Interpreting the retinal neural code for natural scenes: From computations to neurons

A systematic analysis of the joint effects of ganglion cells, lagged LGN cells, and intercortical inhibition on spatiotemporal processing and direction selectivity

Simple cells in the visual cortex process spatial as well as temporal information of the visual stream and enable the perception of motion information. Previous work suggests different mechanisms associated with direction selectivity, such as a temporal offset in thalamocortical input stream through lagged and non-lagged cells of the lateral geniculate nucleus (LGN), or solely from intercortical inhibition, or through a baseline selectivity provided by the thalamocortical connection tuned by intercortical inhibition. While there exists a large corpus of models for spatiotemporal receptive fields, the majority of them built-in the spatiotemporal dynamics by utilizing a combination of spatial and temporal functions and thus, do not explain the emergence of spatiotemporal dynamics on basis of network dynamics emerging in the retina and the LGN. In order to better comprehend the emergence of spatiotemporal processing and direction selectivity, we used a spiking neural network to implement the visual pathway from the retina to the primary visual cortex. By varying different functional parts in our network, we demonstrate how the direction selectivity of simple cells emerges through the interplay between two components: tuned intercortical inhibition and a temporal offset in the feedforward path through lagged LGN cells. In contrast to previous findings, our model simulations suggest an alternative dynamic between these two mechanisms: While intercortical inhibition alone leads to bidirectional selectivity, a temporal shift in the thalamocortical pathway breaks this symmetry in favor of one direction, leading to unidirectional selectivity.

Interpretable deep learning for deconvolutional analysis of neural signals

The widespread adoption of deep learning to model neural activity often relies on "black-box" approaches that lack an interpretable connection between neural activity and network parameters. Here, we propose using algorithm unrolling, a method for interpretable deep learning, to design the architecture of sparse deconvolutional neural networks and obtain a direct interpretation of network weights in relation to stimulus-driven single-neuron activity through a generative model. We introduce our method, deconvolutional unrolled neural learning (DUNL), and demonstrate its versatility by applying it to deconvolve single-trial local signals across multiple brain areas and recording modalities. We uncover multiplexed salience and reward prediction error signals from midbrain dopamine neurons, perform simultaneous event detection and characterization in somatosensory thalamus recordings, and characterize the heterogeneity of neural responses in the piriform cortex and across striatum during unstructured, naturalistic experiments. Our work leverages advances in interpretable deep learning to provide a mechanistic understanding of neural activity.

Neural models for detection and classification of brain states and transitions

Exploring natural or pharmacologically induced brain dynamics, such as sleep, wakefulness, or anesthesia, provides rich functional models for studying brain states. These models allow detailed examination of unique spatiotemporal neural activity patterns that reveal brain function. However, assessing transitions between brain states remains computationally challenging. Here we introduce a pipeline to detect brain states and their transitions in the cerebral cortex using a dual-model Convolutional Neural Network (CNN) and a self-supervised autoencoder-based multimodal clustering algorithm. This approach distinguishes brain states such as slow oscillations, microarousals, and wakefulness with high confidence. Using chronic local field potential recordings from rats, our method achieved a global accuracy of 91%, with up to 96% accuracy for certain states. For the transitions, we report an average accuracy of 74%. Our models were trained using a leave-one-out methodology, allowing for broad applicability across subjects and pre-trained models for deployments. It also features a confidence parameter, ensuring that only highly certain cases are automatically classified, leaving ambiguous cases for the multimodal unsupervised classifier or further expert review. Our approach presents a reliable and efficient tool for brain state labeling and analysis, with applications in basic and clinical neuroscience.

Orthogonal neural representations support perceptual judgments of natural stimuli

In natural visually guided behavior, observers must separate relevant information from a barrage of irrelevant information. Many studies have investigated the neural underpinnings of this ability using artificial stimuli presented on blank backgrounds. Natural images, however, contain task-irrelevant background elements that might interfere with the perception of object features. Recent studies suggest that visual feature estimation can be modeled through the linear decoding of task-relevant information from visual cortex. So, if the representations of task-relevant and irrelevant features are not orthogonal in the neural population, then variation in the task-irrelevant features would impair task performance. We tested this hypothesis using human psychophysics and monkey neurophysiology combined with parametrically variable naturalistic stimuli. We demonstrate that (1) the neural representation of one feature (the position of an object) in visual area V4 is orthogonal to those of several background features, (2) the ability of human observers to precisely judge object position was largely unaffected by those background features, and (3) many features of the object and the background (and of objects from a separate stimulus set) are orthogonally represented in V4 neural population responses. Our observations are consistent with the hypothesis that orthogonal neural representations can support stable perception of object features despite the richness of natural visual scenes.

