Computational Neuroscience seminars

The brain has a hugely diverse and heterogeneous nature. The exact role of heterogeneity has been relatively little explored as most neural models tend to be largely homogeneous. We trained spiking neural networks with varying degrees of heterogeneity on complex real-world tasks and found that heterogeneity resulted in more stable and robust training and improved training performance, especially for tasks with a higher temporal structure. Moreover, the optimal distribution of parameters found by training was found to be similar to experimental observations. These findings suggest that heterogeneity is not simply a result of noisy biological processes, but it may play a crucial role for learning in complex, changing environments.

How can neural networks learn to efficiently represent complex and high-dimensional inputs via local plasticity mechanisms? Classical models of representation learning assume that input weights are learned via pairwise Hebbian-like plasticity. Here, we show that pairwise Hebbian-like plasticity only works under specific requirements on neural dynamics and input statistics. To overcome these limitations, we derive from first principles a learning scheme based on voltage-dependent synaptic plasticity rules. Here, inhibition learns to locally balance excitatory input in individual dendritic compartments, and thereby can modulate excitatory synaptic plasticity to learn efficient representations. We demonstrate in simulations that this learning scheme works robustly even for complex, high-dimensional and correlated inputs. It also works in the presence of inhibitory transmission delays, where Hebbian-like plasticity typically fails. Our results draw a direct connection between dendritic excitatory-inhibitory balance and voltage-dependent synaptic plasticity as observed in vivo, and suggest that both are crucial for representation learning.

Cortical circuits process information by rich recurrent interactions between excitatory neurons and inhibitory interneurons. One of the prime functions of interneurons is to stabilize the circuit by feedback inhibition, but the level of specificity on which inhibitory feedback operates is not fully resolved. We hypothesized that inhibitory circuits could enable separate feedback control loops for different synaptic input streams, by means of specific feedback inhibition to different neuronal compartments. To investigate this hypothesis, we adopted an optimization approach. Leveraging recent advances in training spiking network models, we optimized the connectivity and short-term plasticity of interneuron circuits for compartment-specific feedback inhibition onto pyramidal neurons. Over the course of the optimization, the interneurons diversified into two classes that resembled parvalbumin (PV) and somatostatin (SST) expressing interneurons. The resulting circuit can be understood as a neural decoder that inverts the nonlinear biophysical computations performed within the pyramidal cells. Our model provides a proof of concept for studying structure-function relations in cortical circuits by a combination of gradient-based optimization and biologically plausible phenomenological models

One of Tali's cherished goals was to transform biology into physics. In his view, biologists were far too enamored by the details of the specific models they studied, losing sight of the big principles that may govern the behavior of these models. One such big principle that he suggested was the 'information bottleneck (IB) principle'. The iIB principle is an information-theoretical approach for extracting the relevant information that one random variable carries about another. Tali applied the IB principle to numerous problems in biology, gaining important insights in the process. Here I will describe two applications of the IB principle to neurobiological data. The first is the formalization of the notion of surprise that allowed us to rigorously estimate the memory duration and content of neuronal responses in auditory cortex, and the second is an application to behavior, allowing us to estimate 'optimal policies under information constraints' that shed interesting light on rat behavior.

The basal ganglia are a collection of brain areas that are connected by a variety of synaptic pathways and are a site of significant reward-related dopamine release. These properties suggest a possible role for the basal ganglia in action selection, guided by reinforcement learning. In this talk, I will discuss a framework for how this function might be performed. I will also present some recent experimental results and theory that call for a re-evaluation of certain aspects of this framework. Next, I will turn to the changes in basal ganglia activity observed to occur with the dopamine depletion associated with Parkinson’s disease. I will discuss some of the potential functional implications of some of these changes and, if time permits, will conclude with some new results that focus on delta oscillations under dopamine depletion.

Hydra is an extraordinary creature. Continuously replacing itself, it can live indefinitely, performing a stable repertoire of reasonably sophisticated behaviors. This remarkable stability under plasticity may be due to the uniform nature of its nervous system, which consists of two apparently noncommunicating nerve net layers. We use modeling to understand the role of active muscles and biomechanics interact with neural activity to shape Hydra behaviour. We will discuss our findings and thoughts on how this simple nervous system may self-organize to produce purposeful behavior.

Homeostatic plasticity plays a key role in stabilizing neural network activity. But what is its role in neural information processing? We showed analytically how homeostasis changes collective dynamics and consequently information flow - depending on the input to the network. We then studied how input and homeostasis on a recurrent network of LIF neurons impacts information flow and task performance. We showed how we can tune the working point of the network, and found that, contrary to previous assumptions, there is not one optimal working point for a family of tasks, but each task may require its own working point.

