Neural signatures of natural behaviour in socializing macaques.

Our understanding of the neurobiology of primate behaviour largely derives from artificial tasks in highly controlled laboratory settings, overlooking most natural behaviours that primate brains evolved to produce1-3. How primates navigate the multidimensional social relationships that structure daily life4 and shape survival and reproductive success5 remains largely unclear at the single-neuron level. Here we combine ethological analysis, computer vision and wireless recording technologies to identify neural signatures of natural behaviour in unrestrained, socially interacting pairs of rhesus macaques. Single-neuron and population activity in the prefrontal and temporal cortex robustly encoded 24 species-typical behaviours, as well as social context. Male-female partners demonstrated near-perfect reciprocity in grooming, a key behavioural mechanism supporting friendships and alliances6, and neural activity maintained a running account of these social investments. Confronted with an aggressive intruder, behavioural and neural population responses reflected empathy and were buffered by the presence of a partner. Our findings reveal a highly distributed neurophysiological ledger of social dynamics, a potential computational foundation supporting communal life in primate societies, including our own.

The neuroconnectionist research programme.

Artificial neural networks (ANNs) inspired by biology are beginning to be widely used to model behavioural and neural data, an approach we call 'neuroconnectionism'. ANNs have been not only lauded as the current best models of information processing in the brain but also criticized for failing to account for basic cognitive functions. In this Perspective article, we propose that arguing about the successes and failures of a restricted set of current ANNs is the wrong approach to assess the promise of neuroconnectionism for brain science. Instead, we take inspiration from the philosophy of science, and in particular from Lakatos, who showed that the core of a scientific research programme is often not directly falsifiable but should be assessed by its capacity to generate novel insights. Following this view, we present neuroconnectionism as a general research programme centred around ANNs as a computational language for expressing falsifiable theories about brain computation. We describe the core of the programme, the underlying computational framework and its tools for testing specific neuroscientific hypotheses and deriving novel understanding. Taking a longitudinal view, we review past and present neuroconnectionist projects and their responses to challenges and argue that the research programme is highly progressive, generating new and otherwise unreachable insights into the workings of the brain.

Causal mapping of human brain function.

Mapping human brain function is a long-standing goal of neuroscience that promises to inform the development of new treatments for brain disorders. Early maps of human brain function were based on locations of brain damage or brain stimulation that caused a functional change. Over time, this approach was largely replaced by technologies such as functional neuroimaging, which identify brain regions in which activity is correlated with behaviours or symptoms. Despite their advantages, these technologies reveal correlations, not causation. This creates challenges for interpreting the data generated from these tools and using them to develop treatments for brain disorders. A return to causal mapping of human brain function based on brain lesions and brain stimulation is underway. New approaches can combine these causal sources of information with modern neuroimaging and electrophysiology techniques to gain new insights into the functions of specific brain areas. In this Review, we provide a definition of causality for translational research, propose a continuum along which to assess the relative strength of causal information from human brain mapping studies and discuss recent advances in causal brain mapping and their relevance for developing treatments.

Neural spiking for causal inference and learning.

When a neuron is driven beyond its threshold, it spikes. The fact that it does not communicate its continuous membrane potential is usually seen as a computational liability. Here we show that this spiking mechanism allows neurons to produce an unbiased estimate of their causal influence, and a way of approximating gradient descent-based learning. Importantly, neither activity of upstream neurons, which act as confounders, nor downstream non-linearities bias the results. We show how spiking enables neurons to solve causal estimation problems and that local plasticity can approximate gradient descent using spike discontinuity learning.

A role for cortical interneurons as adversarial discriminators.

The brain learns representations of sensory information from experience, but the algorithms by which it does so remain unknown. One popular theory formalizes representations as inferred factors in a generative model of sensory stimuli, meaning that learning must improve this generative model and inference procedure. This framework underlies many classic computational theories of sensory learning, such as Boltzmann machines, the Wake/Sleep algorithm, and a more recent proposal that the brain learns with an adversarial algorithm that compares waking and dreaming activity. However, in order for such theories to provide insights into the cellular mechanisms of sensory learning, they must be first linked to the cell types in the brain that mediate them. In this study, we examine whether a subtype of cortical interneurons might mediate sensory learning by serving as discriminators, a crucial component in an adversarial algorithm for representation learning. We describe how such interneurons would be characterized by a plasticity rule that switches from Hebbian plasticity during waking states to anti-Hebbian plasticity in dreaming states. Evaluating the computational advantages and disadvantages of this algorithm, we find that it excels at learning representations in networks with recurrent connections but scales poorly with network size. This limitation can be partially addressed if the network also oscillates between evoked activity and generative samples on faster timescales. Consequently, we propose that an adversarial algorithm with interneurons as discriminators is a plausible and testable strategy for sensory learning in biological systems.

