This webinar features presentations from SueYeon Chung (New York University) and Srinivas Turaga (HHMI Janelia Research Campus) on theoretical and computational approaches to sensory cognition. Chung introduced a “neural manifold” framework to capture how high-dimensional neural activity is structured into meaningful manifolds reflecting object representations. She demonstrated that manifold geometry—shaped by radius, dimensionality, and correlations—directly governs a population’s capacity for classifying or separating stimuli under nuisance variations. Applying these ideas as a data analysis tool, she showed how measuring object-manifold geometry can explain transformations along the ventral visual stream and suggested that manifold principles also yield better self-supervised neural network models resembling mammalian visual cortex. Turaga described simulating the entire fruit fly visual pathway using its connectome, modeling 64 key cell types in the optic lobe. His team’s systematic approach—combining sparse connectivity from electron microscopy with simple dynamical parameters—recapitulated known motion-selective responses and produced novel testable predictions. Together, these studies underscore the power of combining connectomic detail, task objectives, and geometric theories to unravel neural computations bridging from stimuli to cognitive functions.

Modern electrophysiological recordings simultaneously capture single-unit spiking activities of hundreds of neurons. These neurons exhibit highly complex coordination patterns. Where does this complexity stem from? One candidate is the ubiquitous heterogeneity in connectivity of local neural circuits. Studying neural network dynamics in the linearized regime and using tools from statistical field theory of disordered systems, we derive relations between structure and dynamics that are readily applicable to subsampled recordings of neural circuits: Measuring the statistics of pairwise covariances allows us to infer statistical properties of the underlying connectivity. Applying our results to spontaneous activity of macaque motor cortex, we find that the underlying network operates in a strongly recurrent regime. In this regime, network connectivity is highly heterogeneous, as quantified by a large radius of bulk connectivity eigenvalues. Being close to the point of linear instability, this dynamical regime predicts a rich correlation structure, a large dynamical repertoire, long-range interaction patterns, relatively low dimensionality and a sensitive control of neuronal coordination. These predictions are verified in analyses of spontaneous activity of macaque motor cortex and mouse visual cortex. Finally, we show that even microscopic features of connectivity, such as connection motifs, systematically scale up to determine the global organization of activity in neural circuits.

Vector-space models are used frequently to compare similarity and dimensionality among entity concepts. What happens when we apply these models to relational concepts? What is the evidence that such models do apply to relational concepts? If we use such a model, then one implication is that maximizing surface feature variation should improve relational concept learning. For example, in STEM instruction, the effectiveness of teaching by analogy is often limited by students’ focus on superficial features of the source and target exemplars. However, in contrast to the prediction of the vector-space computational model, the strategy of progressive alignment (moving from perceptually similar to different targets) has been suggested to address this issue (Gentner & Hoyos, 2017), and human behavioral evidence has shown benefits from progressive alignment. Here I will present some preliminary data that supports the computational approach. Participants were explicitly instructed to match stimuli based on relations while perceptual similarity of stimuli varied parametrically. We found that lower perceptual similarity reduced accurate relational matching. This finding demonstrates that perceptual similarity may interfere with relational judgements, but also hints at why progressive alignment maybe effective. These are preliminary, exploratory data and I to hope receive feedback on the framework and to start a discussion in a group on the utility of vector-space models for relational concepts in general.

Predictive modeling and dimensionality reduction of functional neuroimaging data have provided rich information about the representations and functional architectures of the human brain. While these approaches have been effective in many cases, we will discuss how neglecting the internal dynamics of the brain (e.g., spontaneous activity, global dynamics, effective connectivity) and its underlying computational principles may hinder our progress in understanding and modeling brain functions. By reexamining evidence from our previous and ongoing work, we will propose new hypotheses and directions for research that consider both internal dynamics and the computational principles that may govern brain processes.