Choosing among spatially distributed options is a central challenge for animals, from deciding among alternative potential food sources or refuges, to choosing with whom to associate. Here, using an integrated theoretical and experimental approach (employing immersive Virtual Reality), with both invertebrate and vertebrate models—the fruit fly, desert locust and zebrafish—we consider the recursive interplay between movement and collective vectorial integration in the brain during decision-making regarding options (potential ‘targets’) in space. We reveal that the brain repeatedly breaks multi-choice decisions into a series of abrupt (critical) binary decisions in space-time where organisms switch, spontaneously, from averaging vectorial information among, to suddenly excluding one of, the remaining options. This bifurcation process repeats until only one option—the one ultimately selected—remains. Close to each bifurcation the ‘susceptibility’ of the system exhibits a sharp increase, inevitably causing small differences among the remaining options to become amplified; a property that both comes ‘for free’ and is highly desirable for decision-making. This mechanism facilitates highly effective decision-making, and is shown to be robust both to the number of options available, and to context, such as whether options are static (e.g. refuges) or mobile (e.g. other animals). In addition, we find evidence that the same geometric principles of decision-making occur across scales of biological organisation, from neural dynamics to animal collectives, suggesting they are fundamental features of spatiotemporal computation.

Perception of object motion is fundamentally multisensory, yet little is known about similarities and differences in the computations that give rise to our experience across senses. Insight can be provided by examining auditory motion processing in early blind individuals. In those who become blind early in life, the ‘visual’ motion area hMT+ responds to auditory motion. Meanwhile, the planum temporale, associated with auditory motion in sighted individuals, shows reduced selectivity for auditory motion, suggesting competition between cortical areas for functional role. According to the metamodal hypothesis of cross-modal plasticity developed by Pascual-Leone, the recruitment of hMT+ is driven by it being a metamodal structure containing “operators that execute a given function or computation regardless of sensory input modality”. Thus, the metamodal hypothesis predicts that the computations underlying auditory motion processing in early blind individuals should be analogous to visual motion processing in sighted individuals - relying on non-separable spatiotemporal filters. Inconsistent with the metamodal hypothesis, evidence suggests that the computational algorithms underlying auditory motion processing in early blind individuals fail to undergo a qualitative shift as a result of cross-modal plasticity. Auditory motion filters, in both blind and sighted subjects, are separable in space and time, suggesting that the recruitment of hMT+ to extract motion information from auditory input includes a significant modification of its normal computational operations.

In nature, sensory inputs are often highly structured, and statistical regularities of these signals can be extracted to form expectation about future sensorimotor associations, thereby optimizing behavior. One of the fundamental questions in neuroscience concerns the neural computations that underlie these probabilistic sensorimotor processing. Through a recurrent neural network (RNN) model and human psychophysics and electroencephalography (EEG), the present study investigates circuit mechanisms for processing probabilistic structures of sensory signals to guide behavior. We first constructed and trained a biophysically constrained RNN model to perform a series of probabilistic decision-making tasks similar to paradigms designed for humans. Specifically, the training environment was probabilistic such that one stimulus was more probable than the others. We show that both humans and the RNN model successfully extract information about stimulus probability and integrate this knowledge into their decisions and task strategy in a new environment. Specifically, performance of both humans and the RNN model varied with the degree to which the stimulus probability of the new environment matched the formed expectation. In both cases, this expectation effect was more prominent when the strength of sensory evidence was low, suggesting that like humans, our RNNs placed more emphasis on prior expectation (top-down signals) when the available sensory information (bottom-up signals) was limited, thereby optimizing task performance. Finally, by dissecting the trained RNN model, we demonstrate how competitive inhibition and recurrent excitation form the basis for neural circuitry optimized to perform probabilistic information processing.

Spike-driven adaptation involves intracellular mechanisms that are initiated by spiking and lead to the subsequent reduction of spiking rate. One of its consequences is the temporal patterning of spike trains, as it imparts serial correlations between interspike intervals in baseline activity. Surprisingly the hidden adaptation states that lead to these correlations themselves exhibit quasi-independence. This talk will first discuss recent findings about the role of such adaptation in suppressing noise and extending sensory detection to weak stimuli that leave the firing rate unchanged. Further, a matching of the post-synaptic responses to the pre-synaptic adaptation time scale enables a recovery of the quasi-independence property, and can explain observations of correlations between post-synaptic EPSPs and behavioural detection thresholds. We then consider the involvement of spike-driven adaptation in the representation of intervals between sensory events. We discuss the possible link of this time-stamping mechanism to the conversion of egocentric to allocentric coordinates. The heterogeneity of the population parameters enables the representation and Bayesian decoding of time sequences of events which may be put to good use in path integration and hilus neuron function in hippocampus.