Universality of representation in biological and artificial neural networks

Many artificial neural networks (ANNs) trained with ecologically plausible objectives on naturalistic data align with behavior and neural representations in biological systems. Here, we show that this alignment is a consequence of convergence onto the same representations by high-performing ANNs and by brains. We developed a method to identify stimuli that systematically vary the degree of inter-model representation agreement. Across language and vision, we then showed that stimuli from high- and low-agreement sets predictably modulated model-to-brain alignment. We also examined which stimulus features distinguish high- from low-agreement sentences and images. Our results establish representation universality as a core component in the model-to-brain alignment and provide a new approach for using ANNs to uncover the structure of biological representations and computations.

Stimulus-invariant aspects of the retinal code drive discriminability of natural scenes

Everything that the brain sees must first be encoded by the retina, which maintains a reliable representation of the visual world in many different, complex natural scenes while also adapting to stimulus changes. This study quantifies whether and how the brain selectively encodes stimulus features about scene identity in complex naturalistic environments. While a wealth of previous work has dug into the static and dynamic features of the population code in retinal ganglion cells (RGCs), less is known about how populations form both flexible and reliable encoding in natural moving scenes. We record from the larval salamander retina responding to five different natural movies, over many repeats, and use these data to characterize the population code in terms of single-cell fluctuations in rate and pairwise couplings between cells. Decomposing the population code into independent and cell-cell interactions reveals how broad scene structure is encoded in the retinal output. while the single-cell activity adapts to different stimuli, the population structure captured in the sparse, strong couplings is consistent across natural movies as well as synthetic stimuli. We show that these interactions contribute to encoding scene identity. We also demonstrate that this structure likely arises in part from shared bipolar cell input as well as from gap junctions between RGCs and amacrine cells.

Interpretable deep learning for deconvolutional analysis of neural signals

The widespread adoption of deep learning to build models that capture the dynamics of neural populations is typically based on "black-box" approaches that lack an interpretable link between neural activity and network parameters. Here, we propose to apply algorithm unrolling, a method for interpretable deep learning, to design the architecture of sparse deconvolutional neural networks and obtain a direct interpretation of network weights in relation to stimulus-driven single-neuron activity through a generative model. We characterize our method, referred to as deconvolutional unrolled neural learning (DUNL), and show its versatility by applying it to deconvolve single-trial local signals across multiple brain areas and recording modalities. To exemplify use cases of our decomposition method, we uncover multiplexed salience and reward prediction error signals from midbrain dopamine neurons in an unbiased manner, perform simultaneous event detection and characterization in somatosensory thalamus recordings, and characterize the heterogeneity of neural responses in the piriform cortex and in the striatum during unstructured, naturalistic experiments. Our work leverages the advances in interpretable deep learning to gain a mechanistic understanding of neural activity.

Adaptation of retinal discriminability to natural scenes

Sensory systems discriminate between stimuli to direct behavioral choices, a process governed by two distinct properties - neural sensitivity to specific stimuli, and neural variability that importantly includes correlations between neurons. Two questions that have received extensive investigation and debate are whether visual systems are optimized for natural scenes, and whether correlated neural variability contributes to this optimization. However, the lack of sufficient computational models has made these questions inaccessible in the context of the normal function of the visual system, which is to discriminate between natural stimuli. Here we take a direct approach to analyze discriminability under natural scenes for a population of salamander retinal ganglion cells using a model of the retinal neural code that captures both sensitivity and variability. Using methods of information geometry and generative machine learning, we analyzed the manifolds of natural stimuli and neural responses, finding that discriminability in the ganglion cell population adapts to enhance information transmission about natural scenes, in particular about localized motion. Contrary to previous proposals, correlated noise reduces information transmission and arises simply as a natural consequence of the shared circuitry that generates changing spatiotemporal visual sensitivity. These results address a long-standing debate as to the role of retinal correlations in the encoding of natural stimuli and reveal how the highly nonlinear receptive fields of the retina adapt dynamically to increase information transmission under natural scenes by performing the important ethological function of local motion discrimination.

