Burst-dependent synaptic plasticity can coordinate learning in hierarchical circuits

Vectorized instructive signals in cortical dendrites during a brain-computer interface task

Vectorization of teaching signals is a key element of virtually all modern machine learning algorithms, including backpropagation, target propagation and reinforcement learning. Vectorization allows a scalable and computationally efficient solution to the credit assignment problem by tailoring instructive signals to individual neurons. Recent theoretical models have suggested that neural circuits could implement single-phase vectorized learning at the cellular level by processing feedforward and feedback information streams in separate dendritic compartments1-5. This presents a compelling, but untested, hypothesis for how cortical circuits could solve credit assignment in the brain. We leveraged a neurofeedback brain-computer interface (BCI) task with an experimenter-defined reward function to test for vectorized instructive signals in dendrites. We trained mice to modulate the activity of two spatially intermingled populations (4 or 5 neurons each) of layer 5 pyramidal neurons in the retrosplenial cortex to rotate a visual grating towards a target orientation while we recorded GCaMP activity from somas and corresponding distal apical dendrites. We observed that the relative magnitudes of somatic versus dendritic signals could be predicted using the activity of the surrounding network and contained information about task-related variables that could serve as instructive signals, including reward and error. The signs of these putative teaching signals both depended on the causal role of individual neurons in the task and predicted changes in overall activity over the course of learning. Furthermore, targeted optogenetic perturbation of these signals disrupted learning. These results provide the first biological evidence of a vectorized instructive signal in the brain, implemented via semi-independent computation in cortical dendrites, unveiling a potential mechanism for solving credit assignment in the brain.

Ornstein-Uhlenbeck Adaptation as a Mechanism for Learning in Brains and Machines

Learning is a fundamental property of intelligent systems, observed across biological organisms and engineered systems. While modern intelligent systems typically rely on gradient descent for learning, the need for exact gradients and complex information flow makes its implementation in biological and neuromorphic systems challenging. This has motivated the exploration of alternative learning mechanisms that can operate locally and do not rely on exact gradients. In this work, we introduce a novel approach that leverages noise in the parameters of the system and global reinforcement signals. Using an Ornstein-Uhlenbeck process with adaptive dynamics, our method balances exploration and exploitation during learning, driven by deviations from error predictions, akin to reward prediction error. Operating in continuous time, Ornstein-Uhlenbeck adaptation (OUA) is proposed as a general mechanism for learning in dynamic, time-evolving environments. We validate our approach across a range of different tasks, including supervised learning and reinforcement learning in feedforward and recurrent systems. Additionally, we demonstrate that it can perform meta-learning, adjusting hyper-parameters autonomously. Our results indicate that OUA provides a promising alternative to traditional gradient-based methods, with potential applications in neuromorphic computing. It also hints at a possible mechanism for noise-driven learning in the brain, where stochastic neurotransmitter release may guide synaptic adjustments.

Neuroscientific insights about computer vision models: a concise review

The development of biologically-inspired computational models has been the focus of study ever since the artificial neuron was introduced by McCulloch and Pitts in 1943. However, a scrutiny of literature reveals that most attempts to replicate the highly efficient and complex biological visual system have been futile or have met with limited success. The recent state-of the-art computer vision models, such as pre-trained deep neural networks and vision transformers, may not be biologically inspired per se. Nevertheless, certain aspects of biological vision are still found embedded, knowingly or unknowingly, in the architecture and functioning of these models. This paper explores several principles related to visual neuroscience and the biological visual pathway that resonate, in some manner, in the architectural design and functioning of contemporary computer vision models. The findings of this survey can provide useful insights for building futuristic bio-inspired computer vision models. The survey is conducted from a historical perspective, tracing the biological connections of computer vision models starting with the basic artificial neuron to modern technologies such as deep convolutional neural network (CNN) and spiking neural networks (SNN). One spotlight of the survey is a discussion on biologically plausible neural networks and bio-inspired unsupervised learning mechanisms adapted for computer vision tasks in recent times.

The backpropagation algorithm implemented on spiking neuromorphic hardware

The capabilities of natural neural systems have inspired both new generations of machine learning algorithms as well as neuromorphic, very large-scale integrated circuits capable of fast, low-power information processing. However, it has been argued that most modern machine learning algorithms are not neurophysiologically plausible. In particular, the workhorse of modern deep learning, the backpropagation algorithm, has proven difficult to translate to neuromorphic hardware. This study presents a neuromorphic, spiking backpropagation algorithm based on synfire-gated dynamical information coordination and processing implemented on Intel's Loihi neuromorphic research processor. We demonstrate a proof-of-principle three-layer circuit that learns to classify digits and clothing items from the MNIST and Fashion MNIST datasets. To our knowledge, this is the first work to show a Spiking Neural Network implementation of the exact backpropagation algorithm that is fully on-chip without a computer in the loop. It is competitive in accuracy with off-chip trained SNNs and achieves an energy-delay product suitable for edge computing. This implementation shows a path for using in-memory, massively parallel neuromorphic processors for low-power, low-latency implementation of modern deep learning applications.

Large-scale interactions in predictive processing: oscillatory versus transient dynamics

How do the two main types of neural dynamics, aperiodic transients and oscillations, contribute to the interactions between feedforward (FF) and feedback (FB) pathways in sensory inference and predictive processing? We discuss three theoretical perspectives. First, we critically evaluate the theory that gamma and alpha/beta rhythms play a role in classic hierarchical predictive coding (HPC) by mediating FF and FB communication, respectively. Second, we outline an alternative functional model in which rapid sensory inference is mediated by aperiodic transients, whereas oscillations contribute to the stabilization of neural representations over time and plasticity processes. Third, we propose that the strong dependence of oscillations on predictability can be explained based on a biologically plausible alternative to classic HPC, namely dendritic HPC.

Theoretical framework for learning through structural plasticity

A growing body of research indicates that structural plasticity mechanisms are crucial for learning and memory consolidation. Starting from a simple phenomenological model, we exploit a mean-field approach to develop a theoretical framework of learning through this kind of plasticity, capable of taking into account several features of the connectivity and pattern of activity of biological neural networks, including probability distributions of neuron firing rates, selectivity of the responses of single neurons to multiple stimuli, probabilistic connection rules, and noisy stimuli. More importantly, it describes the effects of stabilization, pruning, and reorganization of synaptic connections. This framework is used to compute the values of some relevant quantities used to characterize the learning and memory capabilities of the neuronal network in training and testing procedures as the number of training patterns and other model parameters vary. The results are then compared with those obtained through simulations with firing-rate-based neuronal network models.

