The human brain is capable of implementing inverse reinforcement learning, where an observer infers the hidden reward structure of a decision problem solely through observing another individual take actions.

Emergent property inference, a novel machine learning methodology, learns distributions of neural circuit model parameters that produce computational properties and provides novel scientific insight through the quantification of the rich parametric structure it captures.

BrainPy presents a general-purpose programming framework that enables efficient implementation, simulation, training, and analysis of brain dynamics models across multiple organization scales, ultimately facilitating the understanding of the complex neural mechanisms underlying brain functions.

The distribution of redundant neural activity is coupled with task-relevant activity, which may limit the extent to which redundancy can be exploited by the brain for computation.

The brain’s projection of linguistic knowledge onto speech is reflected by the automatic, reflexive neural tracking of phrases, while the distribution of phrasal-rate tracking and connectivity, particularly the involvement of inferior frontal cortex, is subject to task demands.

A recurrent network using a simple, biologically plausible learning rule can learn the successor representation, suggesting that long-horizon predictions are computations that are easily accessible in neural circuits.

Transcriptomes of cells in the Drosophila visual system combined with anatomy and other information reveal new functional insight into the visual circuit.

A circuit consisting of motor neurons, inhibitory interneurons and mechanosensory neurons controls copulatory behaviour in male fruit flies.

Neural processes for perception and confidence can be separated thanks to computational electroencephalography, thus revealing a neural circuit specific to metacognition.

Brian 2 is a software package for neural simulations that makes it both easy and computationally efficient to define original models for computational experiment.