Neuroscience is experiencing a data revolution in which simultaneous recording of many hundreds or thousands of neurons is revealing structure in population activity that is not apparent from single-neuron responses. This structure is typically extracted from trial-averaged data. Single-trial analyses are challenging due to incomplete sampling of the neural population, trial-to-trial variability, and fluctuations in action potential timing. Here we introduce Latent Factor Analysis via Dynamical Systems (LFADS), a deep learning method to infer latent dynamics from single-trial neural spiking data. LFADS uses a nonlinear dynamical system (a recurrent neural network) to infer the dynamics underlying observed population activity and to extract 'de-noised' single-trial firing rates from neural spiking data. We apply LFADS to a variety of monkey and human motor cortical datasets, demonstrating its ability to predict observed behavioral variables with unprecedented accuracy, extract precise estimates of neural dynamics on single trials, infer perturbations to those dynamics that correlate with behavioral choices, and combine data from non-overlapping recording sessions (spanning months) to improve inference of underlying dynamics. In summary, LFADS leverages all observations of a neural population's activity to accurately model its dynamics on single trials, opening the door to a detailed understanding of the role of dynamics in performing computation and ultimately driving behavior.Increasing evidence suggests that in many brain areas, such as the motor and prefrontal cortices, the activity of large populations of neurons, termed the neural population state, is often well-described by low-dimensional dynamics [e.g. (Afshar et al. 2011; Harvey, Coen, and Tank 2012; Kaufman et al. 2014;Sadtler et al. 2014;Kobak et al. 2016a) ]. Recovering these dynamics on single trials is essential for illuminating the relationship between neural population activity and behavior, and for advancing therapeutic neurotechnologies such as closed-loop deep brain stimulation and brain-machine interfaces. However, recovering population dynamics All rights reserved. No reuse allowed without permission.(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint . http://dx.doi.org/10.1101/152884 doi: bioRxiv preprint first posted online Jun. 20, 2017; on single trials is difficult due to trial-to-trial variability (e.g. in behavior or arousal) and fluctuations in the spiking of individual neurons. Even with dramatic increases in the numbers of neurons that can be simultaneously recorded using multichannel electrode arrays or optical imaging, accurately recovering population dynamics from single trials remains a significant challenge for data-analysis methods.Standard analyses sacrifice single-trial information for the sake of better estimates of trial-averaged neural states (Ahrens et al. 2012; Kobak et al. 2016b) . Techniques for extrac...

Speech brain-computer interfaces (BCIs) have the potential to restore rapid communication to people with paralysis by decoding neural activity evoked by attempted speaking movements into text or sound. Early demonstrations, while promising, have not yet achieved accuracies high enough for communication of unconstrainted sentences from a large vocabulary. Here, we demonstrate the first speech-to-text BCI that records spiking activity from intracortical microelectrode arrays. Enabled by these high-resolution recordings, our study participant, who can no longer speak intelligibly due amyotrophic lateral sclerosis (ALS), achieved a 9.1% word error rate on a 50 word vocabulary (2.7 times fewer errors than the prior state of the art speech BCI2) and a 23.8% word error rate on a 125,000 word vocabulary (the first successful demonstration of large-vocabulary decoding). Our BCI decoded speech at 62 words per minute, which is 3.4 times faster than the prior record for any kind of BCI and begins to approach the speed of natural conversation (160 words per minute). Finally, we highlight two aspects of the neural code for speech that are encouraging for speech BCIs: spatially intermixed tuning to speech articulators that makes accurate decoding possible from only a small region of cortex, and a detailed articulatory representation of phonemes that persists years after paralysis. These results show a feasible path forward for using intracortical speech BCIs to restore rapid communication to people with paralysis who can no longer speak.

Speaking is a sensorimotor behavior whose neural basis is difficult to study at the resolution of single neurons due to the scarcity of human intracortical measurements and the lack of animal models. We recorded from electrode arrays in the 'hand knob' 30 area of motor cortex in people with tetraplegia. Neurons in this area, which have not previously been implicated in speech, modulated during speaking and during nonspeaking movement of the tongue, lips, and jaw. This challenges whether the conventional model of a 'motor homunculus' division by major body regions extends to the single-neuron scale. Spoken words and syllables could be decoded from single 35 trials, demonstrating the potential utility of intracortical recordings for brain-computer interfaces (BCIs) to restore speech. Two neural population dynamics features previously reported for arm movements were also present during speaking: a large initial condition-invariant signal, followed by rotatory dynamics. This suggests that common neural dynamical motifs may underlie movement of arm and speech 40 articulators.

