Neural computation

Conduct computational neuroscience research in the Laboratory for Neural Computation and Adaptation. The team seeks researchers with strong analytical skills who can connect theoretical models with experiments.

…synthetic psychology and theories of neural computation.

Neural oscillations are often proposed to support brain computation by routing information, organizing cell assemblies, or shaping coding dynamics. Yet these ideas usually assume rhythms that are strong, sustained, and regular, whereas in vivo oscillations are often weak, transient, noisy, and variable in frequency and phase. In this talk, I will argue that such “no-metronome” oscillations are not just noisy fluctuations, but coordinated complex dynamics with functional consequences. Combining analyses of neural activity recordings during actual behavior (mice and non-human-primate LFPs and hu

…dynamics used to describe neural computation.

Predictive processing offers a unifying view of neural computation, proposing that brains continuously anticipate sensory input and update internal models based on prediction errors. In this talk, I will present converging evidence for the computational mechanisms underlying this framework across human neuroscience and deep neural networks. I will begin with recent work showing that large-scale distributed prediction-error encoding in the human brain directly predicts how sensory representations reorganize through predictive learning. I will then turn to PredNet, a popular predictive coding in

Recent advances in recording techniques and behavioral monitoring are revolutionizing systems neuroscience, enabling the study of behavior in naturalistic settings. Using fiber photometry data from dopamine axons (DA) in the striatum, we set out to characterize the homogeneity and heterogeneity across striatal areas during naturalistic behavior. These recordings were performed while a mouse explored multiple items placed inside a box, with which they could interact freely. Specifically, our goal was to test the hypothesis that DA axons in the ventral striatum (VS) solely encode value informati

We can now measure the connectivity of every neuron in a neural circuit, but we are still blind to other biological details, including the dynamical characteristics of each neuron. The degree to which connectivity measurements alone can inform understanding of neural computation is an open question. Here we show that with experimental measurements of only the connectivity of a biological neural network, we can predict the neural activity underlying a specified neural computation. We constructed a model neural network with the experimentally determined connectivity for 64 cell types in the moti

…attractor-based formalism, where the neural dynamics asymptotically converge to one of the system’s stable states, and these states are subsequently associated with the behavioral states of the system [1,2]. Although this framework explains a wide range of phenomena, it cannot explicitly account for transiently stable sequential neuronal activity dynamics, generated for example in olfactory encoding [3]. Recently, we have proposed a complementary theoretical framework that does not rely on (un)stable attractors, but rather on transiently

Most models of neural computation used in artificial intelligence and neuroscience use point neurons, attributing computational complexity to the network level. This approach overlooks the computational complexity of single biological neurons. Neural dendrites exhibit many different nonlinear dynamics, temporal-spatial filtering, and internal feedback and amplification mechanisms (London & Häusser 2005). Abstracting the entire dendritic arbor into a single compartment thus omits useful computational properties (Poirazi et al. 2003). Three main factors contribute to the integration process: (1)

The importance of dendritic structures for neural computation has been recognized since the early days of neuroscience. However, it is only recently, with the increasing availability of data and advances in modelling techniques, that researchers have begun to map out the precise roles of different branching structures and synaptic distributions [1, 2]. Here, we investigate how the nonlinear integration of excitatory and inhibitory currents in dendritic branches can enhance the ability of a neuron to develop input selectivity and E/I balance via Hebbian synaptic plasticity. We model a single ra

In the brain, the collective activity of neurons form trajectories on low dimensional neural manifolds. The neural computation underlying flexible cognition and behavior relies on dynamic control of these structures. For example, different tasks or behaviors are represented on different subspaces, requiring fast timescale subspace rotation to move from one behavior to the next. For flexibility in a particular behavior, the neural trajectory must be dynamically controllable within that behaviorally determined subspace. Here we propose mechanisms for fast timescale control of the neural subspace

…Recent biological findings[1] and computational models suggest that dendritic plateau potentia

Neural computation is inextricably bound to decisions that must be made under time pressure and uncertainty. At the level of neural circuits, single neurons need to decide whether to spike. On longer timescales, the component circuitry needs to decide whether to reconfigure to store memories and adapt to novel situations. In this talk I will focus on two fun ideas in each of these contexts by showing how nonlinearities in neural components naturally form excitable switches that enable reliable decisions to be made in fluctuating environments. I will also issue propaganda that the kind of high

