Xiao‐Jing Wang

This paper trains a recurrent neural network to generate grammatical action sequences and identifies experimentally testable predictions about its internal representation and generalization.

The paper models how mGluRs modulate trans-thalamic communication by altering neuronal dynamics in thalamic and cortical neurons, enhancing sensory information processing and decision-making.

The paper investigates whether the size of visual field maps in the brain correlates with working memory precision and finds a positive association.

This study identifies a dynamical regime in cortical networks that allows reliable signal propagation while preserving local temporal dynamics.

The paper identifies a neural representation of action symbols in the ventral premotor cortex of macaque monkeys performing a drawing-like task.

The paper introduces a three-factor learning rule to model how global signals guide local synaptic updates for mismatch computation in sensory areas.

This paper develops a two-stage neural circuit framework for economic choice using biologically plausible RNNs, showing how value computation and comparison enable generalization and multitasking.

The paper investigates how primate frontal cortex represents action grammar structure during novel problem-solving tasks and identifies preSMA as a key region for encoding grammatical properties.

This paper compares working memory models in macaques and marmosets using connectomic data, finding differences in distractibility and showing that model outcomes can be swapped by adjusting circuit parameters.

The paper identifies a timescale hierarchy in marmoset neocortex using ECoG data and develops a multi-regional model to explain signal integration and propagation.

The paper introduces a predictive coding model that explains neural responses to deviant stimuli across different paradigms and supports findings in mice.

The paper identifies four distinct VIP interneuron populations in the neocortex with unique connectivity and responses to neuromodulation, showing specialized roles in cortical circuit regulation.

The paper investigates bidirectional interactions between hippocampal sharp wave ripples and neocortical slow oscillations during sleep using electrophysiology and imaging in mice.

The paper investigates whether transcriptomic profiles can predict neuronal firing-rate adaptation across brain regions and species using Patch-seq data.

The paper develops a model of neocortex-wide neural activity to explain cognitive network interactions and dynamic routing during cognition.

This paper identifies overconfidence in Bayesian decoders of neural spiking activity and proposes methods to correct miscalibration in various experimental settings.

The paper develops a large-scale mouse brain model to predict working memory activity, showing distributed and modular coding based on cell type-specific connectome features.

This paper develops a neural network model of task switching, showing how rule representation and error monitoring emerge through training and how inhibitory neurons control subspaces of activity.

The paper identifies 35 prefrontal areas in macaques, reveals regional receptor differences, and shows functional connectivity patterns across clusters of frontal lobe areas.

The paper shows that bifurcations in cortical space create functional modules for specific cognitive tasks, with timescale patterns matching experimental observations.

This study investigates learning-to-learn in a neural network model, showing schema formation in a low-dimensional subspace that accelerates learning across similar problems.

The paper systematically examines the electrophysiological features of four L1 interneuron subtypes in mice and builds computational models to explain their firing patterns.

This paper trains biologically plausible RNNs on cognitive tasks with long-term dependencies using specialized skip-connections, achieving faster training and better performance than conventional methods.

This study compares two models of action value encoding in foraging behavior and finds the synaptic model more robust based on experimental predictions.

This paper develops a mesoscale connectome-based model of the macaque cortex to simulate conscious access, showing how receptor dynamics drive ignition-like events in neural networks.

The paper analyzes a primate cortex model to show that timescale hierarchy arises from three circuit properties: synaptic excitation gradient, neuronal population differences, and input balance.

This study develops a computational model of macaque cortex to show that persistent neural activity for working memory can arise from inter-areal reverberation, even when individual areas cannot generate it alone.

This study investigates how disruptions in the excitation-inhibition balance affect decision-making in a cortical circuit model, finding that both elevated and reduced E/I ratios impair decision-making in distinct ways.

This study trains mice to perform a delayed gratification task and shows that VTA dopamine neuron activity increases during waiting, with reinforcement learning modeling supporting the role of sustained dopaminergic signals in delaying gratification.

This study shows that a dopamine gradient in monkey cortex influences working memory by filtering stimuli and determining memory encoding mechanisms.