A neural network model of differentiation and integration of competing memories

What determines when neural representations of memories move together (integrate) or apart (differentiate)? Classic supervised learning models posit that, when two stimuli predict similar outcomes, their representations should integrate. However, these models have recently been challenged by studies showing that pairing two stimuli with a shared associate can sometimes cause differentiation, depending on the parameters of the study and the brain region being examined. Here, we provide a purely unsupervised neural network model that can explain these and other related findings. The model can exhibit integration or differentiation depending on the amount of activity allowed to spread to competitors - inactive memories are not modified, connections to moderately active competitors are weakened (leading to differentiation), and connections to highly active competitors are strengthened (leading to integration). The model also makes several novel predictions - most importantly, that when differentiation occurs as a result of this unsupervised learning mechanism, it will be rapid and asymmetric, and it will give rise to anticorrelated representations in the region of the brain that is the source of the differentiation. Overall, these modeling results provide a computational explanation for a diverse set of seemingly contradictory empirical findings in the memory literature, as well as new insights into the dynamics at play during learning.

Learning to integrate parts for whole through correlated neural variability

Neural activity in the cortex exhibits a wide range of firing variability and rich correlation structures. Studies on neural coding indicate that correlated neural variability can influence the quality of neural codes, either beneficially or adversely. However, the mechanisms by which correlated neural variability is transformed and processed across neural populations to achieve meaningful computation remain largely unclear. Here we propose a theory of covariance computation with spiking neurons which offers a unifying perspective on neural representation and computation with correlated noise. We employ a recently proposed computational framework known as the moment neural network to resolve the nonlinear coupling of correlated neural variability with a task-driven approach to constructing neural network models for performing covariance-based perceptual tasks. In particular, we demonstrate how perceptual information initially encoded entirely within the covariance of upstream neurons' spiking activity can be passed, in a near-lossless manner, to the mean firing rate of downstream neurons, which in turn can be used to inform inference. The proposed theory of covariance computation addresses an important question of how the brain extracts perceptual information from noisy sensory stimuli to generate a stable perceptual whole and indicates a more direct role that correlated variability plays in cortical information processing.

How auditory neurons count temporal intervals and decode information

The numerical sense of animals includes identifying the numerosity of a sequence of events that occur with specific intervals, e.g., notes in a call or bar of music. Across nervous systems, the temporal patterning of spikes can code these events, but how this information is decoded (counted) remains elusive. In the anuran auditory system, temporal information of this type is decoded in the midbrain, where "interval-counting" neurons spike only after at least a threshold number of sound pulses have occurred with specific timing. We show that this decoding process, i.e., interval counting, arises from integrating phasic, onset-type and offset inhibition with excitation that augments across successive intervals, possibly due to a progressive decrease in "shunting" effects of inhibition. Because these physiological properties are ubiquitous within and across central nervous systems, interval counting may be a general mechanism for decoding diverse information coded/encoded in temporal patterns of spikes, including "bursts," and estimating elapsed time.

Learning to use landmarks for navigation amplifies their representation in retrosplenial cortex

Visual landmarks provide powerful reference signals for efficient navigation by altering the activity of spatially tuned neurons, such as place cells, head direction cells, and grid cells. To understand the neural mechanism by which landmarks exert such strong influence, it is necessary to identify how these visual features gain spatial meaning. In this study, we characterized visual landmark representations in mouse retrosplenial cortex (RSC) using chronic two-photon imaging of the same neuronal ensembles over the course of spatial learning. We found a pronounced increase in landmark-referenced activity in RSC neurons that, once established, remained stable across days. Changing behavioral context by uncoupling treadmill motion from visual feedback systematically altered neuronal responses associated with the coherence between visual scene flow speed and self-motion. To explore potential underlying mechanisms, we modeled how burst firing, mediated by supralinear somatodendritic interactions, could efficiently mediate context- and coherence-dependent integration of landmark information. Our results show that visual encoding shifts to landmark-referenced and context-dependent codes as these cues take on spatial meaning during learning.

Half a century legacy of long-term potentiation

In 1973, two papers from Bliss and Lømo and from Bliss and Gardner-Medwin reported that high-frequency synaptic stimulation in the dentate gyrus of rabbits resulted in a long-lasting increase in synaptic strength. This form of synaptic plasticity, commonly referred to as long-term potentiation (LTP), was immediately considered as an attractive mechanism accounting for the ability of the brain to store information. In this historical piece looking back over the past 50 years, we discuss how these two landmark contributions directly motivated a colossal research effort and detail some of the resulting milestones that have shaped our evolving understanding of the molecular and cellular underpinnings of LTP. We highlight the main features of LTP, cover key experiments that defined its induction and expression mechanisms, and outline the evidence supporting a potential role of LTP in learning and memory. We also briefly explore some ramifications of LTP on network stability, consider current limitations of LTP as a model of associative memory, and entertain future research orientations.

Differentiation and Integration of Competing Memories: A Neural Network Model

What determines when neural representations of memories move together (integrate) or apart (differentiate)? Classic supervised learning models posit that, when two stimuli predict similar outcomes, their representations should integrate. However, these models have recently been challenged by studies showing that pairing two stimuli with a shared associate can sometimes cause differentiation, depending on the parameters of the study and the brain region being examined. Here, we provide a purely unsupervised neural network model that can explain these and other related findings. The model can exhibit integration or differentiation depending on the amount of activity allowed to spread to competitors - inactive memories are not modified, connections to moderately active competitors are weakened (leading to differentiation), and connections to highly active competitors are strengthened (leading to integration). The model also makes several novel predictions - most importantly, that when differentiation occurs as a result of this unsupervised learning mechanism, it will be rapid and asymmetric, and it will give rise to anticorrelated representations in the region of the brain that is the source of the differentiation. Overall, these modeling results provide a computational explanation for a diverse set of seemingly contradictory empirical findings in the memory literature, as well as new insights into the dynamics at play during learning.

