A5063041513- Neural dynamics and brain function
- EEG and Brain-Computer Interfaces
- Memory and Neural Mechanisms
- Visual perception and processing mechanisms
- Neuroscience and Neural Engineering
- Glaucoma and retinal disorders
- Ophthalmology and Visual Impairment Studies
- Speech and Audio Processing
- Neurobiology and Insect Physiology Research
- Motor Control and Adaptation
- Neuroscience and Neuropharmacology Research
- Neural and Behavioral Psychology Studies
- Neural Networks and Reservoir Computing
- Advanced Memory and Neural Computing
- Human Motion and Animation
- Functional Brain Connectivity Studies
- Handwritten Text Recognition Techniques
- Hand Gesture Recognition Systems
- Muscle activation and electromyography studies
- Music Technology and Sound Studies
- Reinforcement Learning in Robotics
- Neural Networks and Applications
- Blind Source Separation Techniques
- Music and Audio Processing
- Tactile and Sensory Interactions
Harvard University
2023-2025
Center for the Neural Basis of Cognition
2018-2024
Carnegie Mellon University
2018-2024
Harvard University Press
2023-2024
University of Pittsburgh
2019-2020
The University of Texas at Austin
2010-2015
Society for Neuroscience
2015
Significance Consider a skill you would like to learn, playing the piano. How do progress from “Chopsticks” Chopin? As learn something new with your hands, does brain also new? We found that monkeys learned skilled behavior by generating neural activity patterns. used brain–computer interface (BCI), which directly links movement of computer cursor, encourage animals generate Over several days, began exhibit patterns enabled them control BCI cursor. This suggests learning play piano and other...
Previous work has revealed a remarkably direct neural correlate of decisions in the lateral intraparietal area (LIP). Specifically, firing rate been observed to ramp up or down manner resembling accumulation evidence for perceptual decision reported by making saccade into (or away from) neuron's response field (RF). However, this link between LIP and formation emerged from studies where saccadic target was always stimulating RF during decisions, averaged activity restricted sample neurons....
Associative learning depends on contingency, the degree to which a stimulus predicts an outcome. Despite its importance, neural mechanisms linking contingency behavior remain elusive. Here we examined dopamine activity in ventral striatum - signal implicated associative Pavlovian degradation task mice. We show that both anticipatory licking and responses conditioned decreased when additional rewards were delivered uncued, but remained unchanged if cued. These results conflict with...
Millions of neurons drive the activity hundreds muscles, meaning many different neural population patterns could generate same movement. Studies have suggested that these redundant (i.e. behaviorally equivalent) may be beneficial for computation. However, it is unknown what constraints limit selection patterns. We leveraged a brain-computer interface, allowing us to define precisely which were redundant. Rhesus monkeys made cursor movements by modulating in primary motor cortex. attempted...
To behave adaptively, animals must learn to predict future reward, or value. do this, are thought reward predictions using reinforcement learning. However, in contrast classical models, estimate value only incomplete state information. Previous work suggests that partially observable tasks by first forming "beliefs"-optimal Bayesian estimates of the hidden states task. Although this is one way solve problem partial observability, it not way, nor most computationally scalable solution...
Abstract How are we able to learn new behaviors without disrupting previously learned ones? To understand how the brain achieves this, used a brain-computer interface (BCI) learning paradigm, which enables us detect presence of memory one behavior while performing another. We found that use BCI map altered neural activity monkeys produced when they returned using familiar map, in way was specific experience. That is, left “memory trace.” This trace co-existed with proficient performance...
Temporal integration of visual motion has been studied extensively within the frontoparallel plane (i.e., 2D). However, majority occurs a 3D environment, and it is unknown whether principles from 2D processing generalize to more realistic motion. We therefore characterized compared temporal underlying (left/right) (toward/away) direction discrimination in human observers, varying coherence across range viewing durations. The resulting discrimination-versus-duration functions followed three...
A remarkable demonstration of the flexibility mammalian motor systems is primates’ ability to learn control brain-computer interfaces (BCIs). This constitutes a completely novel behavior, yet primates are capable learning BCIs under wide range conditions. with carefully calibrated decoders, for example, can be learned only minutes hours practice. With few weeks practice, even randomly constructed decoders learned. What biological substrates this process? Here, we develop theory based on...
The variational autoencoder (VAE) is a popular probabilistic generative model. However, one shortcoming of VAEs that the latent variables cannot be discrete, which makes it difficult to generate data from different modes distribution. Here, we propose an extension VAE framework incorporates classifier infer discrete class modeled data. To model sequential data, can combine our Classifying with recurrent neural network such as LSTM. We apply this algorithmic music generation, where learns...
Abstract To behave adaptively, animals must learn to predict future reward, or value. do this, are thought reward predictions using reinforcement learning. However, in contrast classical models, estimate value only incomplete state information. Previous work suggests that partially observable tasks by first forming “beliefs”—optimal Bayesian estimates of the hidden states task. Although this is one way solve problem partial observability, it not way, nor most computationally scalable...
Abstract Internal states such as arousal, attention, and motivation are known to modulate brain-wide neural activity, but how these processes interact with learning is not well understood. During learning, the brain must modify activity it produces improve behavioral performance. How do internal affect evolution of this process? Using a brain-computer interface (BCI) paradigm in non-human primates, we identified large fluctuations population motor cortex (M1) indicative arousal-like state...
The classic aperture problem describes the ambiguity inherent to motion of a frontoparallel (2D) contour (such as line or an edge) viewed through circular aperture. Despite continuum 2D velocities consistent with apertured view, observers consistently perceive direction orthogonal contour. Here we present analogous 3D version where judged slanted planar surface defined by moving random dot stereogram presented behind If is specified single frame lifetimes, only potential factors influencing...