Why does a stored memory sometimes feel out of reach? A study in mice from Nagoya City University, published in Neuron, shows that slow fluctuations in brain histamine neurons shape whether learned memories become accessible when a memory cue is presented. The work suggests that memory access depends not only on stored memory traces, but also on moment-to-moment brain states.

To simulate the properties of materials, researchers use crystal structures of materials stored in databases. Often, however, these lack the positions of the hydrogen atoms. Researchers at the Paul Scherrer Institute PSI have now trained an artificial intelligence system to reconstruct these positions very quickly and efficiently. Their new method, called XtalPaint, will make it possible to simulate materials more accurately for a variety of applications: from efficient hydrogen storage to new batteries.

Age-related hearing loss is very prevalent in older adults and is associated with accelerated cognitive decline and increased dementia risk. Funded by the Fu Tak Iam Foundation Limited, the School of Nursing (SN) of The Hong Kong Polytechnic University (PolyU) has developed the AI-enhanced Gamified Auditory-Cognitive Dual-Task Training System 2.0 (the System) that addresses hearing loss and dementia risk with an integrated innovative solution. In partnership with the Community CareAge Foundation (CCAF), the System has been incorporated into a three-year, territory-wide service under CCAF’s CareAge Hearing Programme and is expected to benefit around 2,400 underprivileged older adults in Hong Kong.

A new study published in Big Earth Data proposes GeoJSON agents, a novel multi-agent large language model (LLM) framework for geospatial analysis that transforms natural language instructions into structured GeoJSON operations through function calling and code generation. Experiments on a hierarchical benchmark of 70 spatial tasks show that the code generation–based agent achieved 97.14% accuracy and the function calling–based agent achieved 85.71%, both significantly outperforming general-purpose models, while highlighting the trade-off between flexibility and execution stability in GeoAI applications. The data that support the findings of this study are openly available in figshare at https://doi.org/10.6084/m9.figshare.29921492.

An international team ihas used artificial intelligence to analyse the climate commitments submitted to the United Nations by 158 countries. Their conclusion is stark: profound inequalities persist within global climate planning. The paper, published in the journal Nature Communications, concludes that high-income nations focus their climate commitments on health, technological transitions and emissions reduction. Conversely, low- and middle-income countries tie climate action to immediate survival challenges – such as access to water, energy, food security and natural resource management.

A new course offered this spring quarter at Cal Poly introduced upper-division undergraduates to the fundamentals of building synthetic cells while collaborating on innovative projects. The cutting-edge course titled “Engineering Synthetic Cells,” taught by Javin Oza, a Cal Poly biochemistry associate professor, involves a partnership with the San Francisco-based company b.next, an open-source synthetic cell platform Nucleus, and an international science network.

Researchers have developed a new magnetic memory material whose information can be rewritten using light instead of electric current. The advance could enable memory devices that are about 1,000 times faster and far more energy efficient, offering a potential path to reducing electricity use in AI systems, data centers and next-generation information infrastructure.