Desiderata for Normative Models of Synaptic Plasticity

Normative models of synaptic plasticity use computational rationales to arrive at predictions of behavioral and network-level adaptive phenomena. In recent years, there has been an explosion of theoretical work in this realm, but experimental confirmation remains limited. In this review, we organize work on normative plasticity models in terms of a set of desiderata that, when satisfied, are designed to ensure that a given model demonstrates a clear link between plasticity and adaptive behavior, is consistent with known biological evidence about neural plasticity and yields specific testable predictions. As a prototype, we include a detailed analysis of the REINFORCE algorithm. We also discuss how new models have begun to improve on the identified criteria and suggest avenues for further development. Overall, we provide a conceptual guide to help develop neural learning theories that are precise, powerful, and experimentally testable.

Information bottleneck-based Hebbian learning rule naturally ties working memory and synaptic updates

Deep neural feedforward networks are effective models for a wide array of problems, but training and deploying such networks presents a significant energy cost. Spiking neural networks (SNNs), which are modeled after biologically realistic neurons, offer a potential solution when deployed correctly on neuromorphic computing hardware. Still, many applications train SNNs offline, and running network training directly on neuromorphic hardware is an ongoing research problem. The primary hurdle is that back-propagation, which makes training such artificial deep networks possible, is biologically implausible. Neuroscientists are uncertain about how the brain would propagate a precise error signal backward through a network of neurons. Recent progress addresses part of this question, e.g., the weight transport problem, but a complete solution remains intangible. In contrast, novel learning rules based on the information bottleneck (IB) train each layer of a network independently, circumventing the need to propagate errors across layers. Instead, propagation is implicit due the layers' feedforward connectivity. These rules take the form of a three-factor Hebbian update a global error signal modulates local synaptic updates within each layer. Unfortunately, the global signal for a given layer requires processing multiple samples concurrently, and the brain only sees a single sample at a time. We propose a new three-factor update rule where the global signal correctly captures information across samples via an auxiliary memory network. The auxiliary network can be trained a priori independently of the dataset being used with the primary network. We demonstrate comparable performance to baselines on image classification tasks. Interestingly, unlike back-propagation-like schemes where there is no link between learning and memory, our rule presents a direct connection between working memory and synaptic updates. To the best of our knowledge, this is the first rule to make this link explicit. We explore these implications in initial experiments examining the effect of memory capacity on learning performance. Moving forward, this work suggests an alternate view of learning where each layer balances memory-informed compression against task performance. This view naturally encompasses several key aspects of neural computation, including memory, efficiency, and locality.

Co-dependent excitatory and inhibitory plasticity accounts for quick, stable and long-lasting memories in biological networks

The brain's functionality is developed and maintained through synaptic plasticity. As synapses undergo plasticity, they also affect each other. The nature of such 'co-dependency' is difficult to disentangle experimentally, because multiple synapses must be monitored simultaneously. To help understand the experimentally observed phenomena, we introduce a framework that formalizes synaptic co-dependency between different connection types. The resulting model explains how inhibition can gate excitatory plasticity while neighboring excitatory-excitatory interactions determine the strength of long-term potentiation. Furthermore, we show how the interplay between excitatory and inhibitory synapses can account for the quick rise and long-term stability of a variety of synaptic weight profiles, such as orientation tuning and dendritic clustering of co-active synapses. In recurrent neuronal networks, co-dependent plasticity produces rich and stable motor cortex-like dynamics with high input sensitivity. Our results suggest an essential role for the neighborly synaptic interaction during learning, connecting micro-level physiology with network-wide phenomena.

Neuromorphic one-shot learning utilizing a phase-transition material

Design of hardware based on biological principles of neuronal computation and plasticity in the brain is a leading approach to realizing energy- and sample-efficient AI and learning machines. An important factor in selection of the hardware building blocks is the identification of candidate materials with physical properties suitable to emulate the large dynamic ranges and varied timescales of neuronal signaling. Previous work has shown that the all-or-none spiking behavior of neurons can be mimicked by threshold switches utilizing material phase transitions. Here, we demonstrate that devices based on a prototypical metal-insulator-transition material, vanadium dioxide (VO2), can be dynamically controlled to access a continuum of intermediate resistance states. Furthermore, the timescale of their intrinsic relaxation can be configured to match a range of biologically relevant timescales from milliseconds to seconds. We exploit these device properties to emulate three aspects of neuronal analog computation: fast (~1 ms) spiking in a neuronal soma compartment, slow (~100 ms) spiking in a dendritic compartment, and ultraslow (~1 s) biochemical signaling involved in temporal credit assignment for a recently discovered biological mechanism of one-shot learning. Simulations show that an artificial neural network using properties of VO2 devices to control an agent navigating a spatial environment can learn an efficient path to a reward in up to fourfold fewer trials than standard methods. The phase relaxations described in our study may be engineered in a variety of materials and can be controlled by thermal, electrical, or optical stimuli, suggesting further opportunities to emulate biological learning in neuromorphic hardware.

Specific connectivity optimizes learning in thalamocortical loops

Thalamocortical loops have a central role in cognition and motor control, but precisely how they contribute to these processes is unclear. Recent studies showing evidence of plasticity in thalamocortical synapses indicate a role for the thalamus in shaping cortical dynamics through learning. Since signals undergo a compression from the cortex to the thalamus, we hypothesized that the computational role of the thalamus depends critically on the structure of corticothalamic connectivity. To test this, we identified the optimal corticothalamic structure that promotes biologically plausible learning in thalamocortical synapses. We found that corticothalamic projections specialized to communicate an efference copy of the cortical output benefit motor control, while communicating the modes of highest variance is optimal for working memory tasks. We analyzed neural recordings from mice performing grasping and delayed discrimination tasks and found corticothalamic communication consistent with these predictions. These results suggest that the thalamus orchestrates cortical dynamics in a functionally precise manner through structured connectivity.