Neural computations are currently investigated using two separate approaches: sorting neurons into functional subpopulations or examining the low-dimensional dynamics of collective activity. Whether and how these two aspects interact to shape computations is currently unclear. Using a novel approach to extract computational mechanisms from networks trained on neuroscience tasks, here we show that the dimensionality of the dynamics and subpopulation structure play fundamentally com- plementary roles. Although various tasks can be implemented by increasing the dimensionality in networks with fully random population structure, flexible input–output mappings instead require a non-random population structure that can be described in terms of multiple subpopulations. Our analyses revealed that such a subpopulation structure enables flexible computations through a mechanism based on gain-controlled modulations that flexibly shape the collective dynamics. Our results lead to task-specific predictions for the structure of neural selectivity, for inactivation experiments and for the implication of different neurons in multi-tasking.

A central goal in neuroscience is to understand how orchestrated computations in the brain arise from the properties of single neurons and networks of such neurons. Answering this question requires theoretical advances that shine light into the ‘black box’ of representations in neural circuits. In this talk, we will demonstrate theoretical approaches that help describe how cognitive and behavioral task implementations emerge from the structure in neural populations and from biologically plausible neural networks. First, we will introduce an analytic theory that connects geometric structures that arise from neural responses (i.e., neural manifolds) to the neural population’s efficiency in implementing a task. In particular, this theory describes a perceptron’s capacity for linearly classifying object categories based on the underlying neural manifolds’ structural properties. Next, we will describe how such methods can, in fact, open the ‘black box’ of distributed neuronal circuits in a range of experimental neural datasets. In particular, our method overcomes the limitations of traditional dimensionality reduction techniques, as it operates directly on the high-dimensional representations, rather than relying on low-dimensionality assumptions for visualization. Furthermore, this method allows for simultaneous multi-level analysis, by measuring geometric properties in neural population data, and estimating the amount of task information embedded in the same population. These geometric frameworks are general and can be used across different brain areas and task modalities, as demonstrated in the work of ours and others, ranging from the visual cortex to parietal cortex to hippocampus, and from calcium imaging to electrophysiology to fMRI datasets. Finally, we will discuss our recent efforts to fully extend this multi-level description of neural populations, by (1) investigating how single neuron properties shape the representation geometry in early sensory areas, and by (2) understanding how task-efficient neural manifolds emerge in biologically-constrained neural networks. By extending our mathematical toolkit for analyzing representations underlying complex neuronal networks, we hope to contribute to the long-term challenge of understanding the neuronal basis of tasks and behaviors.

Biological brains possess an exceptional ability to infer relevant behavioral responses to a wide range of stimuli from only a few examples. This capacity to generalize beyond the training set has been proven particularly challenging to realize in artificial systems. How neural processes enable this capacity to extrapolate to novel stimuli is a fundamental open question. A prominent but underexplored hypothesis suggests that generalization is facilitated by a low-dimensional organization of collective neural activity, yet evidence for the underlying neural mechanisms remains wanting. Combining network modeling, theory and neural data analysis, we tested this hypothesis in the framework of flexible timing tasks, which rely on the interplay between inputs and recurrent dynamics. We first trained recurrent neural networks on a set of timing tasks while minimizing the dimensionality of neural activity by imposing low-rank constraints on the connectivity, and compared the performance and generalization capabilities with networks trained without any constraint. We then examined the trained networks, characterized the dynamical mechanisms underlying the computations, and verified their predictions in neural recordings. Our key finding is that low-dimensional dynamics strongly increases the ability to extrapolate to inputs outside of the range used in training. Critically, this capacity to generalize relies on controlling the low-dimensional dynamics by a parametric contextual input. We found that this parametric control of extrapolation was based on a mechanism where tonic inputs modulate the dynamics along non-linear manifolds in activity space while preserving their geometry. Comparisons with neural recordings in the dorsomedial frontal cortex of macaque monkeys performing flexible timing tasks confirmed the geometric and dynamical signatures of this mechanism. Altogether, our results tie together a number of previous experimental findings and suggest that the low-dimensional organization of neural dynamics plays a central role in generalizable behaviors.