Where in the brain consciousness resides remains unclear. It has been suggested that the subnetworks supporting consciousness should be bidirectionally (recurrently) connected because both feed-forward and feedback processing are necessary for conscious experience. Accordingly, evaluating which subnetworks are bidirectionally connected and the strength of these connections would likely aid the identification of regions essential to consciousness. Here, we propose a method for hierarchically decomposing a network into cores with different strengths of bidirectional connection, as a means of revealing the structure of the complex brain network. We applied the method to a whole-brain mouse connectome. We found that cores with strong bidirectional connections consisted of regions presumably essential to consciousness (e.g., the isocortical and thalamic regions, and claustrum) and did not include regions presumably irrelevant to consciousness (e.g., cerebellum). Contrarily, we could not find such correspondence between cores and consciousness when we applied other simple methods which ignored bidirectionality. These findings suggest that our method provides a novel insight into the relation between bidirectional brain network structures and consciousness. Our recent preprint on this work is here: https://doi.org/10.1101/2021.07.12.452022.

The success of deep learning sparked interest in whether the brain learns by using similar techniques for assigning credit to each synaptic weight for its contribution to the network output. However, the majority of current attempts at biologically-plausible learning methods are either non-local in time, require highly specific connectivity motives, or have no clear link to any known mathematical optimization method. Here, we introduce Deep Feedback Control (DFC), a new learning method that uses a feedback controller to drive a deep neural network to match a desired output target and whose control signal can be used for credit assignment. The resulting learning rule is fully local in space and time and approximates Gauss-Newton optimization for a wide range of feedback connectivity patterns. To further underline its biological plausibility, we relate DFC to a multi-compartment model of cortical pyramidal neurons with a local voltage-dependent synaptic plasticity rule, consistent with recent theories of dendritic processing. By combining dynamical system theory with mathematical optimization theory, we provide a strong theoretical foundation for DFC that we corroborate with detailed results on toy experiments and standard computer-vision benchmarks.

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.

Predictive learning hypotheses posit that the neocortex learns a hierarchical model of the structure of features in the environment. Under these hypotheses, expected or predictable features are differentiated from unexpected ones by comparing bottom-up and top-down streams of data, with unexpected features then driving changes in the representation of incoming stimuli. This is supported by numerous studies in early sensory cortices showing that pyramidal neurons respond particularly strongly to unexpected stimulus events. However, it remains unknown how their responses govern subsequent changes in stimulus representations, and thus, govern learning. Here, I present results from our study of layer 2/3 and layer 5 pyramidal neurons imaged in primary visual cortex of awake, behaving mice using two-photon calcium microscopy at both the somatic and distal apical planes. Our data reveals that individual neurons and distal apical dendrites show distinct, but predictable changes in unexpected event responses when tracked over several days. Considering existing evidence that bottom-up information is primarily targeted to somata, with distal apical dendrites receiving the bulk of top-down inputs, our findings corroborate hypothesized complementary roles for these two neuronal compartments in hierarchical computing. Altogether, our work provides novel evidence that the neocortex indeed instantiates a predictive hierarchical model in which unexpected events drive learning.

Neurological disorders share common high-level alterations, such as cognitive deficits, anxiety, and depression. This raises the possibility of fundamental alterations in the way information conveyed by neural firing is maintained and dispatched in the diseased brain. Using experimental epilepsy as a model of neurological disorder we tested the hypothesis of altered information processing, analyzing how neurons in the hippocampus and the entorhinal cortex store and exchange information during slow and theta oscillations. We equate the storage and sharing of information to low level, or primitive, information processing at the algorithmic level, the theoretical intermediate level between structure and function. We find that these low-level processes are organized into substates during brain states marked by theta and slow oscillations. Their internal composition and organization through time are disrupted in epilepsy, losing brain state-specificity, and shifting towards a regime of disorder in a brain region dependent manner. We propose that the alteration of information processing at an algorithmic level may be a mechanism behind the emergent and widespread co-morbidities associated with epilepsy, and perhaps other disorders.

Given the complex and dynamic nature of our social relationships, the human brain needs to quickly learn and adapt to new social situations. The breakdown of any of these computations could lead to social deficits, as observed in many psychiatric disorders. In this talk, I will present our recent neurocomputational and intracranial work that attempts to model both 1) how humans dynamically adapt beliefs about other people and 2) how individuals can exert influence over social others through model-based forward thinking. Lastly, I will present our findings of how impaired social computations might manifest in different disorders such as addiction, delusion, and autism. Taken together, these findings reveal the dynamic and proactive nature of human interactions as well as the clinical significance of these high-order social processes.

Large-scale, single neuron resolution recordings are inherently high-dimensional, with as many dimensions as neurons. To make sense of them, for many the answer is: reduce the number of dimensions. In this talk I argue we can distinguish weak and strong principles of neural dimension reduction. The weak principle is that dimension reduction is a convenient tool for making sense of complex neural data. The strong principle is that dimension reduction moves us closer to how neural circuits actually operate and compute. Elucidating these principles is crucial, for which we subscribe to provides radically different interpretations of the same dimension reduction techniques applied to the same data. I outline experimental evidence for each principle, but illustrate how we could make either the weak or strong principles appear to be true based on innocuous looking analysis decisions. These insights suggest arguments over low and high-dimensional neural activity need better constraints from both experiment and theory.