Leveraging dendritic properties to advance machine learning and neuro-inspired computing

The brain is a remarkably capable and efficient system. It can process and store huge amounts of noisy and unstructured information, using minimal energy. In contrast, current artificial intelligence (AI) systems require vast resources for training while still struggling to compete in tasks that are trivial for biological agents. Thus, brain-inspired engineering has emerged as a promising new avenue for designing sustainable, next-generation AI systems. Here, we describe how dendritic mechanisms of biological neurons have inspired innovative solutions for significant AI problems, including credit assignment in multi-layer networks, catastrophic forgetting, and high-power consumption. These findings provide exciting alternatives to existing architectures, showing how dendritic research can pave the way for building more powerful and energy efficient artificial learning systems.

Neurobiological Causal Models of Language Processing

The language faculty is physically realized in the neurobiological infrastructure of the human brain. Despite significant efforts, an integrated understanding of this system remains a formidable challenge. What is missing from most theoretical accounts is a specification of the neural mechanisms that implement language function. Computational models that have been put forward generally lack an explicit neurobiological foundation. We propose a neurobiologically informed causal modeling approach which offers a framework for how to bridge this gap. A neurobiological causal model is a mechanistic description of language processing that is grounded in, and constrained by, the characteristics of the neurobiological substrate. It intends to model the generators of language behavior at the level of implementational causality. We describe key features and neurobiological component parts from which causal models can be built and provide guidelines on how to implement them in model simulations. Then we outline how this approach can shed new light on the core computational machinery for language, the long-term storage of words in the mental lexicon and combinatorial processing in sentence comprehension. In contrast to cognitive theories of behavior, causal models are formulated in the "machine language" of neurobiology which is universal to human cognition. We argue that neurobiological causal modeling should be pursued in addition to existing approaches. Eventually, this approach will allow us to develop an explicit computational neurobiology of language.

Frequency Propagation: Multimechanism Learning in Nonlinear Physical Networks

We introduce frequency propagation, a learning algorithm for nonlinear physical networks. In a resistive electrical circuit with variable resistors, an activation current is applied at a set of input nodes at one frequency and an error current is applied at a set of output nodes at another frequency. The voltage response of the circuit to these boundary currents is the superposition of an activation signal and an error signal whose coefficients can be read in different frequencies of the frequency domain. Each conductance is updated proportionally to the product of the two coefficients. The learning rule is local and proved to perform gradient descent on a loss function. We argue that frequency propagation is an instance of a multimechanism learning strategy for physical networks, be it resistive, elastic, or flow networks. Multimechanism learning strategies incorporate at least two physical quantities, potentially governed by independent physical mechanisms, to act as activation and error signals in the training process. Locally available information about these two signals is then used to update the trainable parameters to perform gradient descent. We demonstrate how earlier work implementing learning via chemical signaling in flow networks (Anisetti, Scellier, et al., 2023) also falls under the rubric of multimechanism learning.

Impact on backpropagation of the spatial heterogeneity of sodium channel kinetics in the axon initial segment

In a variety of neurons, action potentials (APs) initiate at the proximal axon, within a region called the axon initial segment (AIS), which has a high density of voltage-gated sodium channels (NaVs) on its membrane. In pyramidal neurons, the proximal AIS has been reported to exhibit a higher proportion of NaVs with gating properties that are "right-shifted" to more depolarized voltages, compared to the distal AIS. Further, recent experiments have revealed that as neurons develop, the spatial distribution of NaV subtypes along the AIS can change substantially, suggesting that neurons tune their excitability by modifying said distribution. When neurons are stimulated axonally, computational modelling has shown that this spatial separation of gating properties in the AIS enhances the backpropagation of APs into the dendrites. In contrast, in the more natural scenario of somatic stimulation, our simulations show that the same distribution can impede backpropagation, suggesting that the choice of orthodromic versus antidromic stimulation can bias or even invert experimental findings regarding the role of NaV subtypes in the AIS. We implemented a range of hypothetical NaV distributions in the AIS of three multicompartmental pyramidal cell models and investigated the precise kinetic mechanisms underlying such effects, as the spatial distribution of NaV subtypes is varied. With axonal stimulation, proximal NaV availability dominates, such that concentrating right-shifted NaVs in the proximal AIS promotes backpropagation. However, with somatic stimulation, the models are insensitive to availability kinetics. Instead, the higher activation threshold of right-shifted NaVs in the AIS impedes backpropagation. Therefore, recently observed developmental changes to the spatial separation and relative proportions of NaV1.2 and NaV1.6 in the AIS differentially impact activation and availability. The observed effects on backpropagation, and potentially learning via its putative role in synaptic plasticity (e.g. through spike-timing-dependent plasticity), are opposite for orthodromic versus antidromic stimulation, which should inform hypotheses about the impact of the developmentally regulated subcellular localization of these NaV subtypes.

Functional subtypes of synaptic dynamics in mouse and human

Synapses preferentially respond to particular temporal patterns of activity with a large degree of heterogeneity that is informally or tacitly separated into classes. Yet, the precise number and properties of such classes are unclear. Do they exist on a continuum and, if so, when is it appropriate to divide that continuum into functional regions? In a large dataset of glutamatergic cortical connections, we perform model-based characterization to infer the number and characteristics of functionally distinct subtypes of synaptic dynamics. In rodent data, we find five clusters that partially converge with transgenic-associated subtypes. Strikingly, the application of the same clustering method in human data infers a highly similar number of clusters, supportive of stable clustering. This nuanced dictionary of functional subtypes shapes the heterogeneity of cortical synaptic dynamics and provides a lens into the basic motifs of information transmission in the brain.

Auditory Corticofugal Neurons Transmit Auditory and Non-auditory Information During Behavior

Layer 5 pyramidal neurons of sensory cortices project "corticofugal" axons to myriad sub-cortical targets, thereby broadcasting high-level signals important for perception and learning. Recent studies suggest dendritic Ca2+ spikes as key biophysical mechanisms supporting corticofugal neuron function: these long-lasting events drive burst firing, thereby initiating uniquely powerful signals to modulate sub-cortical representations and trigger learning-related plasticity. However, the behavioral relevance of corticofugal dendritic spikes is poorly understood. We shed light on this issue using 2-photon Ca2+ imaging of auditory corticofugal dendrites as mice of either sex engage in a GO/NO-GO sound-discrimination task. Unexpectedly, only a minority of dendritic spikes were triggered by behaviorally relevant sounds under our conditions. Task related dendritic activity instead mostly followed sound cue termination and co-occurred with mice's instrumental licking during the answer period of behavioral trials, irrespective of reward consumption. Temporally selective, optogenetic silencing of corticofugal neurons during the trial answer period impaired auditory discrimination learning. Thus, auditory corticofugal systems' contribution to learning and plasticity may be partially nonsensory in nature.