Neurons in the brain communicate through action potentials (spikes) that are transmitted through chemical synapses. Throughout the last decades, the question how networks of spiking neurons represent and process information has remained an important challenge. Some progress has resulted from a recent family of supervised learning rules (tempotrons) for models of spiking neurons. However, these studies have viewed synaptic transmission as static and characterized synaptic efficacies as scalar quantities that change only on slow time scales of learning across trials but remain fixed on the fast time scales of information processing within a trial. By contrast, signal transduction at chemical synapses in the brain results from complex molecular interactions between multiple biochemical processes whose dynamics result in substantial short-term plasticity of most connections. Here we study the computational capabilities of spiking neurons whose synapses are dynamic and plastic, such that each individual synapse can learn its own dynamics. We derive tempotron learning rules for current-based leaky-integrate-and-fire neurons with different types of dynamic synapses. Introducing ordinal synapses whose efficacies depend only on the order of input spikes, we establish an upper capacity bound for spiking neurons with dynamic synapses. We compare this bound to independent synapses, static synapses and to the well established phenomenological Tsodyks-Markram model. We show that synaptic dynamics in principle allow the storage capacity of spiking neurons to scale with the number of input spikes and that this increase in capacity can be traded for greater robustness to input noise, such as spike time jitter. Our work highlights the feasibility of a novel computational paradigm for spiking neural circuits with plastic synaptic dynamics: Rather than being determined by the fixed number of afferents, the dimensionality of a neuron's decision space can be scaled flexibly through the number of input spikes emitted by its input layer.

Neural circuits exhibit complex activity patterns, both spontaneously and in response to external stimuli. Information encoding and learning in neural circuits depend on the ability of time-varying stimuli to control spontaneous network activity. In particular, variability arising from the sensitivity to initial conditions of recurrent cortical circuits can limit the information conveyed about the sensory input. Spiking and firing rate network models can exhibit such sensitivity to initial conditions that are reflected in their dynamic entropy rate and attractor dimensionality computed from their full Lyapunov spectrum. I will show how chaos in both spiking and rate networks depends on biophysical properties of neurons and the statistics of time-varying stimuli. In spiking networks, increasing the input rate or coupling strength aids in controlling the driven target circuit, which is reflected in both a reduced trial-to-trial variability and a decreased dynamic entropy rate. With sufficiently strong input, a transition towards complete network state control occurs. Surprisingly, this transition does not coincide with the transition from chaos to stability but occurs at even larger values of external input strength. Controllability of spiking activity is facilitated when neurons in the target circuit have a sharp spike onset, thus a high speed by which neurons launch into the action potential. I will also discuss chaos and controllability in firing-rate networks in the balanced state. For these, external control of recurrent dynamics strongly depends on correlations in the input. This phenomenon was studied with a non-stationary dynamic mean-field theory that determines how the activity statistics and the largest Lyapunov exponent depend on frequency and amplitude of the input, recurrent coupling strength, and network size. This shows that uncorrelated inputs facilitate learning in balanced networks. The results highlight the potential of Lyapunov spectrum analysis as a diagnostic for machine learning applications of recurrent networks. They are also relevant in light of recent advances in optogenetics that allow for time-dependent stimulation of a select population of neurons.

Neural systems are remarkably robust against various perturbations, a phenomenon that still requires a clear explanation. Here, we graphically illustrate how neural networks can become robust. We study spiking networks that generate low-dimensional representations, and we show that the neurons’ subthreshold voltages are confined to a convex region in a lower-dimensional voltage subspace, which we call a ‘bounding box.’ Any changes in network parameters (such as number of neurons, dimensionality of inputs, firing thresholds, synaptic weights, or transmission delays) can all be understood as deformations of this bounding box. Using these insights, we show that functionality is preserved as long as perturbations do not destroy the integrity of the bounding box. We suggest that the principles underlying robustness in these networks—low-dimensional representations, heterogeneity of tuning, and precise negative feedback—may be key to understanding the robustness of neural systems at the circuit level.

Cognitive control allows us to think and behave flexibly based on our context and goals. Most theories of cognitive control propose a control representation that enables the same input to produce different outputs contingent on contextual factors. In this talk, I will focus on an important property of the control representation's neural code: its representational dimensionality. Dimensionality of a neural representation balances a basic separability/generalizability trade-off in neural computation. This tradeoff has important implications for cognitive control. In this talk, I will present initial evidence from fMRI and EEG showing that task representations in the human brain leverage both ends of this tradeoff during flexible behavior.

