Search

Attention plays a key role in cognition and perception. Here, the authors offer a neuro-inspired mechanism aiming to unify behavioral and neurophysiological findings into a single principled architecture.

Daniel E. Acuna, Stefano Allesina and Konrad P. Kording present a formula to estimate the future h-index of life scientists.

Scientists, research institutions, funders, libraries and publishers must all improve software practices.

How does the neurotransmitter dopamine and Parkinson’s disease (PD) affect decision-making under uncertainty? Vilares and Kording find that dopamine levels, which are affected by PD and the drugs used for its treatment, influence reliance on new versus prior information in decision-making.

The nervous system produces accurate movements by adapting to environmental changes. The authors construct a probabilistic model that compensates for motor errors and estimates their sources, finding that if the motor system used such a strategy, it would explain many previously observed movement-generalization phenomena.

A study shows that reward and punishment have distinct influences on motor adaptation. Punishing mistakes accelerates adaptation, whereas rewarding good behavior improves retention.

In animals, sensory systems appear optimized for the statistics of the external world. Here the authors take an artificial psychophysics approach, analysing sensory responses in artificial neural networks, and show why these demonstrate the same phenomenon as natural sensory systems.

A deep-learning-based software package called DeepLabCut rapidly and easily enables video-based motion tracking in any animal species. Such tracking technology is bound to revolutionize movement science and behavioral tracking in the laboratory and is also poised to find many applications in the real world.

Single-neuron and population activity in the macaque prefrontal and temporal cortex robustly encodes 24 species-typical behaviours, reciprocity in social interactions and social support.

Schulz et al. systematically benchmark performance scaling with increasingly sophisticated prediction algorithms and with increasing sample size in reference machine-learning and biomedical datasets. Complicated nonlinear intervariable relationships remain largely inaccessible for predicting key phenotypes from typical brain scans.

Complex motions can be achieved by chunking together simple movements at the cost of producing smooth, efficient trajectories. Here the authors apply a new algorithm to monkeys learning complex motor sequences and show that optimization initially occurs within small chunks that are later combined.

Progress in neural recording techniques has allowed the number of simultaneously recorded neurons to double approximately every 7 years, mimicking Moore's law. Emerging data analysis techniques should consider both the computational costs and the potential for more accurate models associated with this exponential growth of the number of recorded neurons.

It is debated whether motor cortical activity reflects plans for multiple potential actions. Here, the authors report that in a delayed response task with two potential reach targets, population activity in the dorsal premotor cortex at any moment in time represents only one of the targets.

A computational approach that uses the statistics of movement to find a mapping between neural activity and motor variables decodes the intended movements of monkeys with performance comparable to that of supervised methods.

Movements are continually constrained by the current body position and its relation to the surroundings. Here the authors report that the population activity of monkey dorsal premotor cortex neurons dynamically represents the probability distribution of possible reach directions.

A deep network is best understood in terms of components used to design it—objective functions, architecture and learning rules—rather than unit-by-unit computation. Richards et al. argue that this inspires fruitful approaches to systems neuroscience.

In this Review, Drew Bailey et al. present an accessible, non-technical overview of key challenges for causal inference in studies of human behaviour as well as methodological solutions to these challenges.

One of the ambitions of computational neuroscience is that we will continue to make improvements in the field of artificial intelligence that will be informed by advances in our understanding of how the brains of various species evolved to process information. To that end, here the authors propose an expanded version of the Turing test that involves embodied sensorimotor interactions with the world as a new framework for accelerating progress in artificial intelligence.

In this Review, Siddiqi et al. examine causal approaches to mapping human brain function. They provide a definition of causality for translational research, propose a framework for assessing causality strength in brain mapping studies and cover advances in techniques and their use in developing treatments for brain disorders.

How to establish causal links is a central question across scientific disciplines. Marinescu and colleagues describe methods from empirical economics and how they could be adapted across fields, for example, to psychology and neuroscience, to test causality.

While estimating causality from observational data is challenging, quasi-experiments provide causal inference methods with plausible assumptions that can be practical to a range of real-world problems.

Artificial neural networks are being widely used to model behavioural and neural data. In this Perspective article, Doerig et al. present neuroconnectionism as a Lakatosian research programme using artificial neural networks as a computational language for expressing falsifiable theories and hypotheses about the brain computations underlying cognition.

Comparisons of neural recordings across time, across subsets of neurons and across individuals requires the alignment of low-dimensional latent representations.

Theories of consciousness have a long and controversial history. One well-known proposal — integrated information theory — has recently been labeled as ‘pseudoscience’, which has caused a heated open debate. Here we discuss the case and argue that the theory is indeed unscientific because its core claims are untestable even in principle.