We present a dynamic nonlinear generative model for visual motion based on a latent representation of binary-gated Gaussian variables. Trained on sequences of images, the model learns to represent different movement directions in different variables. We use an online approximate-inference scheme that can be mapped to the dynamics of networks of neurons. Probed with drifting grating stimuli and moving bars of light, neurons in the model show patterns of responses analogous to those of direction-selective simple cells in primary visual cortex. Most model neurons also show speed tuning and respond equally well to a range of motion directions and speeds aligned to the constraint line of their respective preferred speed. We show how these computations are enabled by a specific pattern of recurrent connections learned by the model.

Measuring the dynamics of neural processing across time scales requires following the spiking of thousands of individual neurons over milliseconds and months. To address this need, we introduce the Neuropixels 2.0 probe together with newly designed analysis algorithms. The probe has more than 5000 sites and is miniaturized to facilitate chronic implants in small mammals and recording during unrestrained behavior. High-quality recordings over long time scales were reliably obtained in mice and rats in six laboratories. Improved site density and arrangement combined with newly created data processing methods enable automatic post hoc correction for brain movements, allowing recording from the same neurons for more than 2 months. These probes and algorithms enable stable recordings from thousands of sites during free behavior, even in small animals such as mice.

Neurophysiology has long progressed through exploratory experiments and chance discoveries. Anecdotes abound of researchers listening to spikes in real time and noticing patterns of activity related to ongoing stimuli or behaviors. With the advent of large-scale recordings, such close observation of data has become difficult. To find patterns in large-scale neural data, we developed 'Rastermap', a visualization method that displays neurons as a raster plot after sorting them along a one-dimensional axis based on their activity patterns. We benchmarked Rastermap on realistic simulations and then used it to explore recordings of tens of thousands of neurons from mouse cortex during spontaneous, stimulus-evoked and task-evoked epochs. We also applied Rastermap to whole-brain zebrafish recordings; to wide-field imaging data; to electrophysiological recordings in rat hippocampus, monkey frontal cortex and various cortical and subcortical regions in mice; and to artificial neural networks. Finally, we illustrate high-dimensional scenarios where Rastermap and similar algorithms cannot be used effectively.

Population neural recordings with long-range temporal structure are often best understood in terms of a shared underlying low-dimensional dynamical process. Advances in recording technology provide access to an ever larger fraction of the population, but the standard computational approaches available to identify the collective dynamics scale poorly with the size of the dataset. Here we describe a new, scalable approach to discovering the low-dimensional dynamics that underlie simultaneously recorded spike trains from a neural population. Our method is based on recurrent linear models (RLMs), and relates closely to timeseries models based on recurrent neural networks. We formulate RLMs for neural data by generalising the Kalman-filter-based likelihood calculation for latent linear dynamical systems (LDS) models to incorporate a generalised-linear observation process. We show that RLMs describe motor-cortical population data better than either directly-coupled generalised-linear models or latent linear dynamical system models with generalised-linear observations. We also introduce the cascaded linear model (CLM) to capture low-dimensional instantaneous correlations in neural populations. The CLM describes the cortical recordings better than either Ising or Gaussian models and, like the RLM, can be fit exactly and quickly. The CLM can also be seen as a generalization of a low-rank Gaussian model, in this case factor analysis. The computational tractability of the RLM and CLM allow both to scale to very high-dimensional neural data.

Neural language models (LMs) based on recurrent neural networks (RNN) are some of the most successful word and character-level LMs. Why do they work so well, in particular better than linear neural LMs? Possible explanations are that RNNs have an implicitly better regularization or that RNNs have a higher capacity for storing patterns due to their nonlinearities or both. Here we argue for the first explanation in the limit of little training data and the second explanation for large amounts of text data. We show state-of-the-art performance on the popular and small Penn dataset when RNN LMs are regularized with random dropout. Nonetheless, we show even better performance from a simplified, much less expressive linear RNN model without off-diagonal entries in the recurrent matrix. We call this model an impulse-response LM (IRLM). Using random dropout, column normalization and annealed learning rates, IRLMs develop neurons that keep a memory of up to 50 words in the past and achieve a perplexity of 102.5 on the Penn dataset. On two large datasets however, the same regularization methods are unsuccessful for both models and the RNN's expressivity allows it to overtake the IRLM by 10 and 20 percent perplexity, respectively. Despite the perplexity gap, IRLMs still outperform RNNs on the Microsoft Research Sentence Completion (MRSC) task. We develop a slightly modified IRLM that separates long-context units (LCUs) from short-context units and show that the LCUs alone achieve a state-of-the-art performance on the MRSC task of 60.8%. Our analysis indicates that a fruitful direction of research for neural LMs lies in developing more accessible internal representations, and suggests an optimization regime of very high momentum terms for effectively training such models.

