Latent dynamics

…Models of these neural dynamics should ideally be both interpretable and fit the observed data well. Low-rank recurrent neural networks (RNNs) exhibit such interpretability by having tractable dynamics. However, it is unclear how to best fit low-rank RNNs to data consisting of noisy observations of an underlying stochastic system. Here, we propose to fit stochastic low-rank RNNs with variational sequential Monte Carlo methods. We validate our method on severa

…Models of these neural dynamics should ideally be both interpretable and fit the observed data well. Low-rank recurrent neural networks (RNNs) exhibit such interpretability by having tractable dynamics. However, it is unclear how to best fit low-rank RNNs to data consisting of noisy observations of an underlying stochastic system. Here, we propose to fit stochastic low-rank RNNs with variational sequential Monte Carlo methods. We validate our method on severa

…such activity is constrained to latent dynamics on the population level. Whether other regions exhibit these motor cortical dynamics, and how this distributed activity relate to specific motor behavior remains unclear. We used multiple Neuropixels probes to simultaneously record population activity across cortical and subcortical regions during a goal-directed reaching task in mice. Task-related population activity followed dominant latent dynamics across all rec

Animals of the same species exhibit similar behaviours that are advantageously adapted to their body and environment. These behaviours are shaped at the species level by selection pressures over evolutionary timescales. Yet, it remains unclear how these common behavioural adaptations emerge from the idiosyncratic neural circuitry of each individual. The overall organisation of neural circuits is preserved across individuals because of their common evolutionarily specified developmental programme. Such organisation at the circuit level may constrain neural activity, leading to low-dimensional l

…What are the latent dynamics and circuits that support phase coding with neural oscillations? Here, we propose a new computational hypothesis which is derived from analyzing trained recurrent neural networks (RNNs). We train RNNs on a working memory task, while al

Neural recordings are often high-dimensional while latent dynamics related to behavior are low-dimensional. Similarly, ethological behaviors are described by large numbers of dynamic features and it is important to know both the dimensionality and composition of the neurally relevant behavioral subspace. However, methods for data-driven identification of behavioral subspaces relevant for neural recordings is nascent. Here, we introduce external Dynamics Components Analysis (eDCA), a linear dimensionality reduction method that finds subspaces of past neural data that have maximal mutual informa

Experimental work across a variety of species has demonstrated that spontaneously generated behaviors are robustly correlated to variation in neural activity within the cerebral cortex. Indeed, functional magnetic resonance imaging (fMRI) data suggest that functional connectivity in cortical networks varies across distinct behavioral states, providing for the dynamic reorganization of patterned activity. However, these studies generally lack the temporal resolution to establish links between cortical signals and the continuously varying fluctuations in spontaneous behavior typically observed i

Precise movement execution often relies on adequately planning the upcoming action, a process that is largely mediated by the dorsal premotor (PMd) and primary motor (M1) cortices. Due to their anatomical and functional similarities, PMd and M1 are commonly assumed to contain similar planning- related information. Here we show that when planning movements under uncertainty, the activity of PMd and M1 neural populations show distinct features, suggestive of area-specific computations. We analyzed simultaneous recordings from populations of PMd and M1 neurons from monkeys performing both a stand

Recent investigations into the coordinated activity of neuronal populations in motor circuits have revealed low-dimensional dynamical features hypothesized to support the generation of patterned motor output. However, such studies have typically relied on vast simplifications - trial-averaging, heavy smoothing - and are often far removed from muscle activity, the true output of the motor system. Thus, it is unclear whether the uncovered dynamical features are primarily useful for intuition-building, or reflect computational mechanisms that operate at the temporal precision of motor control. He

…methods to identify low-dimensional latent dynamics without requiring trial-averaging (tensor component analysis; TCA). This approach assumes that the data tensor can be described as a sum of components with fixed neural weights and temporal dynamics that vary over trials only in amplitude. However, recent evidence suggests that

Understanding the neural code depends on capturing latent computational features within the highly complex dynamics of neural activity. While many features of neural dynamics will be relevant for a given computation, other features will not. In contrast with existing methods that describe how latent dynamics depend on experimental variables, we present a model designed to factor out variables that are irrelevant to the latent dynamics. We propose a fluctuating hidden Markov model (FHMM) that simultaneously learns an HMM that captures latent dynamics, and a baseline, dynamic firing rate that ca

…sense', take for example inferring latent dynamics underlying neural population activity; nonparametric approaches such as Gaussian Processes (GPs) are highly flexible, expressive, and allow us to encode broad assumptions such as periodicity, stationarity, and smoothness. Though, these properties make GPs attractive for neural data analysis, their comput

…posteriors over partially observable latent world states. For our task, these beliefs represent the subject's estimated current position and uncertainty, assumed observat

The ability to simultaneously record the activity from tens to thousands to tens of thousands of neurons has allowed us to analyze the computational role of population activity as opposed to single neuron activity. Recent work on a variety of cortical areas suggests that neural function may be built on the activation of population-wide activity patterns, the neural modes, rather than on the independent modulation of individual neural activity. These neural modes, the dominant covariation patterns within the neural population, define a low dimensional neural manifold that captures most of the

The ability to simultaneously record the activity from tens to thousands and maybe even tens of thousands of neurons has allowed us to analyze the computational role of population activity as opposed to single neuron activity. Recent work on a variety of cortical areas suggests that neural function may be built on the activation of population-wide activity patterns, the neural modes, rather than on the independent modulation of individual neural activity. These neural modes, the dominant covariation patterns within the neural population, define a low dimensional neural manifold that captures m

…of the neural modes results in latent dynamics. We hypothesize that the latent dynamics a