Brain2GAN: Feature-disentangled neural encoding and decoding of visual perception in the primate brain.

A challenging goal of neural coding is to characterize the neural representations underlying visual perception. To this end, multi-unit activity (MUA) of macaque visual cortex was recorded in a passive fixation task upon presentation of faces and natural images. We analyzed the relationship between MUA and latent representations of state-of-the-art deep generative models, including the conventional and feature-disentangled representations of generative adversarial networks (GANs) (i.e., z- and w-latents of StyleGAN, respectively) and language-contrastive representations of latent diffusion networks (i.e., CLIP-latents of Stable Diffusion). A mass univariate neural encoding analysis of the latent representations showed that feature-disentangled w representations outperform both z and CLIP representations in explaining neural responses. Further, w-latent features were found to be positioned at the higher end of the complexity gradient which indicates that they capture visual information relevant to high-level neural activity. Subsequently, a multivariate neural decoding analysis of the feature-disentangled representations resulted in state-of-the-art spatiotemporal reconstructions of visual perception. Taken together, our results not only highlight the important role of feature-disentanglement in shaping high-level neural representations underlying visual perception but also serve as an important benchmark for the future of neural coding.

Gaze-contingent processing improves mobility, scene recognition and visual search in simulated head-steered prosthetic vision.

Objective.The enabling technology of visual prosthetics for the blind is making rapid progress. However, there are still uncertainties regarding the functional outcomes, which can depend on many design choices in the development. In visual prostheses with a head-mounted camera, a particularly challenging question is how to deal with the gaze-locked visual percept associated with spatial updating conflicts in the brain. The current study investigates a recently proposed compensation strategy based on gaze-contingent image processing with eye-tracking. Gaze-contingent processing is expected to reinforce natural-like visual scanning and reestablished spatial updating based on eye movements. The beneficial effects remain to be investigated for daily life activities in complex visual environments.Approach.The current study evaluates the benefits of gaze-contingent processing versus gaze-locked and gaze-ignored simulations in the context of mobility, scene recognition and visual search, using a virtual reality simulated prosthetic vision paradigm with sighted subjects.Main results.Compared to gaze-locked vision, gaze-contingent processing was consistently found to improve the speed in all experimental tasks, as well as the subjective quality of vision. Similar or further improvements were found in a control condition that ignores gaze-dependent effects, a simulation that is unattainable in the clinical reality.Significance.Our results suggest that gaze-locked vision and spatial updating conflicts can be debilitating for complex visually-guided activities of daily living such as mobility and orientation. Therefore, for prospective users of head-steered prostheses with an unimpaired oculomotor system, the inclusion of a compensatory eye-tracking system is strongly endorsed.

Towards biologically plausible phosphene simulation for the differentiable optimization of visual cortical prostheses.

Blindness affects millions of people around the world. A promising solution to restoring a form of vision for some individuals are cortical visual prostheses, which bypass part of the impaired visual pathway by converting camera input to electrical stimulation of the visual system. The artificially induced visual percept (a pattern of localized light flashes, or 'phosphenes') has limited resolution, and a great portion of the field's research is devoted to optimizing the efficacy, efficiency, and practical usefulness of the encoding of visual information. A commonly exploited method is non-invasive functional evaluation in sighted subjects or with computational models by using simulated prosthetic vision (SPV) pipelines. An important challenge in this approach is to balance enhanced perceptual realism, biologically plausibility, and real-time performance in the simulation of cortical prosthetic vision. We present a biologically plausible, PyTorch-based phosphene simulator that can run in real-time and uses differentiable operations to allow for gradient-based computational optimization of phosphene encoding models. The simulator integrates a wide range of clinical results with neurophysiological evidence in humans and non-human primates. The pipeline includes a model of the retinotopic organization and cortical magnification of the visual cortex. Moreover, the quantitative effects of stimulation parameters and temporal dynamics on phosphene characteristics are incorporated. Our results demonstrate the simulator's suitability for both computational applications such as end-to-end deep learning-based prosthetic vision optimization as well as behavioral experiments. The modular and open-source software provides a flexible simulation framework for computational, clinical, and behavioral neuroscientists working on visual neuroprosthetics.

Brain2Pix: Fully convolutional naturalistic video frame reconstruction from brain activity.

Reconstructing complex and dynamic visual perception from brain activity remains a major challenge in machine learning applications to neuroscience. Here, we present a new method for reconstructing naturalistic images and videos from very large single-participant functional magnetic resonance imaging data that leverages the recent success of image-to-image transformation networks. This is achieved by exploiting spatial information obtained from retinotopic mappings across the visual system. More specifically, we first determine what position each voxel in a particular region of interest would represent in the visual field based on its corresponding receptive field location. Then, the 2D image representation of the brain activity on the visual field is passed to a fully convolutional image-to-image network trained to recover the original stimuli using VGG feature loss with an adversarial regularizer. In our experiments, we show that our method offers a significant improvement over existing video reconstruction techniques.

