Topic: recurrent neural networks

The hippocampus is thought to build a cognitive map that supports navigation, memory, and planning, but what defines such a map and how it is used remain debated. In this talk, I will present computational models in which hippocampal-like representations emerge in recurrent neural networks trained to predict sequences of sensory observations. While spatially tuned units reliably arise, they are not sufficient to form a cognitive map. Instead, map-like representations emerge when recurrent dynamics support multi-step prediction, yielding a population-level encoding of environmental geometry. On

Continuous attractors offer a unique class of solutions for storing continuous-valued variables in recurrent system states for indefinitely long time intervals. Unfortunately, continuous attractors suffer from severe structural instability in general---they are destroyed by most infinitesimal changes of the dynamical law that defines them. This fragility limits their utility especially in biological systems as their recurrent dynamics are subject to constant perturbations. We observe that the bifurcations from continuous attractors in theoretical neuroscience models display various structurall

Animals and humans rapidly adapt their behavior to dynamic environmental changes, such as predator threats or fluctuating food resources, often without immediate rewards. Existing literature posits that animals rely on internal representations of the environment, termed “beliefs”, for their decision policy. However, previous work ties belief updates to external reward signals, which does not explain adaptation in scenarios where trial-and-error approaches are inefficient or potentially perilous. In this work, we propose that the brain utilize dynamic representations that continuously infer the

Neural oscillations are ubiquitously observed in many brain areas. One proposed functional role of these oscillations is that they serve as an internal clock, or ‘frame of reference’ relative to which information can be encoded. In line with this hypothesis, there have been many empirical observations of this phase code in the brain. What are the latent dynamics and circuits that support phase coding with neural oscillations? Here, we propose a new computational hypothesis which is derived from analyzing trained recurrent neural networks (RNNs). We train RNNs on a working memory task, while al

A widely-accepted tenet of neuroscience is that learning relies on synaptic plasticity. Because plasticity is thought to be governed by local, activity-dependent rules, it is a longstanding mystery how synapses change in concert so that behavioral output improves. This problem is aggravated by the recurrent and multilayered architecture of the cortex, which makes it hard to determine the effects of a given synaptic modification. Here we present a novel principle for learning rooted in optimal control theory, which yields local plasticity rules for any neural dynamics which reaches an equilibri

Recurrent neural networks, a ubiquitous construct in machine intelligence, have become a powerful hypothesis-generating tool in theoretical neuroscience. Such networks can be used to examine potential circuit mechanisms associated with a variety of cognitive functions. Here, we use recurrent networks to engage a new theoretical question: the potential role of astrocytes in neural computation. Astrocytes are highly abundant cells in the cortex that are capable of modulating many facets of neural dynamics, including excitability, synaptic efficacy and plasticity. However, despite this modulatory

The human brain quickly adapts to a rapidly changing environment, often faster than what is thought to be possible by way of synaptic plasticity. Real-time control adaptation can only be achieved through faster processes such as neuromodulation. A number of neuromodulation models based on gain regulation have been proposed recently. But most of them utilize neuron-wide regulation and can not accommodate the simultaneous impact of different types of neurotransmitters on different parts of the dendritic tree. Models of synaptic gating and connectivity-tuning demonstrate that using enough additio

To survive in the uncertain world, humans and animals must pick the right task to perform in each situation and flexibly switch between tasks when they infer that the environment has changed. Yet such flexibility comes at a cost. Task performance decreases when the relevance of the task is uncertain. Here we present multiple lines of evidence from monkey electrophysiology, human psychophysics, and artificial neural networks demonstrating a neuronal mechanism that inflicts a cost of flexibility. In a behavioral paradigm designed to measure and manipulate subjects’ belief about the relevance of

Learning in neural circuits manifests via slow changes in activity driving behaviour changes. Uncovering how the coordination of plasticity mechanisms found in the brain leads to such changes is a difficult problem, chiefly because we do not have access to the synaptic parameters underlying neural computation. Nevertheless, as neural population recording techniques improve and experiments track changes in neural dynamics over long durations, finer details about plastic changes could be inferred by observing and characterizing changes in neural activity. In this work, we present a step towards

In the first part of this talk I will present work showing that recurrent neural networks can replicate broad behavioral patterns associated with dynamic visual object recognition in humans. An analysis of these networks shows that different types of recurrence use different strategies to solve the object recognition problem. The similarities between artificial neural networks and the brain presents another opportunity, beyond using them just as models of biological processing. In the second part of this talk, I will discuss—and solicit feedback on—a proposed research plan for testing a wide r

Artificial neural networks can be useful for studying brain functions. In cognitive neuroscience, recurrent neural networks are often used to model cognitive functions. I will first offer my opinion on what is missing in the classical use of recurrent neural networks. Then I will discuss two lines of ongoing efforts in our group to move beyond the classical recurrent neural networks by studying multi-system neural networks (the talk will focus on two-system networks). These are networks that combine modules for several neural systems, such as vision, audition, prefrontal, hippocampal systems.

