Receptive fields

In sensory systems, stimuli are represented through the diverse firing responses and receptive fields of neurons. These features emerge from the interaction between excitatory (E) and inhibitory (I) neuron populations within the network. Changes in sensory inputs alter this balance, leading to shifts in firing patterns and the input-output properties of individual neurons and the network. While these phenomena have been studied extensively with experiments and theory, the underlying principles for combining E and I inputs are still unclear. Here, the rules for probabilistically combining E and

…continuity is predictive remapping of receptive fields (RF) of neurons: neurons across visual and oculomotor areas predictively respond to visual stimuli that will appear in their receptive field after an impending saccade. This update of visual representations is mediated by an efferent motor-to-sensory signal

Recent large-scale community effort has produced comprehensive mapping of neurons and chemical synapses in both the Drosophila larva and adult brains (‘connectomes’). In addition, the neurotransmitter expression of each neuron can be predicted using a machine learning algorithm. However, interpreting connectomics data has proven challenging: most neurons connect to tens/hundreds of upstream/downstream neurons, and removing weak connections eliminates most connections in the brain. This could lead to biassed usage of connectomics data, favouring neurons previously studied, or those accessible w

…Here we characterize L4 and L2/3 receptive fields (RFs) and the L4 orientation map at eye-opening. We then model the connectivity by which L4 i

Understanding how cortical circuits encode sensory information remains a key challenge. Given a stimulus, the central question from the prespecitive of probablistic theories is how posterior probablities are neurally encoded. In recent years, standard approaches like sampling-based and variational approximations have been extended by deep neural networks (DNNs). For variational approaches, DNNs are frequently used to define encoders in variational autoencoders (VAE) settings. It remains unclear, however, what aspects of posterior approximations are important to explain neural encoding . Unlik

…has two sub-units and corresponding receptive fields (RFs), and factorizes its amplitude and phase response through a recurrent inference procedure. The model also learns localized, oriented, and band-pass RFs from natural images, and exhibits signature V1 complex cell phase inv

Relational invariance---a network’s ability to capture the geometric relationships between input patterns abstracted away from explicit external variables---has the potential to enable robustness and generalizability in neural networks. Previous work from our group has shown that a combinatorial neuronal code preserves the relational geometry of acoustic stimuli in the songbird brain. Here we show that combinatorial codes in autoencoding spiking neural networks (SNNs) can maintain relational invariance and display a hierarchical structure that optimizes the network’s internal representation. B

While sensory coding is traditionally thought of as a process independent of behavioral state, modern studies showcase potent modulatory effects of movement on sensory neurons. Intriguingly, these effects are not uniform and can differ between species. In insects and rodents, visual responses increase dramatically when the animal is moving, whereas in primates locomotion has a weak impact on responses in the primary visual cortex (V1). These differences raise questions about the existence and universality of computational principles underlying sensory coding during behavior. We take a normativ

Recent research in awake, behaving organisms revealed the strong modulatory effects of movement on sensory coding. Surprisingly, these effects are not consistent across species. For example, in rodents and insects locomotion increases the magnitude of visual responses [1-3], while in primates, locomotion has a weak suppressive influence [4]. These differences raise intriguing questions about the computational purpose of such modulations and the generality of the underlying principles of sensory processing. Here, we address these questions from a theoretical perspective. Our approach is groun

Spontaneous activity in young ferrets (p22) is organised into modular, low-dimensional patterns across multiple cortical areas [1]. In the following two weeks (p22-43), these activity patterns undergo a transition similar in all areas, and show a gradual increase in dimensionality in most areas, suggesting an expansion of representational capacity [2]. However, it is unclear what the functional role of having low-dimensional activity patterns at the initial stage of development is, since one would expect the limited dimensionality to restrict representational learning. In this work, we investi

…nonlinearly to stimuli inside their receptive fields (RFs) [1-3], and these neurophysiological properties could mediate spatial suppression of motion obs

The role of sensory systems is to interpret real-world events to facilitate behavior. For instance, the visual system enacts compensatory eye movements based on a global motion representation, generated by combining local motion signals from the retina. However, due to the aperture problem, local motion signals are inherently ambiguous, and how the visual system solves the aperture problem is unclear. We studied how the mouse Nucleus of the Optic Tract (NOT) integrates retinal visual information and how this supports behavioral responses. NOT mediates the horizontal optokinetic reflex, the com