Inferring causal connectivity from pairwise recordings and optogenetics.

To understand the neural mechanisms underlying brain function, neuroscientists aim to quantify causal interactions between neurons, for instance by perturbing the activity of neuron A and measuring the effect on neuron B. Recently, manipulating neuron activity using light-sensitive opsins, optogenetics, has increased the specificity of neural perturbation. However, using widefield optogenetic interventions, multiple neurons are usually perturbed, producing a confound-any of the stimulated neurons can have affected the postsynaptic neuron making it challenging to discern which neurons produced the causal effect. Here, we show how such confounds produce large biases in interpretations. We explain how confounding can be reduced by combining instrumental variables (IV) and difference in differences (DiD) techniques from econometrics. Combined, these methods can estimate (causal) effective connectivity by exploiting the weak, approximately random signal resulting from the interaction between stimulation and the absolute refractory period of the neuron. In simulated neural networks, we find that estimates using ideas from IV and DiD outperform naïve techniques suggesting that methods from causal inference can be useful to disentangle neural interactions in the brain.

The study of plasticity has always been about gradients.

The experimental study of learning and plasticity has always been driven by an implicit question: how can physiological changes be adaptive and improve performance? For example, in Hebbian plasticity only synapses from presynaptic neurons that were active are changed, avoiding useless changes. Similarly, in dopamine-gated learning synapse changes depend on reward or lack thereof and do not change when everything is predictable. Within machine learning we can make the question of which changes are adaptive concrete: performance improves when changes correlate with the gradient of an objective function quantifying performance. This result is general for any system that improves through small changes. As such, physiology has always implicitly been seeking mechanisms that allow the brain to approximate gradients. Coming from this perspective we review the existing literature on plasticity-related mechanisms, and we show how these mechanisms relate to gradient estimation. We argue that gradients are a unifying idea to explain the many facets of neuronal plasticity.

Deep neural networks are not a single hypothesis but a language for expressing computational hypotheses.

An ideal vision model accounts for behavior and neurophysiology in both naturalistic conditions and designed lab experiments. Unlike psychological theories, artificial neural networks (ANNs) actually perform visual tasks and generate testable predictions for arbitrary inputs. These advantages enable ANNs to engage the entire spectrum of the evidence. Failures of particular models drive progress in a vibrant ANN research program of human vision.

Energy expenditure does not solely explain step length-width choices during walking.

Healthy young adults have a most preferred walking speed, step length and step width that are close to energetically optimal. However, people can choose to walk with a multitude of different step lengths and widths, which can vary in both energy expenditure and preference. Here, we further investigated step length-width preferences and their relationship to energy expenditure. In line with a growing body of research, we hypothesized that people's preferred stepping patterns would not be fully explained by metabolic energy expenditure. To test this hypothesis, we used a two-alternative forced-choice paradigm. Fifteen participants walked on an oversized treadmill. Each trial, participants performed two prescribed stepping patterns and then chose the pattern they preferred. Over time, we adapted the choices such that there was 50% chance of choosing one pattern over another (equally preferred). If people's preferences are based solely on metabolic energy expenditure, then these equally preferred stepping patterns should have equal energy expenditure. In contrast, we found that energy expenditure differed across equally preferred step length-width patterns (P<0.001). On average, longer steps with higher energy expenditure were preferred over shorter and wider steps with lower energy expenditure (P<0.001). We also asked participants to rank a set of shorter, wider and longer steps from most preferred to least preferred, and from most energy expended to least energy expended. Only 7/15 participants had the same rankings for their preferences and perceived energy expenditure. Our results suggest that energy expenditure is not the only factor influencing a person's conscious gait choices.

Recording Temporal Signals with Minutes Resolution Using Enzymatic DNA Synthesis.

