Byron M. Yu

This study investigates how rare rewards affect performance and neural activity in rhesus monkeys during a color change detection task, finding detrimental changes in V4 population activity linked to perceptual choking.

This study uses a brain-computer interface to show animals can regulate prefrontal cortex activity through real-time neurofeedback, reducing neural fluctuations and linking this to arousal regulation.

The paper uses recurrent network models to show how input alignment with connectivity shapes directional interactions in cortical circuits, explaining shifts in signal flow directionality.

The paper uses dual hemisphere recordings and a novel statistical method to show shared neuronal variability across prefrontal cortex populations linked to arousal.

The paper identifies distinct neural modes in the sensorimotor cortex that encode grasp force and timing during isometric grasps in tetraplegic individuals.

The paper shows that posture and goal information in motor cortex are organized in separate, orthogonal subspaces, with posture information being stable across tasks.

This study investigates how working memory in the dlPFC remains robust to microstimulation perturbations, finding that task performance is largely unaffected despite changes in neural activity.

This study challenges monkeys to reverse natural neural activity time courses in motor cortex using a BCI, finding they cannot violate these time courses, supporting the idea that neural activity reflects network-level computations.

This paper introduces SNOPS, a method to automatically customize spiking network models to match large-scale neuronal recordings, revealing limitations of current models.

The paper investigates how high rewards affect neural activity in Rhesus monkeys during a reaching task, finding that large rewards reduce neural information about movement plans.

This paper develops a theory explaining BCI learning through low-dimensional control strategies, validated with experimental data.

The paper shows that motor cortex maintains similar neural dynamics during motor execution and imagery by recreating missing components in a unique imagery subspace.

This study evaluates dimensionality reduction methods for calcium imaging data and introduces CILDS, which outperforms existing methods in recovering low-dimensional neural activity.

This paper introduces DLAG, a framework to disentangle bidirectional signal flow between neuronal populations, demonstrating its effectiveness on synthetic data and primate visual areas.

The paper shows that learning a new BCI map alters neural activity in a way that preserves performance on a familiar map, suggesting a mechanism for learning multiple motor behaviors without interference.

The paper examines feedforward and feedback interactions between visual cortical areas and finds they use distinct population activity patterns.

The paper establishes mathematical and empirical links between pairwise neuronal correlations and dimensionality reduction metrics, showing that attention-related changes in mean pairwise correlation arise from distinct population-wide covariability shifts.

The paper investigates how internal states like arousal affect neural engagement during learning in non-human primates using a BCI paradigm, finding that abrupt changes in engagement influence learning speed for different task goals.

The paper identifies slow neural activity drift in macaque visual and prefrontal cortex as an impulsivity signal influencing perceptual decisions over time.

Task complexity sets how many dimensions multineuronal activity occupies and when dynamical portraits recovered from a tiny fraction of neurons can be trusted; tested on monkey reaching recordings.