Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Deep neural networks may offer theories of perception, cognition and action for biological brains. Here, Saxe, Nelli and Summerfield offer a road map of how neuroscientists can use deep networks to model and understand biological brains.
Mounting evidence suggests that the gut microbiome impacts brain function, and mechanistic connections between specific microbial by-products and the brain have begun to emerge. In this Perspective, Mazmanian and colleagues discuss the assortment of microbial molecules currently thought to mediate these gut–brain connections.
Major compelling questions about the functional role of the locus coeruleus nucleus that had been difficult to answer, given its remote location and diminutive size, have now become accessible via new neuroscience tools. In this Perspective, 14 investigators provide a historical context for recent discoveries and outline new vistas for investigation.
Reinforcement learning has been suggested to come in two flavours: model-free and model-based. In this Perspective, Collins and Cockburn explain why viewing reinforcement learning through this dichotomous lens is not always accurate or helpful, and suggest paths forward.
Although inputs and outputs that carry social signals are anatomically restricted to distinct subnuclear regions of the amygdala, social behaviours are not. This fact may be explained by the operation of multidimensional processing in parallel with subcircuits of genetically identical neurons that serve specialized and functionally dissociable functions.
In this Perspective, Hanno Würbel and colleagues argue that a disregard for incorporating biological variation in study design is an important cause of poor reproducibility in animal research. They put the case for the use of systematic heterogenization of study samples and conditions in studies to improve reproducibility.
The backpropagation of error (backprop) algorithm is frequently used to train deep neural networks in machine learning, but it has not been viewed as being implemented by the brain. In this Perspective, however, Lillicrap and colleagues argue that the key principles underlying backprop may indeed have a role in brain function.
Prior experience is incorporated into the brain’s predictive models of the world, enabling the accurate interpretation of and responses to new sensory information. In this Perspective, Teufel and Fletcher make the case for an important distinction between two forms of prediction that may advance our understanding of brain function.
Certain biological properties vary across different areas of the cerebral cortex. In this Perspective, Xiao-Jing Wang proposes that macroscopic gradients in some properties align with functional hierarchy and can lead to qualitative differences in function.