No abstract

For children who depend on devices to communicate, the rate of communication is a primary determinant of success. For children with motor impairments, the rate of communication may be limited by inability to contact buttons or cells rapidly or accurately. It is, therefore, essential to know how to adjust the device interface in order to maximize each child's rate of communication. The optimal rate of communication is determined by the channel capacity, which is the maximum value of the information rate for all possible keyboard button or cell layouts for the communication device. We construct a mathematical model for the information rate based on the relationship between movement time and the number of buttons per screen, the size of the buttons, and the length of a sequence of buttons that must be pressed to communicate each word in the vocabulary. We measure the parameters of the model using a custom-programmed touchscreen interface in 10 children with disorders of arm movement due to cerebral palsy who use a DynaVox communication device. We measure the same parameters in 20 healthy control subjects. We show that the model approximates the measured information rate and that the information rate is lower in children with motor impairments compared with control subjects. The theory predicts that for each child there is a combination of button size and number that maximizes the predicted information rate and thereby achieves communication at the optimal channel capacity. Programming communication devices with each child's predicted optimal parameters improved the communication rate in five of the ten children, compared with programming by professionals. Therefore, measurement of information rate may provide an assessment of the effect of motor disorders on success in assisted communication. Optimization of the information rate may be useful for programming assisted communication devices.

Background: During deep brain stimulation implant surgery, microelectrode recordings are used to map the location of targeted neurons. The effects produced by propofol or remifentanil on discharge activity of subthalamic neurons were studied intraoperatively to determine whether they alter neuronal activity. Methods: Microelectrode recordings from 11 neurons, each from individual patients, were discriminated and analyzed before and after administration of either propofol or remifentanil. Subthalamic neurons in rat brain slices were recorded in patch-clamp to investigate cellular level effects. Results: Neurons discharged at 42 Ϯ 9 spikes/s (mean Ϯ SD) and showed a common pattern of inhibition that lasted 4.3 ms. Unique discharge profiles were evident for each neuron, seen using joint-interval analysis. Propofol (intravenous bolus 0.3 mg/kg) produced sedation, with minor effects on discharge activity (less than 2.0% change in frequency). A prolongation of recurrent inhibition was evident from jointinterval analysis, and propofol's effect peaked within 2 min, with recovery evident at 10 min. Subthalamic neurons recorded in rat brain slices exhibited inhibitory synaptic currents that were prolonged by propofol (155%) but appeared to lack tonic inhibitory currents. Propofol did not alter membrane potential, membrane resistance, current-evoked discharge, or holding current during voltage clamp. Remifentanil (0.05 mg/kg) had little effect on overall subthalamic neuron discharge activity and did not prolong recurrent inhibition.

24Decades after the motor homunculus was first proposed, it is still unknown how different body 25 parts are intermixed and interrelated in human motor cortex at single-neuron resolution. Using 26 microelectrode arrays, we studied how face, head, arm and leg movements on both sides of the 27 body are represented in hand knob area of precentral gyrus in people with tetraplegia. Contrary 28 to the traditional somatotopy, we found strong representation of all movements. Probing further, 29 we found that ipsilateral and contralateral movements, and homologous arm and leg 30 movements (e.g. wrist and ankle), had a correlated representation. Additionally, there were 31 neural dimensions where the limb was represented independently of the movement. Together, 32 these patterns formed a "modular" code that might facilitate skill transfer across limbs. We also 33 investigated dual-effector movement, finding that more strongly represented effectors 34 suppressed the activity of weaker effectors. Finally, we leveraged these results to improve 35 discrete brain-computer interfaces by spreading targets across all limbs. 42intermixing between widely distinct areas of the body within any one individual (Leyton and 43 Sherrington 1917; W. Penfield and Boldrey 1937; Wilder Penfield and Rasmussen 1950). fMRI 44 studies also support the existence of an orderly map with largely separate face, arm and leg 45 areas along the precentral gyrus and the anterior bank of the central sulcus (Lotze et al. 2000; computer interface (BCI) that can decode movements across all four limbs. We show that this 86 "full body" BCI improves information throughput relative to a single-effector approach. 87 Results 89Tuning to Face, Head, Arm and Leg Movements 90 91We used microelectrode array recordings from participants T5 and T7 to assess tuning to face, 92 head, arm and leg movements in hand knob area of precentral gyrus. Participant T5 had a C4 93 spinal cord injury and was paralyzed from the neck down; he could move his face and head, but 94 attempted arm and leg movements resulted in little or no overt motion. Participant T7 had ALS 95and could move all joints tested, although some of his arm movements were limited due to 96 weakness. 98In this experiment, T5 and T7 made (or attempted to make) movements in sync with visual cues 99 displayed on a computer screen ( Fig. 1A). T5 completed an instructed delay version of the task 100 where each trial randomly cued one of 32 possible movements spanning the face, head, arm 101 and legs. For face and head movements, T5 was instructed to move normally; for arm and leg 102 movements, T5 was instructed to attempt to make the movement as if he were not paralyzed. 103T7, whose more restricted data was collected earlier in a different study (and who is no longer 104 enrolled), completed an alternating paired movement task with a block design. Each block 105 tested a different movement pair, during which T7 alternated between making each of the paired 106 movements every 3 seconds. 108Despite recording from micr...

This chapter addresses the following question: If we adopt the view that it is the brain that feels, thinks, and decides, then how do we accommodate commonsense explanations of human behavior and the notion that we are intentional rational agents capable of voluntary action? It argues that there are limits to the coexistence of folk psychology (and the notion that we are intentional rational agents) and neuroscience. It explores how neuroethics must accommodate both science and ethics and, drawing on contemporary studies of deception, lies, and others, urges an awareness of the limitations of neuroscience in determining thought and defining responsibility for actions.