Short-term Hebbian learning can implement transformer-like attention

Transformers have revolutionized machine learning models of language and vision, but their connection with neuroscience remains tenuous. Built from attention layers, they require a mass comparison of queries and keys that is difficult to perform using traditional neural circuits. Here, we show that neurons can implement attention-like computations using short-term, Hebbian synaptic potentiation. We call our mechanism the match-and-control principle and it proposes that when activity in an axon is synchronous, or matched, with the somatic activity of a neuron that it synapses onto, the synapse can be briefly strongly potentiated, allowing the axon to take over, or control, the activity of the downstream neuron for a short time. In our scheme, the keys and queries are represented as spike trains and comparisons between the two are performed in individual spines allowing for hundreds of key comparisons per query and roughly as many keys and queries as there are neurons in the network.

Beyond spiking networks: The computational advantages of dendritic amplification and input segregation

The brain can efficiently learn a wide range of tasks, motivating the search for biologically inspired learning rules for improving current artificial intelligence technology. Most biological models are composed of point neurons and cannot achieve state-of-the-art performance in machine learning. Recent works have proposed that input segregation (neurons receive sensory information and higher-order feedback in segregated compartments), and nonlinear dendritic computation would support error backpropagation in biological neurons. However, these approaches require propagating errors with a fine spatiotemporal structure to all the neurons, which is unlikely to be feasible in a biological network. To relax this assumption, we suggest that bursts and dendritic input segregation provide a natural support for target-based learning, which propagates targets rather than errors. A coincidence mechanism between the basal and the apical compartments allows for generating high-frequency bursts of spikes. This architecture supports a burst-dependent learning rule, based on the comparison between the target bursting activity triggered by the teaching signal and the one caused by the recurrent connections, providing support for target-based learning. We show that this framework can be used to efficiently solve spatiotemporal tasks, such as context-dependent store and recall of three-dimensional trajectories, and navigation tasks. Finally, we suggest that this neuronal architecture naturally allows for orchestrating "hierarchical imitation learning", enabling the decomposition of challenging long-horizon decision-making tasks into simpler subtasks. We show a possible implementation of this in a two-level network, where the high network produces the contextual signal for the low network.

Dendrites and efficiency: Optimizing performance and resource utilization

The brain is a highly efficient system that has evolved to optimize performance under limited resources. In this review, we highlight recent theoretical and experimental studies that support the view that dendrites make information processing and storage in the brain more efficient. This is achieved through the dynamic modulation of integration versus segregation of inputs and activity within a neuron. We argue that under conditions of limited energy and space, dendrites help biological networks to implement complex functions such as processing natural stimuli on behavioral timescales, performing the inference process on those stimuli in a context-specific manner, and storing the information in overlapping populations of neurons. A global picture starts to emerge, in which dendrites help the brain achieve efficiency through a combination of optimization strategies that balance the tradeoff between performance and resource utilization.

Predictive Coding as a Neuromorphic Alternative to Backpropagation: A Critical Evaluation

Backpropagation has rapidly become the workhorse credit assignment algorithm for modern deep learning methods. Recently, modified forms of predictive coding (PC), an algorithm with origins in computational neuroscience, have been shown to result in approximately or exactly equal parameter updates to those under backpropagation. Due to this connection, it has been suggested that PC can act as an alternative to backpropagation with desirable properties that may facilitate implementation in neuromorphic systems. Here, we explore these claims using the different contemporary PC variants proposed in the literature. We obtain time complexity bounds for these PC variants, which we show are lower bounded by backpropagation. We also present key properties of these variants that have implications for neurobiological plausibility and their interpretations, particularly from the perspective of standard PC as a variational Bayes algorithm for latent probabilistic models. Our findings shed new light on the connection between the two learning frameworks and suggest that in its current forms, PC may have more limited potential as a direct replacement of backpropagation than previously envisioned.

The combination of Hebbian and predictive plasticity learns invariant object representations in deep sensory networks

Recognition of objects from sensory stimuli is essential for survival. To that end, sensory networks in the brain must form object representations invariant to stimulus changes, such as size, orientation and context. Although Hebbian plasticity is known to shape sensory networks, it fails to create invariant object representations in computational models, raising the question of how the brain achieves such processing. In the present study, we show that combining Hebbian plasticity with a predictive form of plasticity leads to invariant representations in deep neural network models. We derive a local learning rule that generalizes to spiking neural networks and naturally accounts for several experimentally observed properties of synaptic plasticity, including metaplasticity and spike-timing-dependent plasticity. Finally, our model accurately captures neuronal selectivity changes observed in the primate inferotemporal cortex in response to altered visual experience. Thus, we provide a plausible normative theory emphasizing the importance of predictive plasticity mechanisms for successful representational learning.

A GPU-based computational framework that bridges neuron simulation and artificial intelligence

Biophysically detailed multi-compartment models are powerful tools to explore computational principles of the brain and also serve as a theoretical framework to generate algorithms for artificial intelligence (AI) systems. However, the expensive computational cost severely limits the applications in both the neuroscience and AI fields. The major bottleneck during simulating detailed compartment models is the ability of a simulator to solve large systems of linear equations. Here, we present a novel Dendritic Hierarchical Scheduling (DHS) method to markedly accelerate such a process. We theoretically prove that the DHS implementation is computationally optimal and accurate. This GPU-based method performs with 2-3 orders of magnitude higher speed than that of the classic serial Hines method in the conventional CPU platform. We build a DeepDendrite framework, which integrates the DHS method and the GPU computing engine of the NEURON simulator and demonstrate applications of DeepDendrite in neuroscience tasks. We investigate how spatial patterns of spine inputs affect neuronal excitability in a detailed human pyramidal neuron model with 25,000 spines. Furthermore, we provide a brief discussion on the potential of DeepDendrite for AI, specifically highlighting its ability to enable the efficient training of biophysically detailed models in typical image classification tasks.

