The authors demonstrate that optical fibers with tapered tips can homogenously illuminate either elongated brain structures or dynamically selected subregions. Tapered fibers achieve efficient optogen...

Modelling biological decision-making with tiny recurrent neural networks enables more accurate predictions of animal choices than classical cognitive models and offers insights into the underlying cog...

Comrie et al. show that the hippocampus flexibly represents different alternative possibilities—including paths ahead, behind, and far away—to meet distinct cognitive demands related to decision-makin...

Efficient processing of high-dimensional datasets demands minimal information loss and reduced computational complexity. Ding et al. report a memristor-based system that boosts data simplification eff...

Using experiments and theory, the authors examine how olfactory memories are encoded in the zebrafish olfactory cortex. They find that experience reshapes the internal map of odor space, making releva...

Chericoni et al. show that neurons in the human hippocampus track the positions of multiple agents in a simple pursuit-based video game. They identified place codes for each agent and for gaze. The co...

GluA2-containing AMPA receptors (AMPARs) are not Ca2+ impermeable, and their ability to transport Ca2+ is shaped by the subunit composition of AMPAR tetramers as well as the orientation of transm...

Nature Methods - Getting over ANOVA: estimation graphics for multi-group comparisons

Approaches for the development of future at-scale neuromorphic systems based on principles of biointelligence are described, along with potential applications of scalable neuromorphic architectures an...

This Perspective sets out an approach to electron transport in nanoscale devices based on statistical fields — specifically the spatial and temporal variations in temperature and chemical potential th...

We demonstrate the first-ever nontrivial, biologically realistic connectome simulated on neuromorphic computing hardware. Specifically, we implement the whole-brain connectome of the adult Drosophila ...

Explore Taiwan’s neuroscience history through this 50 Reflections campaign, showcasing impactful contributions to global brain research.

The authors found that dopamine signals move gradually from time of reward to time of cue through intermediate timepoints, similarly to the evaluation signals used in temporal difference learning. The...

Null space theory predicts that neurons generate spikes not only to produce behavior but also to prevent the undesirable effect of other neurons on behavior. In this work, we show that this competitiv...

Nature spoke to three students from China’s first cohort of PhD candidates who are allowed to graduate with products instead of papers.

Behavioural experiments to study decision-making in response to context-dependent accumulation of evidence provide testable models that are consistent with the heterogeneity in neural signatures among...

Astera recently hosted its first essay competition, focused around metascience.

Renderers, Simulators, Planners, and the Loop That Connects Them

We present Probabilistic Structure Integration (PSI), a system for learning richly controllable and flexibly promptable world models from data. PSI consists of a three-step cycle. The first step, Prob...

The authors developed Cheese3D, a hardware–software framework for precise and sensitive measurement of whole-face movements in mice that enables quantitative inference of neural and physiological proc...

Beta and gamma inhibitory rhythms are preferentially tuned to govern synaptic integration in layer 5 pyramidal neurons by differentially modulating responses to inputs targeting distal dendritic and p...

The brain is a dissipative, far-from-equilibrium system that continuously consumes energy to sustain structured neural activity while producing entropy. How system-level interactions among energy, entropy, structure, and physiological constraints give rise to cognition through distributed computation remains a central challenge in neuroscience. Existing free-energy frameworks have provided valuable conceptual insights, yet many remain largely metaphorical and lack explicit biological grounding, limiting the specificity and falsifiability of their predictions. Here, we propose a biologically grounded, thermodynamically inspired framework in which brain dynamics emerge from a continual balance between stability and flexibility, formalized as minimizing a Gibbs-free-energy-like function (ΔG = ΔH − TΔS). Stable neural configurations correspond to low-enthalpy, ordered states that support reliable function, whereas flexible configurations correspond to higher-entropy states that enable adaptive behavior. We further propose that neural population dynamics can be described as evolving on a hyperbolic statistical landscape shaped by metabolic, biophysical, anatomical, and bodily constraints. Neural activity samples this landscape, and experience gradually reshapes it over time. Metastable brain states occupy local free-energy minima, with transitions governed by Boltzmann-like probabilities. Learning sculpts the landscape via synaptic optimization and pruning, reducing entropy and stabilizing task-relevant dynamics. Importantly, this framework accounts for adaptive reallocation of neural resources under physiological stress, whereby higher-order cognitive functions may be transiently constrained to prioritize survival-essential processes, thereby predicting that cognitive function can remain rescuable in some cases despite structural damage. Together, this quantitative thermodynamic and geometric framework provides a unified foundation for understanding cognition as an emergent, system-level property of whole-brain dynamics.

Backpropagation through time (BPTT) is the de facto standard for training recurrent neural networks (RNNs), but it is non-causal and non-local. Real-time recurrent learning is a causal alternative, bu...

Recent data has observed sequential, rather than persistent, neural activity during evidence accumulation for decision-making. Here, the authors develop models that accumulate evidence through sequenc...

Entorhinal-hippocampal afferents innervate intrahippocampal circuitry at both behavioral and synaptic levels, including the spatial learning and heterosynaptic plasticity.

“Most observable patterns of nature arise from aggregation of numerous small-scale processes… Aggregation tends to smooth fluctuations.” ~S. A. Frank[1] “Nature is both periodic and perpetual. One of ...

Language models show a surprising range of capabilities, but the source of their apparent competence is unclear. Do these networks just memorize a collection of surface statistics, or do they rely...

Toward identifying the protein landscape of memory-encoding hippocampal synapses.

The latent circuit model identifies low-dimensional mechanisms of task execution from heterogenous neural responses. This approach reveals a latent inhibitory mechanism for context-dependent decisions...

Various theories have been developed for the biological and physical basis of consciousness. In this Review, Anil Seth and Tim Bayne discuss four prominent theoretical approaches to consciousness, nam...

Read chapter 2 The Bidirectionality of Neuroscience and Artificial Intelligence: Since artificial intelligence's (AI) inception, AI and neuroscience have...

Artificial neural networks are becoming more advanced and human-like in detail and behavior. The notion that machines mimicking human brain computations migh...

Flexible learning relies on integrating sensory and contextual information to adjust behavioral output in different environments. The anterolateral...