Employing DNA as a high-density data storage medium has paved the way for next-generation digital storage and biosensing technologies. However, the multipart architecture of current DNA-based recording techniques renders them inherently slow and incapable of recording fluctuating signals with subhour frequencies. To address this limitation, we developed a simplified system employing a single enzyme, terminal deoxynucleotidyl transferase (TdT), to transduce environmental signals into DNA. TdT adds nucleotides to the 3'-ends of single-stranded DNA (ssDNA) in a template-independent manner, selecting bases according to inherent preferences and environmental conditions. By characterizing TdT nucleotide selectivity under different conditions, we show that TdT can encode various physiologically relevant signals such as Co2+, Ca2+, and Zn2+ concentrations and temperature changes in vitro. Further, by considering the average rate of nucleotide incorporation, we show that the resulting ssDNA functions as a molecular ticker tape. With this method we accurately encode a temporal record of fluctuations in Co2+ concentration to within 1 min over a 60 min period. Finally, we engineer TdT to allosterically turn off in the presence of a physiologically relevant concentration of calcium. We use this engineered TdT in concert with a reference TdT to develop a two-polymerase system capable of recording a single-step change in the Ca2+ signal to within 1 min over a 60 min period. This work expands the repertoire of DNA-based recording techniques by developing a novel DNA synthesis-based system that can record temporal environmental signals into DNA with a resolution of minutes.

On PDE Characterization of Smooth Hierarchical Functions Computed by Neural Networks.

Neural networks are versatile tools for computation, having the ability to approximate a broad range of functions. An important problem in the theory of deep neural networks is expressivity; that is, we want to understand the functions that are computable by a given network. We study real, infinitely differentiable (smooth) hierarchical functions implemented by feedforward neural networks via composing simpler functions in two cases: (1) each constituent function of the composition has fewer inputs than the resulting function and (2) constituent functions are in the more specific yet prevalent form of a nonlinear univariate function (e.g., tanh) applied to a linear multivariate function. We establish that in each of these regimes, there exist nontrivial algebraic partial differential equations (PDEs) that are satisfied by the computed functions. These PDEs are purely in terms of the partial derivatives and are dependent only on the topology of the network. Conversely, we conjecture that such PDE constraints, once accompanied by appropriate nonsingularity conditions and perhaps certain inequalities involving partial derivatives, guarantee that the smooth function under consideration can be represented by the network. The conjecture is verified in numerous examples, including the case of tree architectures, which are of neuroscientific interest. Our approach is a step toward formulating an algebraic description of functional spaces associated with specific neural networks, and may provide useful new tools for constructing neural networks.

Might a Single Neuron Solve Interesting Machine Learning Problems Through Successive Computations on Its Dendritic Tree?

Physiological experiments have highlighted how the dendrites of biological neurons can nonlinearly process distributed synaptic inputs. However, it is unclear how aspects of a dendritic tree, such as its branched morphology or its repetition of presynaptic inputs, determine neural computation beyond this apparent nonlinearity. Here we use a simple model where the dendrite is implemented as a sequence of thresholded linear units. We manipulate the architecture of this model to investigate the impacts of binary branching constraints and repetition of synaptic inputs on neural computation. We find that models with such manipulations can perform well on machine learning tasks, such as Fashion MNIST or Extended MNIST. We find that model performance on these tasks is limited by binary tree branching and dendritic asymmetry and is improved by the repetition of synaptic inputs to different dendritic branches. These computational experiments further neuroscience theory on how different dendritic properties might determine neural computation of clearly defined tasks.

Machine Learning and Surgical Outcomes Prediction: A Systematic Review.

BACKGROUND: Machine learning (ML) has garnered increasing attention as a means to quantitatively analyze the growing and complex medical data to improve individualized patient care. We herein aim to critically examine the current state of ML in predicting surgical outcomes, evaluate the quality of currently available research, and propose areas of improvement for future uses of ML in surgery. METHODS: A systematic review was conducted in accordance with the Preferred Reporting Items for a Systematic Review and Meta-Analysis (PRISMA) checklist. PubMed, MEDLINE, and Embase databases were reviewed under search syntax "machine learning" and "surgery" for papers published between 2015 and 2020. RESULTS: Of the initial 2677 studies, 45 papers met inclusion and exclusion criteria. Fourteen different subspecialties were represented with neurosurgery being most common. The most frequently used ML algorithms were random forest (n = 19), artificial neural network (n = 17), and logistic regression (n = 17). Common outcomes included postoperative mortality, complications, patient reported quality of life and pain improvement. All studies which compared ML algorithms to conventional studies which used area under the curve (AUC) to measure accuracy found improved outcome prediction with ML models. CONCLUSIONS: While still in its early stages, ML models offer surgeons an opportunity to capitalize on the myriad of clinical data available and improve individualized patient care. Limitations included heterogeneous outcome and imperfect quality of some of the papers. We therefore urge future research to agree upon methods of outcome reporting and require basic quality standards.