Every thought is associated with a unique pattern of brain activity. Thus, in principle, it should be possible to use these activity patterns as "brain fingerprints" for different thoughts and to read out what a person is thinking based on their brain activity alone. Indeed, using machine learning considerable progress has been made in such "brainreading" in recent years. It is now possible to decode which image a person is viewing, which film sequence they are watching, which emotional state they are in or which intentions they hold in mind. This talk will provide an overview of the current state of the art in brain reading. It will also highlight the main challenges and limitations of this research field. For example, mathematical models are needed to cope with the high dimensionality of potential mental states. Furthermore, the ethical concerns raised by (often premature) commercial applications of brain reading will also be discussed.

Networks are widely used as mathematical models of complex systems across many scientific disciplines, and in particular within neuroscience. In this talk, we introduce two aspects of our collaborative research: (1) machine learning and networks, and (2) graph dimensionality. Machine learning and networks. Decades of work have produced a vast corpus of research characterising the topological, combinatorial, statistical and spectral properties of graphs. Each graph property can be thought of as a feature that captures important (and sometimes overlapping) characteristics of a network. We have developed hcga, a framework for highly comparative analysis of graph data sets that computes several thousands of graph features from any given network. Taking inspiration from hctsa, hcga offers a suite of statistical learning and data analysis tools for automated identification and selection of important and interpretable features underpinning the characterisation of graph data sets. We show that hcga outperforms other methodologies (including deep learning) on supervised classification tasks on benchmark data sets whilst retaining the interpretability of network features, which we exemplify on a dataset of neuronal morphologies images. Graph dimensionality. Dimension is a fundamental property of objects and the space in which they are embedded. Yet ideal notions of dimension, as in Euclidean spaces, do not always translate to physical spaces, which can be constrained by boundaries and distorted by inhomogeneities, or to intrinsically discrete systems such as networks. Deviating from approaches based on fractals, here, we present a new framework to define intrinsic notions of dimension on networks, the relative, local and global dimension. We showcase our method on various physical systems.

A central issue in reinforcement learning (RL) is the ‘curse-of-dimensionality’, arising when the degrees-of-freedom are much larger than the number of training samples. In such circumstances, the learning process becomes too slow to be plausible. In the brain, higher cognitive functions (such as abstraction or metacognition) may be part of the solution by generating low dimensional representations on which RL can operate. In this talk I will discuss a series of studies in which we used functional magnetic resonance imaging (fMRI) and computational modeling to investigate the neuro-computational basis of efficient RL. We found that people can learn remarkably complex task structures non-consciously, but also that - intriguingly - metacognition appears tightly coupled to this learning ability. Furthermore, when people use an explicit (conscious) policy to select relevant information, learning is accelerated by abstractions. At the neural level, prefrontal cortex subregions are differentially involved in separate aspects of learning: dorsolateral prefrontal cortex pairs with metacognitive processes, while ventromedial prefrontal cortex with valuation and abstraction. I will discuss the implications of these findings, in particular new questions on the function of metacognition in adaptive behavior and the link with abstraction.

Technological advances have increased the availability of recordings from large populations of neurons across multiple brain areas. Coupling these recordings with dimensionality reduction techniques, recent work has led to new proposals for how populations of neurons can send and receive signals selectively and flexibly. Advancement of these proposals depends, however, on untangling the bidirectional, parallel communication between neuronal populations. Because our current data analytic tools struggle to achieve this task, we have recently validated and presented a novel dimensionality reduction framework: DLAG, or Delayed Latents Across Groups. DLAG decomposes the time-varying activity in each area into within- and across-area latent variables. Across-area variables can be decomposed further into feedforward and feedback components using automatically estimated time delays. In this talk, I will review the DLAG framework. Then I will discuss new insights into the moment-by-moment nature of feedforward and feedback communication between visual cortical areas V1 and V2 of macaque monkeys. Overall, this work lays the foundation for dissecting the dynamic flow of signals across populations of neurons, and how it might change across brain areas and behavioral contexts.

The link between behavior, learning and the underlying connectome is a fundamental open problem in neuroscience. In my talk I will show how it is possible to develop a theory that bridges across these three levels (animal behavior, learning and network connectivity) based on the geometrical properties of neural activity. The central tool in my approach is the dimensionality of neural activity. I will link animal complex behavior to the geometry of neural representations, specifically their dimensionality; I will then show how learning shapes changes in such geometrical properties and how local connectivity properties can further regulate them. As a result, I will explain how the complexity of neural representations emerges from both behavioral demands (top-down approach) and learning or connectivity features (bottom-up approach). I will build these results regarding neural dynamics and representations starting from the analysis of neural recordings, by means of theoretical and computational tools that blend dynamical systems, artificial intelligence and statistical physics approaches.