…of image elements within their small receptive fields, while neurons in higher visual areas represent the visual scene in a more global and abstract manner (e.g., by identification and segmentation of objects). However, when processing ambiguous visual stimuli, such as those generated by Kanizsa figures, the local analysis by lower levels might be insufficient for

…axon terminals and their corresponding receptive

…suggested that macaque V1 neurons’ receptive fields are more complex and diverse in feature selectivity (Tang et al 2017) rather than just oriented Gabor filters. However, that finding was based on extensive parametric artificial stimuli used and might be biased. Hence, the nature of the neural codes of macaque V1 neurons remains controversial. To resolve this question, we performed 2-photon calcium imaging on three macaque monkeys to obtain 1689 V1 neur

Normative principles describe how to efficiently encode visual scenes, and supervised learning models show how formal tasks shape visual processing, yet animal vision ultimately adapts to encode survival-relevant features. To study how ecology constrains visual processing, we present a reinforcement learning (RL) framework in which an agent aims to survive in a 3-d environment that it perceives through a vision model. In particular, we consider a visually-guided foraging task: the environment is populated with randomly placed objects, and the agent must learn to identify those that increase or

The processing of natural visual environments rich with motion is central to animal survival. Many aspects of perceiving and detecting visual motion are already mature before the onset of visual experience. How the visual system develops to enable these functions in the absence of external stimuli remains unclear. A prominent transient feature of developing circuits is the ability to generate complex spatiotemporal patterns of spontaneous activity, even before sensory organs mature. For example, before vision onset in the mammalian retina, activity propagates in the form of spatiotemporally st

…to this challenge is to compute receptive fields by correlating neural activity with external variables drawn from sensory signals. But these receptive fields are only meaningful to the experimenter, not the organism, because only the experimenter has access to both the neural activity and knowledge of the external variables. To understand neural representation more directly, recent methodological advances have sought to capture the intrinsic geometry of sensory driven neural responses

…While excitatory HD cells had narrow receptive fields, fast-spiking (FS) interneurons had broad an

The upper visual layers of the mouse superior colliculus (SC) exhibit an orientation preference map (OPM) with a retinotopy-based concentric pattern (Ahmadlou and Heimel, 2015; Feinberg and Meister, 2015; but see Chen et al., 2020). The underlying developmental mechanisms that give rise to this concentric OPM in the mouse SC are still unclear. During visual system development, the stage III (S3) retinal waves are known to play important roles in circuit formation. Specifically, the S3 waves that tend to propagate towards the caudal direction (Gribizis et al., 2019) have impacts on the directio

The responses of mammalian visual system to the increment (ON) and decrement (OFF) of light are asymmetrical. Previous studies have shown that the OFF responses in early visual systems of higher mammals are dominant. Recently, the OFF dominance had been shown in mouse primary visual cortex (V1), but it is unclear whether the feedforward inputs from dorsal lateral geniculate nucleus (dLGN) to V1 in mouse are OFF dominant. Here, we labeled thalamocortical projecting cells in dLGN with calcium indicator and imaged their axonal bouton activity at different depths in V1. By using locally sparse noi

Sensory coding in olfaction remains poorly understood. Unlike in vision, where the stimulus space is well-characterized, the complexity of the "odor space" and the complexity of the olfactory receptor (OR) repertoire has hindered our understanding of odor coding and processing. There is no consensus on key defining features of odor sensation, such as the dimensionality of odor perceptual or odor receptor space. It is commonly conjectured that the ORs transform the high-dimensional "chemical space" to a relatively low-dimensional space of neuronal representations. We developed a novel method f

…functions whose shapes resemble the receptive fields (RFs) of simple cells in the primary visual cortex (V1). However, few studies have considered how these basis functions develop during training. In particular, it is unclear whether certain types of basis functions emerge more quickly than others, or whether they develop simultaneously. In this work, we train an overcomplete sparse coding model (Sparsenet) on natural images and find that there is indeed order in the development of its basis functions, with lower

…of convolutions, namely localised receptive fields that tile the input space, can be implemented with fully-connected neural networks, learning convolutions directly from inputs in a fully-connected network has so far proven