Low-dimensional dynamics for working memory and time encoding

Proceedings of the National Academy of Sciences, Год журнала: 2020, Номер 117(37), С. 23021 - 23032

Опубликована: Авг. 28, 2020

Our decisions often depend on multiple sensory experiences separated by time delays. The brain can remember these and, simultaneously, estimate the timing between events. To understand mechanisms underlying working memory and encoding, we analyze neural activity recorded during delays in four experiments nonhuman primates. disambiguate potential mechanisms, propose two analyses, namely, decoding passage of from data computing cumulative dimensionality trajectory over time. Time be decoded with high precision tasks where information is relevant lower when irrelevant for performing task. Neural trajectories are always observed to low-dimensional. In addition, our results further constrain encoding as find that linear “ramping” component each neuron’s firing rate strongly contributes slow timescale variations make possible. These constraints rule out models rely constant, sustained networks high-dimensional trajectories, like reservoir networks. Instead, recurrent trained backpropagation capture time-encoding properties data.

Язык: Английский

---

A Dynamical Systems Perspective on Flexible Motor Timing

Trends in Cognitive Sciences, Год журнала: 2018, Номер 22(10), С. 938 - 952

Опубликована: Сен. 25, 2018

Язык: Английский

---

Cortical-like dynamics in recurrent circuits optimized for sampling-based probabilistic inference

Nature Neuroscience, Год журнала: 2020, Номер 23(9), С. 1138 - 1149

Опубликована: Авг. 10, 2020

Язык: Английский

---

Latent Factors and Dynamics in Motor Cortex and Their Application to Brain–Machine Interfaces

Journal of Neuroscience, Год журнала: 2018, Номер 38(44), С. 9390 - 9401

Опубликована: Окт. 31, 2018

In the 1960s, Evarts first recorded activity of single neurons in motor cortex behaving monkeys (Evarts, 1968). 50 years since, great effort has been devoted to understanding how neuron relates movement. Yet these exist within a vast network, nature which largely inaccessible. With advances recording technologies, algorithms, and computational power, ability study networks is increasing exponentially. Recent experimental results suggest that dynamical properties are critical movement planning execution. Here we discuss this systems perspective it reshaping our cortices. Following an overview key studies cortex, techniques uncover "latent factors" underlying observed neural population activity. Finally, efforts use factors improve performance brain-machine interfaces, promising make findings broadly relevant neuroengineering as well neuroscience.

Язык: Английский

---

More than Just a “Motor”: Recent Surprises from the Frontal Cortex

Journal of Neuroscience, Год журнала: 2018, Номер 38(44), С. 9402 - 9413

Опубликована: Окт. 31, 2018

Motor and premotor cortices are crucial for the control of movements. However, we still know little about how these areas contribute to higher-order motor control, such as deciding which movements make when them. Here focus on rodent studies review recent findings, suggest that-in addition control-neurons in play a role sensory integration, behavioral strategizing, working memory, decision-making. We that seemingly disparate functions may subserve an evolutionarily conserved sensorimotor cognition further study could major contribution our understanding evolution function mammalian frontal cortex.

Язык: Английский

---