Artificial Neuronal Devices Based on Emerging Materials: Neuronal Dynamics and Applications

Artificial neuronal devices are critical building blocks of neuromorphic computing systems and currently the subject of intense research motivated by application needs from new computing technology and more realistic brain emulation. Researchers have proposed a range of device concepts that can mimic neuronal dynamics and functions. Although the switching physics and device structures of these artificial neurons are largely different, their behaviors can be described by several neuron models in a more unified manner. In this paper, the reports of artificial neuronal devices based on emerging volatile switching materials are reviewed from the perspective of the demonstrated neuron models, with a focus on the neuronal functions implemented in these devices and the exploitation of these functions for computational and sensing applications. Furthermore, the neuroscience inspirations and engineering methods to enrich the neuronal dynamics that remain to be implemented in artificial neuronal devices and networks toward realizing the full functionalities of biological neurons are discussed.

Physics-based character animation and human motor control

Motor neuroscience and physics-based character animation (PBCA) approach human and humanoid control from different perspectives. The primary goal of PBCA is to control the movement of a ragdoll (humanoid or animal) applying forces and torques within a physical simulation. The primary goal of motor neuroscience is to understand the contribution of different parts of the nervous system to generate coordinated movements. We review the functional principles and the functional anatomy of human motor control and the main strategies used in PBCA. We then explore common research points by discussing the functional anatomy and ongoing debates in motor neuroscience from the perspective of PBCA. We also suggest there are several benefits to be found in studying sensorimotor integration and human-character coordination through closer collaboration between these two fields.

How to incorporate biological insights into network models and why it matters

Due to the staggering complexity of the brain and its neural circuitry, neuroscientists rely on the analysis of mathematical models to elucidate its function. From Hodgkin and Huxley's detailed description of the action potential in 1952 to today, new theories and increasing computational power have opened up novel avenues to study how neural circuits implement the computations that underlie behaviour. Computational neuroscientists have developed many models of neural circuits that differ in complexity, biological realism or emergent network properties. With recent advances in experimental techniques for detailed anatomical reconstructions or large-scale activity recordings, rich biological data have become more available. The challenge when building network models is to reflect experimental results, either through a high level of detail or by finding an appropriate level of abstraction. Meanwhile, machine learning has facilitated the development of artificial neural networks, which are trained to perform specific tasks. While they have proven successful at achieving task-oriented behaviour, they are often abstract constructs that differ in many features from the physiology of brain circuits. Thus, it is unclear whether the mechanisms underlying computation in biological circuits can be investigated by analysing artificial networks that accomplish the same function but differ in their mechanisms. Here, we argue that building biologically realistic network models is crucial to establishing causal relationships between neurons, synapses, circuits and behaviour. More specifically, we advocate for network models that consider the connectivity structure and the recorded activity dynamics while evaluating task performance.

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.

Logarithmically scaled, gamma distributed neuronal spiking

Naturally log-scaled quantities abound in the nervous system. Distributions of these quantities have non-intuitive properties, which have implications for data analysis and the understanding of neural circuits. Here, we review the log-scaled statistics of neuronal spiking and the relevant analytical probability distributions. Recent work using log-scaling revealed that interspike intervals of forebrain neurons segregate into discrete modes reflecting spiking at different timescales and are each well-approximated by a gamma distribution. Each neuron spends most of the time in an irregular spiking 'ground state' with the longest intervals, which determines the mean firing rate of the neuron. Across the entire neuronal population, firing rates are log-scaled and well approximated by the gamma distribution, with a small number of highly active neurons and an overabundance of low rate neurons (the 'dark matter'). These results are intricately linked to a heterogeneous balanced operating regime, which confers upon neuronal circuits multiple computational advantages and has evolutionarily ancient origins.

Voltage dynamics of dendritic integration and back-propagation in vivo

Neurons integrate synaptic inputs within their dendrites and produce spiking outputs, which then propagate down the axon and back into the dendrites where they contribute to plasticity. Mapping the voltage dynamics in dendritic arbors of live animals is crucial for understanding neuronal computation and plasticity rules. Here we combine patterned channelrhodopsin activation with dual-plane structured illumination voltage imaging, for simultaneous perturbation and monitoring of dendritic and somatic voltage in Layer 2/3 pyramidal neurons in anesthetized and awake mice. We examined the integration of synaptic inputs and compared the dynamics of optogenetically evoked, spontaneous, and sensory-evoked back-propagating action potentials (bAPs). Our measurements revealed a broadly shared membrane voltage throughout the dendritic arbor, and few signatures of electrical compartmentalization among synaptic inputs. However, we observed spike rate acceleration-dependent propagation of bAPs into distal dendrites. We propose that this dendritic filtering of bAPs may play a critical role in activity-dependent plasticity.

Learning cortical hierarchies with temporal Hebbian updates

A key driver of mammalian intelligence is the ability to represent incoming sensory information across multiple abstraction levels. For example, in the visual ventral stream, incoming signals are first represented as low-level edge filters and then transformed into high-level object representations. Similar hierarchical structures routinely emerge in artificial neural networks (ANNs) trained for object recognition tasks, suggesting that similar structures may underlie biological neural networks. However, the classical ANN training algorithm, backpropagation, is considered biologically implausible, and thus alternative biologically plausible training methods have been developed such as Equilibrium Propagation, Deep Feedback Control, Supervised Predictive Coding, and Dendritic Error Backpropagation. Several of those models propose that local errors are calculated for each neuron by comparing apical and somatic activities. Notwithstanding, from a neuroscience perspective, it is not clear how a neuron could compare compartmental signals. Here, we propose a solution to this problem in that we let the apical feedback signal change the postsynaptic firing rate and combine this with a differential Hebbian update, a rate-based version of classical spiking time-dependent plasticity (STDP). We prove that weight updates of this form minimize two alternative loss functions that we prove to be equivalent to the error-based losses used in machine learning: the inference latency and the amount of top-down feedback necessary. Moreover, we show that the use of differential Hebbian updates works similarly well in other feedback-based deep learning frameworks such as Predictive Coding or Equilibrium Propagation. Finally, our work removes a key requirement of biologically plausible models for deep learning and proposes a learning mechanism that would explain how temporal Hebbian learning rules can implement supervised hierarchical learning.