…needs to be computed by subsequent neural circuits, involving a comparison of the signals from neighboring photoreceptors over time. The exact nature of this process represents a classic example of neural computation and has been a longstanding question in the field. Much progress has been made in recent years

An emerging perspective in neuroscience connects neural dynamics, or the rules that govern how neural population activity develops over time, to functional computations performed by the brain [1]. Unfortunately, neural dynamics cannot be observed directly and must instead be inferred from neural recordings. One promising method to infer these dynamics is to use a latent dynamics model, commonly a Sequential Auto-Encoder (SAE), to reconstruct neural activity [2] with the assumption that better reconstruction performance corresponds to better dynamical accuracy. Unfortunately, most SAE readouts,

Attractor dynamics are essential components of neural computation— they have given rise to elegant, interpretable models of population coding and working memory. Frequently, attractors are classified based on whether their invariant manifolds can represent discrete/continuous variables, with point attractors being an example of the former and ‘continuous attractors’ of the latter. Here we highlight that ‘learnability’ is another quality that may serve to differentiate neural dynamics while offering novel insight. Counterintuitively, good working memory does not necessarily imply that a well-de

Neuronal computation is mediated by spikes, yet it is unclear what the benefits of discrete spiking dynamics over continuous firing rates are. Many theoretical and computational studies treat single neurons as continuous units and their output as an effective firing rate. These works view spiking dynamics as a biophysical constraint, e.g., for energy efficiency. Conversely, other theories suggest that the exact timings of single spikes are meaningful. We propose a novel theory that shows the benefits of spiking dynamics on neural computation. In particular, we show that spiking neural dynamics

The physiological properties of neurons influence their population dynamics to shape neural computation and behavior. These physiological properties are diverse: the brain is composed of many genetically distinct neuronal cell types with pronounced physiological and functional differences. But neurons of the same cell type are also not identical, and exhibit heterogeneity in their physiological properties. We call this "within-type" heterogeneity, in contrast to the former, "between-type" heterogeneity. Whereas the contributions of distinct neuron types to neural network dynamics has been stud

Predictive coding (PC) has emerged as an influential normative model of neural computation, with numerous extensions and applications. As such, much effort has been put into mapping PC faithfully onto the cortex, but there are issues that remain unresolved or controversial. In particular, current implementations often involve biologically unrealistic features such as separate value and error neurons and symmetric forward and backward weights across different brain regions. In this work, we show that the PC framework in the linear regime can be modified to map faithfully onto the cortical hier

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

…and complex synapses, determine neural computation beyond this apparent nonlinearity. While it has been speculated that the dendritic tree of a neuron can be seen as a multi-layer neural network and it has been shown that such an architecture could be computationally strong,

…needs to be computed by subsequent neural circuits. The exact nature of this process represents a classic example of neural computation and has been a longstanding question in the field. Our results obtained in the fruit fly Drosophila demonstrate that the local direction of motion is computed in two parallel

Neural computation depends on inter-areal signal transformations, which are determined by the input-output (I/O) functions of individual and networks of neurons. Due to intrinsic neuronal properties and inter-neuronal interactions, networks can show preferences for synaptic inputs in certain frequencies, e.g. in the form of lowpass filtering or resonance [Izhikevich 2003, Cardin et al 2009, Lewis et al 2021, Pike et al 2000]. High-density silicon-probe recordings were made from area V1 and V2 in awake mice. We used optogenetic stimulation (continuous (1s), pulses (5ms), sinusoids) with an opsi

Substantial progress towards understanding neural computation has been made by studying the dynamical systems that govern neural population activity. Because dynamical models make predictions about the temporal evolution of neural states, perturbations of neural activity are a powerful tool to evaluate and refine these models. Here, using electrodes and Neuropixels in the motor cortices of macaques engaged in a reaching task, we probed how local population dynamics are affected by two perturbation modalities: optogenetic excitation and electrical intracortical microstimulation (ICMS). Both st