Plaid masking explained with input-dependent dendritic nonlinearities

A serious obstacle for understanding early spatial vision comes from the failure of the so-called standard model (SM) to predict the perception of plaid masking. But the SM originated from a major oversimplification of single neuron computations, ignoring fundamental properties of dendrites. Here we show that a spatial vision model including computations mimicking the input-dependent nature of dendritic nonlinearities, i.e. including nonlinear neural summation, has the potential to explain plaid masking data.

A chromatic feature detector in the retina signals visual context changes

The retina transforms patterns of light into visual feature representations supporting behaviour. These representations are distributed across various types of retinal ganglion cells (RGCs), whose spatial and temporal tuning properties have been studied extensively in many model organisms, including the mouse. However, it has been difficult to link the potentially nonlinear retinal transformations of natural visual inputs to specific ethological purposes. Here, we discover a nonlinear selectivity to chromatic contrast in an RGC type that allows the detection of changes in visual context. We trained a convolutional neural network (CNN) model on large-scale functional recordings of RGC responses to natural mouse movies, and then used this model to search in silico for stimuli that maximally excite distinct types of RGCs. This procedure predicted centre colour opponency in transient suppressed-by-contrast (tSbC) RGCs, a cell type whose function is being debated. We confirmed experimentally that these cells indeed responded very selectively to Green-OFF, UV-ON contrasts. This type of chromatic contrast was characteristic of transitions from ground to sky in the visual scene, as might be elicited by head or eye movements across the horizon. Because tSbC cells performed best among all RGC types at reliably detecting these transitions, we suggest a role for this RGC type in providing contextual information (i.e. sky or ground) necessary for the selection of appropriate behavioural responses to other stimuli, such as looming objects. Our work showcases how a combination of experiments with natural stimuli and computational modelling allows discovering novel types of stimulus selectivity and identifying their potential ethological relevance.

Applying Super-Resolution and Tomography Concepts to Identify Receptive Field Subunits in the Retina

Spatially nonlinear stimulus integration by retinal ganglion cells lies at the heart of various computations performed by the retina. It arises from the nonlinear transmission of signals that ganglion cells receive from bipolar cells, which thereby constitute functional subunits within a ganglion cell's receptive field. Inferring these subunits from recorded ganglion cell activity promises a new avenue for studying the functional architecture of the retina. This calls for efficient methods, which leave sufficient experimental time to leverage the acquired knowledge for further investigating identified subunits. Here, we combine concepts from super-resolution microscopy and computed tomography and introduce super-resolved tomographic reconstruction (STR) as a technique to efficiently stimulate and locate receptive field subunits. Simulations demonstrate that this approach can reliably identify subunits across a wide range of model variations, and application in recordings of primate parasol ganglion cells validates the experimental feasibility. STR can potentially reveal comprehensive subunit layouts within only a few tens of minutes of recording time, making it ideal for online analysis and closed-loop investigations of receptive field substructure in retina recordings.

Biophysical neural adaptation mechanisms enable artificial neural networks to capture dynamic retinal computation

Adaptation is a universal aspect of neural systems that changes circuit computations to match prevailing inputs. These changes facilitate efficient encoding of sensory inputs while avoiding saturation. Conventional artificial neural networks (ANNs) have limited adaptive capabilities, hindering their ability to reliably predict neural output under dynamic input conditions. Can embedding neural adaptive mechanisms in ANNs improve their performance? To answer this question, we develop a new deep learning model of the retina that incorporates the biophysics of photoreceptor adaptation at the front-end of conventional convolutional neural networks (CNNs). These conventional CNNs build on 'Deep Retina,' a previously developed model of retinal ganglion cell (RGC) activity. CNNs that include this new photoreceptor layer outperform conventional CNN models at predicting male and female primate and rat RGC responses to naturalistic stimuli that include dynamic local intensity changes and large changes in the ambient illumination. These improved predictions result directly from adaptation within the phototransduction cascade. This research underscores the potential of embedding models of neural adaptation in ANNs and using them to determine how neural circuits manage the complexities of encoding natural inputs that are dynamic and span a large range of light levels.