The remarkable properties of information-processing of biological and of artificial neuronal networks alike arise from the interaction of large numbers of neurons. A central quest is thus to characterize their collective states. The directed coupling between pairs of neurons and their continuous dissipation of energy, moreover, cause dynamics of neuronal networks outside thermodynamic equilibrium. Tools from non-equilibrium statistical mechanics and field theory are thus instrumental to obtain a quantitative understanding. We here present progress with this recent approach [1]. On the experimental side, we show how correlations between pairs of neurons are informative on the dynamics of cortical networks: they are poised near a transition to chaos [2]. Close to this transition, we find prolongued sequential memory for past signals [3]. In the chaotic regime, networks offer representations of information whose dimensionality expands with time. We show how this mechanism aids classification performance [4]. Together these works illustrate the fruitful interplay between theoretical physics, neuronal networks, and neural information processing.

Humans and other primates exhibit rich and versatile behaviour, switching nimbly between tasks as the environmental context requires. I will discuss the neural coding patterns that make this possible in humans and deep networks. First, using deep network simulations, I will characterise two distinct solutions to task acquisition (“lazy” and “rich” learning) which trade off learning speed for robustness, and depend on the initial weights scale and network sparsity. I will chart the predictions of these two schemes for a context-dependent decision-making task, showing that the rich solution is to project task representations onto orthogonal planes on a low-dimensional embedding space. Using behavioural testing and functional neuroimaging in humans, we observe BOLD signals in human prefrontal cortex whose dimensionality and neural geometry are consistent with the rich learning regime. Next, I will discuss the problem of continual learning, showing that behaviourally, humans (unlike vanilla neural networks) learn more effectively when conditions are blocked than interleaved. I will show how this counterintuitive pattern of behaviour can be recreated in neural networks by assuming that information is normalised and temporally clustered (via Hebbian learning) alongside supervised training. Together, this work offers a picture of how humans learn to partition knowledge in the service of structured behaviour, and offers a roadmap for building neural networks that adopt similar principles in the service of multitask learning. This is work with Andrew Saxe, Timo Flesch, David Nagy, and others.

Cortical circuits receive multiple inputs from upstream populations with non-overlapping stimulus tuning preferences. Both the feedforward and recurrent architectures of the receiving cortical layer will reflect this diverse input tuning. We study how population-wide neuronal variability propagates through a hierarchical cortical network receiving multiple, independent, tuned inputs. We present new analysis of in vivo neural data from the primate visual system showing that the number of latent variables (dimension) needed to describe population shared variability is smaller in V4 populations compared to those of its downstream visual area PFC. We successfully reproduce this dimensionality expansion from our V4 to PFC neural data using a multi-layer spiking network with structured, feedforward projections and recurrent assemblies of multiple, tuned neuron populations. We show that tuning-structured connectivity generates attractor dynamics within the recurrent PFC current, where attractor competition is reflected in the high dimensional shared variability across the population. Indeed, restricting the dimensionality analysis to activity from one attractor state recovers the low-dimensional structure inherited from each of our tuned inputs. Our model thus introduces a framework where high-dimensional cortical variability is understood as ``time-sharing’’ between distinct low-dimensional, tuning-specific circuit dynamics.

Analogy-making is considered to be one of the cognitive processes which are hard to be accounted for in connectionist terms. A number of models have been proposed, but they are either tailed for specific analogical tasks or require complicated mechanisms which don’t fit into the mainstream connectionist modelling paradigm. In this talk I will present a new connectionist account of analogy-making based on the vector approach to representing symbols (VARS). This approach allows representing relational structures of varying complexity by numeric vectors with fixed dimensionality. I will also present a simple and computationally efficient mechanism of aligning VARS representations, which integrates both semantic similarity and structural constraints. The results of a series of simulations will demonstrate that VARS can account for basic analogical phenomena.