Classification of behavior from multi-regional neural data sequentially in time can help in the early detection of behavior, which may provide important ways to provide corrective neural stimulation before the onset of behavior. Recurrent Neural Networks (RNNs) are designed for time-series data: they take in sequential inputs and predict the class of the sequence using recurrent hidden states that are able to retain a memory of previous inputs. However, standard RNNs cannot guarantee the classification at earlier points of the sequence; they are traditionally designed to achieve correct classi

How does the motor cortex (MC) achieve generalizable and purposeful movements from the complex non-linear musculoskeletal system? This understanding can help us uncover how does the cortical activity relates to the movement. To elucidate this relationship, we build a model of the MC with realistic properties. To mimic the recurrent connections between cortical neurons, we use recurrent neural networks (RNNs) to model the MC. The MC model receives sensory feedback and delivers muscle excitations to anatomically accurate musculoskeletal models to produce experimentally observed movements. The se

Neural activity in response to sensory stimuli leads to a reorganization of neuronal ensemble dynamics, which can be captured by lower-dimensional latent dynamics. Recurrent Neural Networks (RNN) have been proven successful at extracting such latent representations of neural dynamics. Here, we applied an RNN to the data available in the Steinmetz Neuropixels dataset, containing electrophysiological and behavioral data from mice engaged in a visual go/no go contrast detection task. First, we chose as our input (‘seed’) region a subset of spike rates recorded from mouse primary visual cortex (VI

Experimental evidence suggests that the brain is able to encode memories in the phase of ongoing low-frequency oscillations. Recently, trainable recurrent neural networks (RNNs) have emerged as promising models of how neural circuits process information. However, short-term memory maintenance in these networks has mostly been implemented through static attractors. Thus, it remains unclear if and how RNNs can encode memories in the relative phase of oscillations. Here we show that such phase-dependent coding can arise through the training of RNNs on memory tasks. We characterize how the emerg

Flexible computation is a hallmark of intelligent behavior. Yet, little is known about how neural networks contextually reconfigure for different computations. Humans are able to perform a new task without extensive training, presumably through the composition of elementary processes that were previously learned. Cognitive scientists have long hypothesized the possibility of a compositional neural code, where complex neural computations are made up of constituent components; however, the neural substrate underlying this structure remains elusive in biological and artificial neural networks. He

Working memory involves the short-term maintenance of information and is critical in many tasks. The neural circuit dynamics underlying working memory remain poorly understood, with different aspects of prefrontal cortical (PFC) responses explained by different putative mechanisms. By mathematical analysis, numerical simulations, and using recordings from monkey PFC, we investigate a critical but hitherto ignored aspect of working memory dynamics: information loading. We find that, contrary to common assumptions, optimal information loading involves inputs that are largely orthogonal, rather t

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

Predictive coding represents a promising framework for understanding brain function, postulating that the brain continuously inhibits predictable sensory input, ensuring a preferential processing of surprising elements. A central aspect of this view on cortical computation is its hierarchical connectivity, involving recurrent message passing between excitatory bottom-up signals and inhibitory top-down feedback. Here we use computational modelling to demonstrate that such architectural hard-wiring is not necessary. Rather, predictive coding is shown to emerge as a consequence of energy efficien

Low-dimensional population dynamics can be observed in neural activity recorded from the prefrontal cortex (PFC) of subjects performing simple cognitive tasks. Many studies have shown that recurrent neural networks (RNNs) trained on the same tasks can reproduce qualitatively these state space trajectories, and have used them as models of how neuronal dynamics implement task computations. The PFC is also viewed as a conductor that organizes the communication between cortical areas and provides contextual information. It is then not clear what is its role in solving simple cognitive tasks. Do th

Learning in real life is often sequential: A learner first learns task A, then task B. If the tasks are related, the learner may adapt the previously learned representation instead of generating a new one from scratch. Adaptation may ease learning task B but may also decrease the performance on task A. Such interference has been observed in experimental and machine learning studies. In the latter case, it is mediated by correlations between weight updates for the different tasks. In typical applications, like image classification with feed-forward networks, these correlated weight updates can

Rajanlab designs neural network models constrained by experimental data, and reverse engineers them to figure out how brain circuits function in health and disease. Recently, we have been developing a powerful new theory-based framework for “in-vivo tract tracing” from multi-regional neural activity collected experimentally. We call this framework CURrent-Based Decomposition (CURBD). CURBD employs recurrent neural networks (RNNs) directly constrained, from the outset, by time series measurements acquired experimentally, such as Ca2+ imaging or electrophysiological data. Once trained, these da

Distinct synapses influence one another when they undergo changes, with unclear consequences for neuronal dynamics and function. Here we show that synapses can interact such that excitatory currents are naturally normalised and balanced by inhibitory inputs. This happens when classical spike-timing dependent synaptic plasticity rules are extended by additional mechanisms that incorporate the influence of neighbouring synaptic currents and regulate the amplitude of efficacy changes accordingly. The resulting control of excitatory plasticity by inhibitory activation, and vice versa, gives rise t

Rajanlab designs neural network models constrained by experimental data, and reverse engineers them to figure out how brain circuits function in health and disease. Recently, we have been developing a powerful new theory-based framework for “in-vivo tract tracing” from multi-regional neural activity collected experimentally. We call this framework CURrent-Based Decomposition (CURBD). CURBD employs recurrent neural networks (RNNs) directly constrained, from the outset, by time series measurements acquired experimentally, such as Ca2+ imaging or electrophysiological data. Once trained, these dat