Synaptic turnover promotes efficient learning in bio-realistic spiking neural networks

While artificial machine learning systems achieve superhuman performance in specific tasks such as language processing, image and video recognition, they do so use extremely large datasets and huge amounts of power. On the other hand, the brain remains superior in several cognitively challenging tasks while operating with the energy of a small lightbulb. We use a biologically constrained spiking neural network model to explore how the neural tissue achieves such high efficiency and assess its learning capacity on discrimination tasks. We found that synaptic turnover, a form of structural plasticity, which is the ability of the brain to form and eliminate synapses continuously, increases both the speed and the performance of our network on all tasks tested. Moreover, it allows accurate learning using a smaller number of examples. Importantly, these improvements are most significant under conditions of resource scarcity, such as when the number of trainable parameters is halved and when the task difficulty is increased. Our findings provide new insights into the mechanisms that underlie efficient learning in the brain and can inspire the development of more efficient and flexible machine learning algorithms.

Responses of pyramidal cell somata and apical dendrites in mouse visual cortex over multiple days

The apical dendrites of pyramidal neurons in sensory cortex receive primarily top-down signals from associative and motor regions, while cell bodies and nearby dendrites are heavily targeted by locally recurrent or bottom-up inputs from the sensory periphery. Based on these differences, a number of theories in computational neuroscience postulate a unique role for apical dendrites in learning. However, due to technical challenges in data collection, little data is available for comparing the responses of apical dendrites to cell bodies over multiple days. Here we present a dataset collected through the Allen Institute Mindscope's OpenScope program that addresses this need. This dataset comprises high-quality two-photon calcium imaging from the apical dendrites and the cell bodies of visual cortical pyramidal neurons, acquired over multiple days in awake, behaving mice that were presented with visual stimuli. Many of the cell bodies and dendrite segments were tracked over days, enabling analyses of how their responses change over time. This dataset allows neuroscientists to explore the differences between apical and somatic processing and plasticity.

Transcriptomic cell type structures in vivo neuronal activity across multiple timescales

Cell type is hypothesized to be a key determinant of a neuron's role within a circuit. Here, we examine whether a neuron's transcriptomic type influences the timing of its activity. We develop a deep-learning architecture that learns features of interevent intervals across timescales (ms to >30 min). We show that transcriptomic cell-class information is embedded in the timing of single neuron activity in the intact brain of behaving animals (calcium imaging and extracellular electrophysiology) as well as in a bio-realistic model of the visual cortex. Further, a subset of excitatory cell types are distinguishable but can be classified with higher accuracy when considering cortical layer and projection class. Finally, we show that computational fingerprints of cell types may be universalizable across structured stimuli and naturalistic movies. Our results indicate that transcriptomic class and type may be imprinted in the timing of single neuron activity across diverse stimuli.

Efficient shallow learning as an alternative to deep learning

The realization of complex classification tasks requires training of deep learning (DL) architectures consisting of tens or even hundreds of convolutional and fully connected hidden layers, which is far from the reality of the human brain. According to the DL rationale, the first convolutional layer reveals localized patterns in the input and large-scale patterns in the following layers, until it reliably characterizes a class of inputs. Here, we demonstrate that with a fixed ratio between the depths of the first and second convolutional layers, the error rates of the generalized shallow LeNet architecture, consisting of only five layers, decay as a power law with the number of filters in the first convolutional layer. The extrapolation of this power law indicates that the generalized LeNet can achieve small error rates that were previously obtained for the CIFAR-10 database using DL architectures. A power law with a similar exponent also characterizes the generalized VGG-16 architecture. However, this results in a significantly increased number of operations required to achieve a given error rate with respect to LeNet. This power law phenomenon governs various generalized LeNet and VGG-16 architectures, hinting at its universal behavior and suggesting a quantitative hierarchical time-space complexity among machine learning architectures. Additionally, the conservation law along the convolutional layers, which is the square-root of their size times their depth, is found to asymptotically minimize error rates. The efficient shallow learning that is demonstrated in this study calls for further quantitative examination using various databases and architectures and its accelerated implementation using future dedicated hardware developments.

Meta-learning biologically plausible plasticity rules with random feedback pathways

Backpropagation is widely used to train artificial neural networks, but its relationship to synaptic plasticity in the brain is unknown. Some biological models of backpropagation rely on feedback projections that are symmetric with feedforward connections, but experiments do not corroborate the existence of such symmetric backward connectivity. Random feedback alignment offers an alternative model in which errors are propagated backward through fixed, random backward connections. This approach successfully trains shallow models, but learns slowly and does not perform well with deeper models or online learning. In this study, we develop a meta-learning approach to discover interpretable, biologically plausible plasticity rules that improve online learning performance with fixed random feedback connections. The resulting plasticity rules show improved online training of deep models in the low data regime. Our results highlight the potential of meta-learning to discover effective, interpretable learning rules satisfying biological constraints.

Evolution of neural activity in circuits bridging sensory and abstract knowledge

The ability to associate sensory stimuli with abstract classes is critical for survival. How are these associations implemented in brain circuits? And what governs how neural activity evolves during abstract knowledge acquisition? To investigate these questions, we consider a circuit model that learns to map sensory input to abstract classes via gradient-descent synaptic plasticity. We focus on typical neuroscience tasks (simple, and context-dependent, categorization), and study how both synaptic connectivity and neural activity evolve during learning. To make contact with the current generation of experiments, we analyze activity via standard measures such as selectivity, correlations, and tuning symmetry. We find that the model is able to recapitulate experimental observations, including seemingly disparate ones. We determine how, in the model, the behaviour of these measures depends on details of the circuit and the task. These dependencies make experimentally testable predictions about the circuitry supporting abstract knowledge acquisition in the brain.

Discrimination and learning of temporal input sequences in a cerebellar Purkinje cell model

Introduction

Temporal information processing is essential for sequential contraction of various muscles with the appropriate timing and amplitude for fast and smooth motor control. These functions depend on dynamics of neural circuits, which consist of simple neurons that accumulate incoming spikes and emit other spikes. However, recent studies indicate that individual neurons can perform complex information processing through the nonlinear dynamics of dendrites with complex shapes and ion channels. Although we have extensive evidence that cerebellar circuits play a vital role in motor control, studies investigating the computational ability of single Purkinje cells are few.