From Prior Information to Saccade Selection: Evolution of Frontal Eye Field Activity during Natural Scene Search.

Prior knowledge about our environment influences our actions. How does this knowledge evolve into a final action plan and how does the brain represent this? Here, we investigated this question in the monkey oculomotor system during self-guided search of natural scenes. In the frontal eye field (FEF), we found a subset of neurons, "Early neurons," that contain information about the upcoming saccade long before it is executed, often before the previous saccade had even ended. Crucially, much of this early information did not relate to the actual saccade that would eventually be selected. Rather, it related to prior information about the probabilities of possible upcoming saccades based on the presaccade fixation location. Nearer to the time of saccade onset, a greater proportion of these neurons' activities related to the saccade selection, although prior information continued to influence activity throughout. A separate subset of FEF neurons, "Late neurons," only represented the final action plan near saccade onset and not prior information. Our results demonstrate how, across the population of FEF neurons, prior information evolves into definitive saccade plans.

Evaluation of Changes in Depression, Anxiety, and Social Anxiety Using Smartphone Sensor Features: Longitudinal Cohort Study.

BACKGROUND: The assessment of behaviors related to mental health typically relies on self-report data. Networked sensors embedded in smartphones can measure some behaviors objectively and continuously, with no ongoing effort. OBJECTIVE: This study aims to evaluate whether changes in phone sensor-derived behavioral features were associated with subsequent changes in mental health symptoms. METHODS: This longitudinal cohort study examined continuously collected phone sensor data and symptom severity data, collected every 3 weeks, over 16 weeks. The participants were recruited through national research registries. Primary outcomes included depression (8-item Patient Health Questionnaire), generalized anxiety (Generalized Anxiety Disorder 7-item scale), and social anxiety (Social Phobia Inventory) severity. Participants were adults who owned Android smartphones. Participants clustered into 4 groups: multiple comorbidities, depression and generalized anxiety, depression and social anxiety, and minimal symptoms. RESULTS: A total of 282 participants were aged 19-69 years (mean 38.9, SD 11.9 years), and the majority were female (223/282, 79.1%) and White participants (226/282, 80.1%). Among the multiple comorbidities group, depression changes were preceded by changes in GPS features (Time: r=-0.23, P=.02; Locations: r=-0.36, P<.001), exercise duration (r=0.39; P=.03) and use of active apps (r=-0.31; P<.001). Among the depression and anxiety groups, changes in depression were preceded by changes in GPS features for Locations (r=-0.20; P=.03) and Transitions (r=-0.21; P=.03). Depression changes were not related to subsequent sensor-derived features. The minimal symptoms group showed no significant relationships. There were no associations between sensor-based features and anxiety and minimal associations between sensor-based features and social anxiety. CONCLUSIONS: Changes in sensor-derived behavioral features are associated with subsequent depression changes, but not vice versa, suggesting a directional relationship in which changes in sensed behaviors are associated with subsequent changes in symptoms.

A smartphone-based online system for fall detection with alert notifications and contextual information of real-life falls.