Calcium imaging is a powerful method to record the activity of neural populations, but inferring spike times from calcium signals is a challenging problem. We compared multiple approaches using multiple datasets with ground truth electrophysiology, and found that simple non-negative deconvolution (NND) outperformed all other algorithms. We introduce a novel benchmark applicable to recordings without electrophysiological ground truth, based on the correlation of responses to two stimulus repeats, and used this to show that unconstrained NND also outperformed the other algorithms when run on 'zoomed out' datasets of ~10,000 cell recordings. Finally, we show that NND-based methods match the performance of a supervised method based on convolutional neural networks, while avoiding some of the biases of such methods, and at much faster running times. We therefore recommend that spikes be inferred from calcium traces using simple NND, due to its simplicity, efficiency and accuracy.

Calcium imaging is a powerful method to record the activity of neural populations in many species, but inferring spike times from calcium signals is a challenging problem. We compared multiple approaches using multiple datasets with ground truth electrophysiology, and found that simple non-negative deconvolution (NND) outperformed all other algorithms on out-of-sample test data. We introduce a novel benchmark applicable to recordings without electrophysiological ground truth, based on the correlation of responses to two stimulus repeats, and used this to show that unconstrained NND also outperformed the other algorithms when run on "zoomed out" datasets of ∼10,000 cell recordings from the visual cortex of mice of either sex. Finally, we show that NND-based methods match the performance of a supervised method based on convolutional neural networks, while avoiding some of the biases of such methods, and at much faster running times. We therefore recommend that spikes be inferred from calcium traces using simple NND, due to its simplicity, efficiency and accuracy.The experimental method that currently allows for recordings of the largest numbers of cells simultaneously is two-photon calcium imaging. However, use of this powerful method requires that neuronal firing times be inferred correctly from the large resulting datasets. Previous studies have claimed that complex supervised learning algorithms outperform simple deconvolution methods at this task. Unfortunately, these studies suffered from several problems and biases. When we repeated the analysis, using the same data and correcting these problems, we found that simpler spike inference methods perform better. Even more importantly, we found that supervised learning methods can introduce artifactual structure into spike trains, that can in turn lead to erroneous scientific conclusions. Of the algorithms we evaluated, we found that an extremely simple method performed best in all circumstances tested, was much faster to run, and was insensitive to parameter choices, making incorrect scientific conclusions much less likely.

Spike sorting is the computational process of extracting the firing times of single neurons from recordings of local electrical fields. This is an important but hard problem in neuroscience, complicated by the non-stationarity of the recordings and the dense overlap in electrical fields between nearby neurons. To solve the spike sorting problem, we have continuously developed over the past eight years a framework known as Kilosort. This paper describes the various algorithmic steps introduced in different versions of Kilosort. We also report the development of Kilosort4, a new version with substantially improved performance due to new clustering algorithms inspired by graph-based approaches. To test the performance of Kilosort, we developed a realistic simulation framework which uses densely sampled electrical fields from real experiments to generate non-stationary spike waveforms and realistic noise. We find that nearly all versions of Kilosort outperform other algorithms on a variety of simulated conditions, and Kilosort4 performs best in all cases, correctly identifying even neurons with low amplitudes and small spatial extents in high drift conditions.

Predictive coding is a theoretical framework that can explain how animals build internal models of their sensory environments by predicting sensory inputs. Predictive coding may capture either spatial or temporal relationships between sensory objects. While the original theory by Rao and Ballard, 1999 described spatial predictive coding, much of the recent experimental data has been interpreted as evidence for temporal predictive coding. Here we directly tested whether the “mismatch” neural responses in sensory cortex are due to a spatial or a temporal internal model. We adopted two common paradigms to study predictive coding: one based on virtual-reality and one based on static images. After training mice with repeated visual stimulation for several days, we performed multiple manipulations, including: 1) we introduced a novel stimulus, 2) we replaced a stimulus with a novel gray wall, 3) we duplicated a trained stimulus, or 4) we altered the order of the stimuli. The first two manipulations induced a substantial mismatch response in neural populations of up to 20,000 neurons recorded across primary and higher-order visual cortex, while the third and fourth ones did not. Thus, a mismatch response only occurred if a new spatial – not temporal – pattern was introduced.

Spike sorting is the computational process of extracting the firing times of single neurons from recordings of local electrical fields. This is an important but hard problem in neuroscience, made complicated by the nonstationarity of the recordings and the dense overlap in electrical fields between nearby neurons. To address the spike-sorting problem, we have been openly developing the Kilosort framework. Here we describe the various algorithmic steps introduced in different versions of Kilosort. We also report the development of Kilosort4, a version with substantially improved performance due to clustering algorithms inspired by graph-based approaches. To test the performance of Kilosort, we developed a realistic simulation framework that uses densely sampled electrical fields from real experiments to generate nonstationary spike waveforms and realistic noise. We found that nearly all versions of Kilosort outperformed other algorithms on a variety of simulated conditions and that Kilosort4 performed best in all cases, correctly identifying even neurons with low amplitudes and small spatial extents in high drift conditions.