published in the Proceedings of the National Academy of Sciences of the USA — 113, 2591–2596 (2016). Authors: Dan V. Nicolau Jr., Mercy Lard, Till Korten, Falco C. M. J. M. van Delft, Malin Persson, Elina Bengtsson, Alf Månsson, Stefan Diez, Heiner Linke, and Dan V. Nicolau For more information on the ABACUS Project you may visit http://abacus4eu.com/
A study conducted by researchers from the University of São Paulo sheds light on new discoveries about the mechanisms of oxidative phosphorylation in ATP production. Recent findings highlight the involvement of sodium in mitochondrial respiration.
In an article published in Trends in Biochemical Sciences, Alicia Kowaltowski, a full professor at the University of São Paulo’s Institute of Chemistry (IQ-USP) in Brazil, calls for a “rewriting” of textbooks regarding the location of the electron transport chain in mitochondria and the role of sodium in mitochondrial respiration.
Kowaltowski is also a member of the Research Center for Redox Processes in Biomedicine (Redoxoma), a Research, Innovation, and Dissemination Center (RIDC) funded by FAPESP and based at IQ-USP.
Four minutes. Imagine what you can accomplish in four minutes. Make coffee? Read half an article? Send a few text messages?
For most of us, four minutes pass in a heartbeat. Yet during those same four minutes, a quantum computer recently performed calculations that would have kept a conventional supercomputer busy for 2.6 billion years.
Scientists achieved something magical—compressing billions of years of computation into minutes. Such power shifts our understanding of what’s possible. Quantum computing won’t just change how we process information; it will transform medicine, climate science, materials design, and countless other fields we rely on daily.
Join our free newsletter for weekly updates on the latest innovations improving our lives and shaping our future, and don’t miss this cool list of easy ways to help yourself while helping the planet.