BACKGROUND: Falls are a leading cause of accidental deaths and injuries worldwide. The risk of falling is especially high for individuals suffering from balance impairments. Retrospective surveys and studies of simulated falling in lab conditions are frequently used and are informative, but prospective information about real-life falls remains sparse. Such data are essential to address fall risks and develop fall detection and alert systems. Here we present the results of a prospective study investigating a proof-of-concept, smartphone-based, online system for fall detection and notification. METHODS: The system uses the smartphone's accelerometer and gyroscope to monitor the participants' motion, and falls are detected using a regularized logistic regression. Data on falls and near-fall events (i.e., stumbles) is stored in a cloud server and fall-related variables are logged onto a web portal developed for data exploration, including the event time and weather, fall probability, and the faller's location and activity before the fall. RESULTS: In total, 23 individuals with an elevated risk of falling carried the phones for 2070 days in which the model classified 14,904,000 events. The system detected 27 of the 37 falls that occurred (sensitivity = 73.0 %) and resulted in one false alarm every 46 days (specificity > 99.9 %, precision = 37.5 %). 42.2 % of the events falsely classified as falls were validated as stumbles. CONCLUSIONS: The system's performance shows the potential of using smartphones for fall detection and notification in real-life. Apart from functioning as a practical fall monitoring instrument, this system may serve as a valuable research tool, enable future studies to scale their ability to capture fall-related data, and help researchers and clinicians to investigate real-falls.

High-resolution mapping of DNA polymerase fidelity using nucleotide imbalances and next-generation sequencing.

DNA polymerase fidelity is affected by both intrinsic properties and environmental conditions. Current strategies for measuring DNA polymerase error rate in vitro are constrained by low error subtype sensitivity, poor scalability, and lack of flexibility in types of sequence contexts that can be tested. We have developed the Magnification via Nucleotide Imbalance Fidelity (MagNIFi) assay, a scalable next-generation sequencing assay that uses a biased deoxynucleotide pool to quantitatively shift error rates into a range where errors are frequent and hence measurement is robust, while still allowing for accurate mapping to error rates under typical conditions. This assay is compatible with a wide range of fidelity-modulating conditions, and enables high-throughput analysis of sequence context effects on base substitution and single nucleotide deletion fidelity using a built-in template library. We validate this assay by comparing to previously established fidelity metrics, and use it to investigate neighboring sequence-mediated effects on fidelity for several DNA polymerases. Through these demonstrations, we establish the MagNIFi assay for robust, high-throughput analysis of DNA polymerase fidelity.

Policies or knowledge: priors differ between a perceptual and sensorimotor task.

If the brain abstractly represents probability distributions as knowledge, then the modality of a decision, e.g., movement vs. perception, should not matter. If, on the other hand, learned representations are policies, they may be specific to the task where learning takes place. Here, we test this by asking whether a learned spatial prior generalizes from a sensorimotor estimation task to a two-alternative-forced choice (2-Afc) perceptual comparison task. A model and simulation-based analysis revealed that while participants learn prior distribution in the sensorimotor estimation task, measured priors are consistently broader than sensorimotor priors in the 2-Afc task. That the prior does not fully generalize suggests that sensorimotor priors are more like policies than knowledge. In disagreement with standard Bayesian thought, the modality of the decision has a strong influence on the implied prior distributions. NEW & NOTEWORTHY We do not know whether the brain represents abstract and generalizable knowledge or task-specific policies that map internal states to actions. We find that learning in a sensorimotor task does not generalize strongly to a perceptual task, suggesting that humans learned policies and did not truly acquire knowledge. Priors differ across tasks, thus casting doubt on the central tenet of many Bayesian models, that the brain's representation of the world is built on generalizable knowledge.

Sensorimotor priors are effector dependent.

During sensorimotor tasks, subjects use sensory feedback but also prior information. It is often assumed that the sensorimotor prior is just given by the experiment and that the details for acquiring this prior (e.g., the effector) are irrelevant. However, recent research has suggested that the construction of priors is nontrivial. To test if the sensorimotor details matter for the construction of a prior, we designed two tasks that differ only in the effectors that subjects use to indicate their estimate. For both a typical reaching setting and an atypical wrist rotation setting, prior and feedback uncertainty matter as quantitatively predicted by Bayesian statistics. However, in violation of simple Bayesian models, the importance of the prior differs across effectors. Subjects overly rely on their prior in the typical reaching case compared with the wrist case. The brain is not naively Bayesian with a single and veridical prior. NEW & NOTEWORTHY Traditional Bayesian models often assume that we learn statistics of movements and use the information as a prior to guide subsequent movements. The effector is merely a reporting modality for information processing. We asked subjects to perform a visuomotor learning task with different effectors (finger or wrist). Surprisingly, we found that prior information is used differently between the effectors, suggesting that learning of the prior is related to the movement context such as the effector involved or that naive models of Bayesian behavior need to be extended.