Optimization of Neuroprosthetic Vision via End-to-End Deep Reinforcement Learning.

Visual neuroprostheses are a promising approach to restore basic sight in visually impaired people. A major challenge is to condense the sensory information contained in a complex environment into meaningful stimulation patterns at low spatial and temporal resolution. Previous approaches considered task-agnostic feature extractors such as edge detectors or semantic segmentation, which are likely suboptimal for specific tasks in complex dynamic environments. As an alternative approach, we propose to optimize stimulation patterns by end-to-end training of a feature extractor using deep reinforcement learning agents in virtual environments. We present a task-oriented evaluation framework to compare different stimulus generation mechanisms, such as static edge-based and adaptive end-to-end approaches like the one introduced here. Our experiments in Atari games show that stimulation patterns obtained via task-dependent end-to-end optimized reinforcement learning result in equivalent or improved performance compared to fixed feature extractors on high difficulty levels. These findings signify the relevance of adaptive reinforcement learning for neuroprosthetic vision in complex environments.

A partially nested cortical hierarchy of neural states underlies event segmentation in the human brain.

A fundamental aspect of human experience is that it is segmented into discrete events. This may be underpinned by transitions between distinct neural states. Using an innovative data-driven state segmentation method, we investigate how neural states are organized across the cortical hierarchy and where in the cortex neural state boundaries and perceived event boundaries overlap. Our results show that neural state boundaries are organized in a temporal cortical hierarchy, with short states in primary sensory regions, and long states in lateral and medial prefrontal cortex. State boundaries are shared within and between groups of brain regions that resemble well-known functional networks. Perceived event boundaries overlap with neural state boundaries across large parts of the cortical hierarchy, particularly when those state boundaries demarcate a strong transition or are shared between brain regions. Taken together, these findings suggest that a partially nested cortical hierarchy of neural states forms the basis of event segmentation.

A 10-hour within-participant magnetoencephalography narrative dataset to test models of language comprehension.

Recently, cognitive neuroscientists have increasingly studied the brain responses to narratives. At the same time, we are witnessing exciting developments in natural language processing where large-scale neural network models can be used to instantiate cognitive hypotheses in narrative processing. Yet, they learn from text alone and we lack ways of incorporating biological constraints during training. To mitigate this gap, we provide a narrative comprehension magnetoencephalography (MEG) data resource that can be used to train neural network models directly on brain data. We recorded from 3 participants, 10 separate recording hour-long sessions each, while they listened to audiobooks in English. After story listening, participants answered short questions about their experience. To minimize head movement, the participants wore MEG-compatible head casts, which immobilized their head position during recording. We report a basic evoked-response analysis showing that the responses accurately localize to primary auditory areas. The responses are robust and conserved across 10 sessions for every participant. We also provide usage notes and briefly outline possible future uses of the resource.

End-to-end optimization of prosthetic vision.

Neural prosthetics may provide a promising solution to restore visual perception in some forms of blindness. The restored prosthetic percept is rudimentary compared to normal vision and can be optimized with a variety of image preprocessing techniques to maximize relevant information transfer. Extracting the most useful features from a visual scene is a nontrivial task and optimal preprocessing choices strongly depend on the context. Despite rapid advancements in deep learning, research currently faces a difficult challenge in finding a general and automated preprocessing strategy that can be tailored to specific tasks or user requirements. In this paper, we present a novel deep learning approach that explicitly addresses this issue by optimizing the entire process of phosphene generation in an end-to-end fashion. The proposed model is based on a deep auto-encoder architecture and includes a highly adjustable simulation module of prosthetic vision. In computational validation experiments, we show that such an approach is able to automatically find a task-specific stimulation protocol. The results of these proof-of-principle experiments illustrate the potential of end-to-end optimization for prosthetic vision. The presented approach is highly modular and our approach could be extended to automated dynamic optimization of prosthetic vision for everyday tasks, given any specific constraints, accommodating individual requirements of the end-user.

Real-world indoor mobility with simulated prosthetic vision: The benefits and feasibility of contour-based scene simplification at different phosphene resolutions.