Orthogonal neural representations support perceptual judgements of natural stimuli

In natural behavior, observers must separate relevant information from a barrage of irrelevant information. Many studies have investigated the neural underpinnings of this ability using artificial stimuli presented on simple backgrounds. Natural viewing, however, carries a set of challenges that are inaccessible using artificial stimuli, including neural responses to background objects that are task-irrelevant. An emerging body of evidence suggests that the visual abilities of humans and animals can be modeled through the linear decoding of task-relevant information from visual cortex. This idea suggests the hypothesis that irrelevant features of a natural scene should impair performance on a visual task only if their neural representations intrude on the linear readout of the task relevant feature, as would occur if the representations of task-relevant and irrelevant features are not orthogonal in the underlying neural population. We tested this hypothesis using human psychophysics and monkey neurophysiology, in response to parametrically variable naturalistic stimuli. We demonstrate that 1) the neural representation of one feature (the position of a central object) in visual area V4 is orthogonal to those of several background features, 2) the ability of human observers to precisely judge object position was largely unaffected by task-irrelevant variation in those background features, and 3) many features of the object and the background are orthogonally represented by V4 neural responses. Our observations are consistent with the hypothesis that orthogonal neural representations can support stable perception of objects and features despite the tremendous richness of natural visual scenes.

Classification and analysis of retinal interneurons by computational structure under natural scenes

Inhibitory neurons are diverse across the brain, but for the visual system we lack the ability to functionally classify these neurons under complex natural stimuli. Here we take the approach of classifying retinal amacrine cell responses to natural scenes using optical recording and an interpretable neural network model. We fit mouse amacrine cell responses to a two-layer convolutional neural network model of a class shown previously to accurately capture salamander ganglion cell responses to natural scenes. Using an approach from interpretable machine learning, we determined for each stimulus the model interneurons that generated each amacrine response, analogous to the set of bipolar cells that target the amacrine population. From this analysis we clustered amacrine cells not by their natural scene responses, but by the model presynaptic neurons that constructed those responses, conservatively finding approximately seven groups by this approach. By analyzing the set of model presynaptic input neurons for each amacrine cluster, we find that distributed rather than dedicated inputs generate natural scene responses for different amacrine cell types. Additional analyses revealed distinct transient and sustained modes exhibited by the network during the response to simple flashes. These results give insight into the computational structure of how the diverse amacrine cell population responds to natural scenes, and generate multiple quantitative hypotheses for how synaptic inputs generate those responses.

Encoding surprise by retinal ganglion cells

The efficient coding hypothesis posits that early sensory neurons transmit maximal information about sensory stimuli, given internal constraints. A central prediction of this theory is that neurons should preferentially encode stimuli that are most surprising. Previous studies suggest this may be the case in early visual areas, where many neurons respond strongly to rare or surprising stimuli. For example, previous research showed that when presented with a rhythmic sequence of full-field flashes, many retinal ganglion cells (RGCs) respond strongly at the instance the flash sequence stops, and when another flash would be expected. This phenomenon is called the 'omitted stimulus response'. However, it is not known whether the responses of these cells varies in a graded way depending on the level of stimulus surprise. To investigate this, we presented retinal neurons with extended sequences of stochastic flashes. With this stimulus, the surprise associated with a particular flash/silence, could be quantified analytically, and varied in a graded manner depending on the previous sequences of flashes and silences. Interestingly, we found that RGC responses could be well explained by a simple normative model, which described how they optimally combined their prior expectations and recent stimulus history, so as to encode surprise. Further, much of the diversity in RGC responses could be explained by the model, due to the different prior expectations that different neurons had about the stimulus statistics. These results suggest that even as early as the retina many cells encode surprise, relative to their own, internally generated expectations.

Response sub-additivity and variability quenching in visual cortex

Sub-additivity and variability are ubiquitous response motifs in the primary visual cortex (V1). Response sub-additivity enables the construction of useful interpretations of the visual environment, whereas response variability indicates the factors that limit the precision with which the brain can do this. There is increasing evidence that experimental manipulations that elicit response sub-additivity often also quench response variability. Here, we provide an overview of these phenomena and suggest that they may have common origins. We discuss empirical findings and recent model-based insights into the functional operations, computational objectives and circuit mechanisms underlying V1 activity. These different modelling approaches all predict that response sub-additivity and variability quenching often co-occur. The phenomenology of these two response motifs, as well as many of the insights obtained about them in V1, generalize to other cortical areas. Thus, the connection between response sub-additivity and variability quenching may be a canonical motif across the cortex.