The curse of dimensionality plagues models of reinforcement learning and decision-making. The process of abstraction solves this by constructing abstract variables describing features shared by different specific instances, reducing dimensionality and enabling generalization in novel situations. Here we characterized neural representations in monkeys performing a task where a hidden variable described the temporal statistics of stimulus-response-outcome mappings. Abstraction was defined operationally using the generalization performance of neural decoders across task conditions not used for training. This type of generalization requires a particular geometric format of neural representations. Neural ensembles in dorsolateral pre-frontal cortex, anterior cingulate cortex and hippocampus, and in simulated neural networks, simultaneously represented multiple hidden and explicit variables in a format reflecting abstraction. Task events engaging cognitive operations modulated this format. These findings elucidate how the brain and artificial systems represent abstract variables, variables critical for generalization that in turn confers cognitive flexibility.

Many networks in the brain are sparsely connected, and the brain eliminates synapses during development and learning. How could the brain decide which synapses to prune? In a recurrent network, determining the importance of a synapse between two neurons is a difficult computational problem, depending on the role that both neurons play and on all possible pathways of information flow between them. Noise is ubiquitous in neural systems, and often considered an irritant to be overcome. In the first part of this talk, I will suggest that noise could play a functional role in synaptic pruning, allowing the brain to probe network structure and determine which synapses are redundant. I will introduce a simple, local, unsupervised plasticity rule that either strengthens or prunes synapses using only synaptic weight and the noise-driven covariance of the neighboring neurons. For a subset of linear and rectified-linear networks, this rule provably preserves the spectrum of the original matrix and hence preserves network dynamics even when the fraction of pruned synapses asymptotically approaches 1. The plasticity rule is biologically-plausible and may suggest a new role for noise in neural computation. Time permitting, I will then turn to the problem of extracting structure from neural population data sets using dimensionality reduction methods. I will argue that nonlinear structures naturally arise in neural data and show how these nonlinearities cause linear methods of dimensionality reduction, such as Principal Components Analysis, to fail dramatically in identifying low-dimensional structure.

The description of neural computations currently relies on two competing views: (i) a classical single-cell view that aims to relate the activity of individual neurons to sensory or behavioural variables, and organize them into functional classes; (ii) a more recent population view that instead characterises computations in terms of collective neural trajectories, and focuses on the dimensionality of these trajectories as animals perform tasks. How the two key concepts of functional cell classes and low-dimensional trajectories interact to shape neural computations is however at present not understood. Here I will address this question by combining machine-learning tools for training recurrent neural networks with reverse-engineering and theoretical analyses of network dynamics.

The curse of dimensionality plagues models of reinforcement learning and decision-making. The process of abstraction solves this by constructing abstract variables describing features shared by different specific instances, reducing dimensionality and enabling generalization in novel situations. We characterized neural representations in monkeys performing a task where a hidden variable described the temporal statistics of stimulus-response-outcome mappings. Abstraction was defined operationally using the generalization performance of neural decoders across task conditions not used for training. This type of generalization requires a particular geometric format of neural representations. Neural ensembles in dorsolateral pre-frontal cortex, anterior cingulate cortex and hippocampus, and in simulated neural networks, simultaneously represented multiple hidden and explicit variables in a format reflecting abstraction. Task events engaging cognitive operations modulated this format. These findings elucidate how the brain and artificial systems represent abstract variables, variables critical for generalization that in turn confers cognitive flexibility.

We will organize the workshop around one question: “Now that we can track (most) everything, what can we do with the data?” Given the recent dramatic advances in technology, we now have behavioral data sets with orders of magnitude more accuracy, dimensionality, diversity, and size than we had even a few years ago. That being said, there is still little agreement as to what theoretical frameworks can inform our understanding of these data sets and suggest new experiments we can perform. We hope that after this workshop we’ll see a variety of new ideas and perhaps gain some inspiration. We have invited eight speakers, each studying different systems, scales, and topics, to provide 10 minute presentations focused on the above question, with another 10 minutes set aside for questions/discussions (moderated by the two of us). Although we naturally expect speakers to include aspects of their own work, we have encouraged all of them to think broadly and provocatively. We are also hoping to organize some breakout sessions after the talks so that we can have some more expanded discussions about topics arising during the meeting.