Neuroprosthetic implants are a promising technology for restoring some form of vision in people with visual impairments via electrical neurostimulation in the visual pathway. Although an artificially generated prosthetic percept is relatively limited compared with normal vision, it may provide some elementary perception of the surroundings, re-enabling daily living functionality. For mobility in particular, various studies have investigated the benefits of visual neuroprosthetics in a simulated prosthetic vision paradigm with varying outcomes. The previous literature suggests that scene simplification via image processing, and particularly contour extraction, may potentially improve the mobility performance in a virtual environment. In the current simulation study with sighted participants, we explore both the theoretically attainable benefits of strict scene simplification in an indoor environment by controlling the environmental complexity, as well as the practically achieved improvement with a deep learning-based surface boundary detection implementation compared with traditional edge detection. A simulated electrode resolution of 26 × 26 was found to provide sufficient information for mobility in a simple environment. Our results suggest that, for a lower number of implanted electrodes, the removal of background textures and within-surface gradients may be beneficial in theory. However, the deep learning-based implementation for surface boundary detection did not improve mobility performance in the current study. Furthermore, our findings indicate that, for a greater number of electrodes, the removal of within-surface gradients and background textures may deteriorate, rather than improve, mobility. Therefore, finding a balanced amount of scene simplification requires a careful tradeoff between informativity and interpretability that may depend on the number of implanted electrodes.

Hyperrealistic neural decoding for reconstructing faces from fMRI activations via the GAN latent space.

Neural decoding can be conceptualized as the problem of mapping brain responses back to sensory stimuli via a feature space. We introduce (i) a novel experimental paradigm that uses well-controlled yet highly naturalistic stimuli with a priori known feature representations and (ii) an implementation thereof for HYPerrealistic reconstruction of PERception (HYPER) of faces from brain recordings. To this end, we embrace the use of generative adversarial networks (GANs) at the earliest step of our neural decoding pipeline by acquiring fMRI data as participants perceive face images synthesized by the generator network of a GAN. We show that the latent vectors used for generation effectively capture the same defining stimulus properties as the fMRI measurements. As such, these latents (conditioned on the GAN) are used as the in-between feature representations underlying the perceived images that can be predicted in neural decoding for (re-)generation of the originally perceived stimuli, leading to the most accurate reconstructions of perception to date.

Detecting neural state transitions underlying event segmentation.

Segmenting perceptual experience into meaningful events is a key cognitive process that helps us make sense of what is happening around us in the moment, as well as helping us recall past events. Nevertheless, little is known about the underlying neural mechanisms of the event segmentation process. Recent work has suggested that event segmentation can be linked to regional changes in neural activity patterns. Accurate methods for identifying such activity changes are important to allow further investigation of the neural basis of event segmentation and its link to the temporal processing hierarchy of the brain. In this study, we introduce a new set of elegant and simple methods to study these mechanisms. We introduce a method for identifying the boundaries between neural states in a brain area and a complementary one for identifying the number of neural states. Furthermore, we present the results of a comprehensive set of simulations and analyses of empirical fMRI data to provide guidelines for reliable estimation of neural states and show that our proposed methods outperform the current state-of-the-art in the literature. This methodological innovation will allow researchers to make headway in investigating the neural basis of event segmentation and information processing during naturalistic stimulation.

Cortical network responses map onto data-driven features that capture visual semantics of movie fragments.

Research on how the human brain extracts meaning from sensory input relies in principle on methodological reductionism. In the present study, we adopt a more holistic approach by modeling the cortical responses to semantic information that was extracted from the visual stream of a feature film, employing artificial neural network models. Advances in both computer vision and natural language processing were utilized to extract the semantic representations from the film by combining perceptual and linguistic information. We tested whether these representations were useful in studying the human brain data. To this end, we collected electrocorticography responses to a short movie from 37 subjects and fitted their cortical patterns across multiple regions using the semantic components extracted from film frames. We found that individual semantic components reflected fundamental semantic distinctions in the visual input, such as presence or absence of people, human movement, landscape scenes, human faces, etc. Moreover, each semantic component mapped onto a distinct functional cortical network involving high-level cognitive regions in occipitotemporal, frontal and parietal cortices. The present work demonstrates the potential of the data-driven methods from information processing fields to explain patterns of cortical responses, and contributes to the overall discussion about the encoding of high-level perceptual information in the human brain.

Brain-optimized extraction of complex sound features that drive continuous auditory perception.