Modeling responses of macaque and human retinal ganglion cells to natural images using a convolutional neural network

Linear-nonlinear (LN) cascade models provide a simple way to capture retinal ganglion cell (RGC) responses to artificial stimuli such as white noise, but their ability to model responses to natural images is limited. Recently, convolutional neural network (CNN) models have been shown to produce light response predictions that were substantially more accurate than those of a LN model. However, this modeling approach has not yet been applied to responses of macaque or human RGCs to natural images. Here, we train and test a CNN model on responses to natural images of the four numerically dominant RGC types in the macaque and human retina - ON parasol, OFF parasol, ON midget and OFF midget cells. Compared with the LN model, the CNN model provided substantially more accurate response predictions. Linear reconstructions of the visual stimulus were more accurate for CNN compared to LN model-generated responses, relative to reconstructions obtained from the recorded data. These findings demonstrate the effectiveness of a CNN model in capturing light responses of major RGC types in the macaque and human retinas in natural conditions.

Distributed feature representations of natural stimuli across parallel retinal pathways

How sensory systems extract salient features from natural environments and organize them across neural pathways is unclear. Combining single-cell and population two-photon calcium imaging in mice, we discover that retinal ON bipolar cells (second-order neurons of the visual system) are divided into two blocks of four types. The two blocks distribute temporal and spatial information encoding, respectively. ON bipolar cell axons co-stratify within each block, but separate laminarly between them (upper block: diverse temporal, uniform spatial tuning; lower block: diverse spatial, uniform temporal tuning). ON bipolar cells extract temporal and spatial features similarly from artificial and naturalistic stimuli. In addition, they differ in sensitivity to coherent motion in naturalistic movies. Motion information is distributed across ON bipolar cells in the upper and the lower blocks, multiplexed with temporal and spatial contrast, independent features of natural scenes. Comparing the responses of different boutons within the same arbor, we find that axons of all ON bipolar cell types function as computational units. Thus, our results provide insights into the visual feature extraction from naturalistic stimuli and reveal how structural and functional organization cooperate to generate parallel ON pathways for temporal and spatial information in the mammalian retina.

Compact deep neural network models of visual cortex

A powerful approach to understanding the computations carried out in visual cortex is to develop models that predict neural responses to arbitrary images. Deep neural network (DNN) models have worked remarkably well at predicting neural responses [1, 2, 3], yet their underlying computations remain buried in millions of parameters. Have we simply replaced one complicated system in vivo with another in silico? Here, we train a data-driven deep ensemble model that predicts macaque V4 responses ~50% more accurately than currently-used task-driven DNN models. We then compress this deep ensemble to identify compact models that have 5,000x fewer parameters yet equivalent accuracy as the deep ensemble. We verified that the stimulus preferences of the compact models matched those of the real V4 neurons by measuring V4 responses to both 'maximizing' and adversarial images generated using compact models. We then analyzed the inner workings of the compact models and discovered a common circuit motif: Compact models share a similar set of filters in early stages of processing but then specialize by heavily consolidating this shared representation with a precise readout. This suggests that a V4 neuron's stimulus preference is determined entirely by its consolidation step. To demonstrate this, we investigated the compression step of a dot-detecting compact model and found a set of simple computations that may be carried out by dot-selective V4 neurons. Overall, our work demonstrates that the DNN models currently used in computational neuroscience are needlessly large; our approach provides a new way forward for obtaining explainable, high-accuracy models of visual cortical neurons.

Large-scale electrophysiology and deep learning reveal distorted neural signal dynamics after hearing loss

Listeners with hearing loss often struggle to understand speech in noise, even with a hearing aid. To better understand the auditory processing deficits that underlie this problem, we made large-scale brain recordings from gerbils, a common animal model for human hearing, while presenting a large database of speech and noise sounds. We first used manifold learning to identify the neural subspace in which speech is encoded and found that it is low-dimensional and that the dynamics within it are profoundly distorted by hearing loss. We then trained a deep neural network (DNN) to replicate the neural coding of speech with and without hearing loss and analyzed the underlying network dynamics. We found that hearing loss primarily impacts spectral processing, creating nonlinear distortions in cross-frequency interactions that result in a hypersensitivity to background noise that persists even after amplification with a hearing aid. Our results identify a new focus for efforts to design improved hearing aids and demonstrate the power of DNNs as a tool for the study of central brain structures.