Evidence for neuromodulatory control of flexibility in human neural models.

Deep neural networks can be trained to automatically find mechanistic models which quantitatively agree with experimental data, providing new opportunities for building and visualizing interpretable models of neural dynamics.

Model identification of neural encoding is an accessible system for the analysis of neural data that allows identifying and characterizing arbitrary relationships between neural activity and task-related variables such as behavior, stimuli, or internal states.

A data-driven model of the neural activity of a circuit essential to swimming orientation in zebrafish can be mathematically interpreted to unveil the dynamical mechanism leading to persistence in tail orientation.

Psychophysical measurement and computational modeling show that sensory information cannot contribute directly to a cognitive judgment, but must first be integrated into resource-limited working memory.

Visualization methods display the dynamics of the currents in conductance-based model neurons, and show how their contribution changes in response to perturbation.

A novel analysis of neural activity recorded in monkeys performing a “brain-machine interface” task reveals that a mismatch between motor effectors and the brains’ internal models of those effectors can explain a substantial portion of movement errors.

A neural network model of the hippocampus exhibits a division of labor across its two main pathways during category learning, with one pathway specializing in extracting systematic category information and another in encoding arbitrary details.

A model of in vitro human corticogenesis identifies alterations in gene expression caused by loss of 16p11.2 CNV genes in hiPSC-derived progenitor cells.

The neural representation of position in the medial entorhinal cortex may be stabilized by synaptic connectivity across modules, which enforces coherent updates in their states.