Methods

We found, through computer simulations, that a Purkinje cell can discriminate a series of pulses in two directions (from dendrite tip to soma, and from soma to dendrite), as cortical pyramidal cells do. Such direction sensitivity was observed in whatever compartment types of dendrites (spiny, smooth, and main), although they have dierent sets of ion channels.

Results

We found that the shortest and longest discriminable sequences lasted for 60 ms (6 pulses with 10 ms interval) and 4,000 ms (20 pulses with 200 ms interval), respectively. and that the ratio of discriminable sequences within the region of the interesting parameter space was, on average, 3.3% (spiny), 3.2% (smooth), and 1.0% (main). For the direction sensitivity, a T-type Ca²⁺ channel was necessary, in contrast with cortical pyramidal cells that have N-methyl-D-aspartate receptors (NMDARs). Furthermore, we tested whether the stimulus direction can be reversed by learning, specifically by simulated long-term depression, and obtained positive results.

Discussion

Our results show that individual Purkinje cells can perform more complex information processing than is conventionally assumed for a single neuron, and suggest that Purkinje cells act as sequence discriminators, a useful role in motor control and learning.

Mental navigation and the neural mechanisms of insight

How do new ideas come about? The central hypothesis presented here states that insights might happen during mental navigation and correspond to rapid plasticity at the cellular level. We highlight the differences between neocortical and hippocampal mechanisms of insight. We argue that the suddenness of insight can be related to the sudden emergence of place fields in the hippocampus. According to our hypothesis, insights are supported by a state of mind-wandering that can be tied to the process of combining knowledge pieces during sharp-wave ripples (SWRs). Our framework connects the dots between research on creativity, mental navigation, and specific synaptic plasticity mechanisms in the hippocampus.

The plasticitome of cortical interneurons

Hebb postulated that, to store information in the brain, assemblies of excitatory neurons coding for a percept are bound together via associative long-term synaptic plasticity. In this view, it is unclear what role, if any, is carried out by inhibitory interneurons. Indeed, some have argued that inhibitory interneurons are not plastic. Yet numerous recent studies have demonstrated that, similar to excitatory neurons, inhibitory interneurons also undergo long-term plasticity. Here, we discuss the many diverse forms of long-term plasticity that are found at inputs to and outputs from several types of cortical inhibitory interneuron, including their plasticity of intrinsic excitability and their homeostatic plasticity. We explain key plasticity terminology, highlight key interneuron plasticity mechanisms, extract overarching principles and point out implications for healthy brain functionality as well as for neuropathology. We introduce the concept of the plasticitome - the synaptic plasticity counterpart to the genome or the connectome - as well as nomenclature and definitions for dealing with this rich diversity of plasticity. We argue that the great diversity of interneuron plasticity rules is best understood at the circuit level, for example as a way of elucidating how the credit-assignment problem is solved in deep biological neural networks.

Learning on tree architectures outperforms a convolutional feedforward network

Advanced deep learning architectures consist of tens of fully connected and convolutional hidden layers, currently extended to hundreds, are far from their biological realization. Their implausible biological dynamics relies on changing a weight in a non-local manner, as the number of routes between an output unit and a weight is typically large, using the backpropagation technique. Here, a 3-layer tree architecture inspired by experimental-based dendritic tree adaptations is developed and applied to the offline and online learning of the CIFAR-10 database. The proposed architecture outperforms the achievable success rates of the 5-layer convolutional LeNet. Moreover, the highly pruned tree backpropagation approach of the proposed architecture, where a single route connects an output unit and a weight, represents an efficient dendritic deep learning.

Introducing the Dendrify framework for incorporating dendrites to spiking neural networks

Computational modeling has been indispensable for understanding how subcellular neuronal features influence circuit processing. However, the role of dendritic computations in network-level operations remains largely unexplored. This is partly because existing tools do not allow the development of realistic and efficient network models that account for dendrites. Current spiking neural networks, although efficient, are usually quite simplistic, overlooking essential dendritic properties. Conversely, circuit models with morphologically detailed neuron models are computationally costly, thus impractical for large-network simulations. To bridge the gap between these two extremes and facilitate the adoption of dendritic features in spiking neural networks, we introduce Dendrify, an open-source Python package based on Brian 2. Dendrify, through simple commands, automatically generates reduced compartmental neuron models with simplified yet biologically relevant dendritic and synaptic integrative properties. Such models strike a good balance between flexibility, performance, and biological accuracy, allowing us to explore dendritic contributions to network-level functions while paving the way for developing more powerful neuromorphic systems.

Cerebro-cerebellar networks facilitate learning through feedback decoupling

Behavioural feedback is critical for learning in the cerebral cortex. However, such feedback is often not readily available. How the cerebral cortex learns efficiently despite the sparse nature of feedback remains unclear. Inspired by recent deep learning algorithms, we introduce a systems-level computational model of cerebro-cerebellar interactions. In this model a cerebral recurrent network receives feedback predictions from a cerebellar network, thereby decoupling learning in cerebral networks from future feedback. When trained in a simple sensorimotor task the model shows faster learning and reduced dysmetria-like behaviours, in line with the widely observed functional impact of the cerebellum. Next, we demonstrate that these results generalise to more complex motor and cognitive tasks. Finally, the model makes several experimentally testable predictions regarding cerebro-cerebellar task-specific representations over learning, task-specific benefits of cerebellar predictions and the differential impact of cerebellar and inferior olive lesions. Overall, our work offers a theoretical framework of cerebro-cerebellar networks as feedback decoupling machines.

Recent Advances at the Interface of Neuroscience and Artificial Neural Networks

Biological neural networks adapt and learn in diverse behavioral contexts. Artificial neural networks (ANNs) have exploited biological properties to solve complex problems. However, despite their effectiveness for specific tasks, ANNs are yet to realize the flexibility and adaptability of biological cognition. This review highlights recent advances in computational and experimental research to advance our understanding of biological and artificial intelligence. In particular, we discuss critical mechanisms from the cellular, systems, and cognitive neuroscience fields that have contributed to refining the architecture and training algorithms of ANNs. Additionally, we discuss how recent work used ANNs to understand complex neuronal correlates of cognition and to process high throughput behavioral data.