Understanding how the human brain processes auditory input remains a challenge. Traditionally, a distinction between lower- and higher-level sound features is made, but their definition depends on a specific theoretical framework and might not match the neural representation of sound. Here, we postulate that constructing a data-driven neural model of auditory perception, with a minimum of theoretical assumptions about the relevant sound features, could provide an alternative approach and possibly a better match to the neural responses. We collected electrocorticography recordings from six patients who watched a long-duration feature film. The raw movie soundtrack was used to train an artificial neural network model to predict the associated neural responses. The model achieved high prediction accuracy and generalized well to a second dataset, where new participants watched a different film. The extracted bottom-up features captured acoustic properties that were specific to the type of sound and were associated with various response latency profiles and distinct cortical distributions. Specifically, several features encoded speech-related acoustic properties with some features exhibiting shorter latency profiles (associated with responses in posterior perisylvian cortex) and others exhibiting longer latency profiles (associated with responses in anterior perisylvian cortex). Our results support and extend the current view on speech perception by demonstrating the presence of temporal hierarchies in the perisylvian cortex and involvement of cortical sites outside of this region during audiovisual speech perception.

Representations of naturalistic stimulus complexity in early and associative visual and auditory cortices.

The complexity of sensory stimuli has an important role in perception and cognition. However, its neural representation is not well understood. Here, we characterize the representations of naturalistic visual and auditory stimulus complexity in early and associative visual and auditory cortices. This is realized by means of encoding and decoding analyses of two fMRI datasets in the visual and auditory modalities. Our results implicate most early and some associative sensory areas in representing the complexity of naturalistic sensory stimuli. For example, parahippocampal place area, which was previously shown to represent scene features, is shown to also represent scene complexity. Similarly, posterior regions of superior temporal gyrus and superior temporal sulcus, which were previously shown to represent syntactic (language) complexity, are shown to also represent music (auditory) complexity. Furthermore, our results suggest the existence of gradients in sensitivity to naturalistic sensory stimulus complexity in these areas.

Neural Tuning to Low-Level Features of Speech throughout the Perisylvian Cortex.

Despite a large body of research, we continue to lack a detailed account of how auditory processing of continuous speech unfolds in the human brain. Previous research showed the propagation of low-level acoustic features of speech from posterior superior temporal gyrus toward anterior superior temporal gyrus in the human brain (Hullett et al., 2016). In this study, we investigate what happens to these neural representations past the superior temporal gyrus and how they engage higher-level language processing areas such as inferior frontal gyrus. We used low-level sound features to model neural responses to speech outside of the primary auditory cortex. Two complementary imaging techniques were used with human participants (both males and females): electrocorticography (ECoG) and fMRI. Both imaging techniques showed tuning of the perisylvian cortex to low-level speech features. With ECoG, we found evidence of propagation of the temporal features of speech sounds along the ventral pathway of language processing in the brain toward inferior frontal gyrus. Increasingly coarse temporal features of speech spreading from posterior superior temporal cortex toward inferior frontal gyrus were associated with linguistic features such as voice onset time, duration of the formant transitions, and phoneme, syllable, and word boundaries. The present findings provide the groundwork for a comprehensive bottom-up account of speech comprehension in the human brain.SIGNIFICANCE STATEMENT We know that, during natural speech comprehension, a broad network of perisylvian cortical regions is involved in sound and language processing. Here, we investigated the tuning to low-level sound features within these regions using neural responses to a short feature film. We also looked at whether the tuning organization along these brain regions showed any parallel to the hierarchy of language structures in continuous speech. Our results show that low-level speech features propagate throughout the perisylvian cortex and potentially contribute to the emergence of "coarse" speech representations in inferior frontal gyrus typically associated with high-level language processing. These findings add to the previous work on auditory processing and underline a distinctive role of inferior frontal gyrus in natural speech comprehension.

Modeling the Dynamics of Human Brain Activity with Recurrent Neural Networks.

Encoding models are used for predicting brain activity in response to sensory stimuli with the objective of elucidating how sensory information is represented in the brain. Encoding models typically comprise a nonlinear transformation of stimuli to features (feature model) and a linear convolution of features to responses (response model). While there has been extensive work on developing better feature models, the work on developing better response models has been rather limited. Here, we investigate the extent to which recurrent neural network models can use their internal memories for nonlinear processing of arbitrary feature sequences to predict feature-evoked response sequences as measured by functional magnetic resonance imaging. We show that the proposed recurrent neural network models can significantly outperform established response models by accurately estimating long-term dependencies that drive hemodynamic responses. The results open a new window into modeling the dynamics of brain activity in response to sensory stimuli.

Increasingly complex representations of natural movies across the dorsal stream are shared between subjects.

Recently, deep neural networks (DNNs) have been shown to provide accurate predictions of neural responses across the ventral visual pathway. We here explore whether they also provide accurate predictions of neural responses across the dorsal visual pathway, which is thought to be devoted to motion processing and action recognition. This is achieved by training deep neural networks to recognize actions in videos and subsequently using them to predict neural responses while subjects are watching natural movies. Moreover, we explore whether dorsal stream representations are shared between subjects. In order to address this question, we examine if individual subject predictions can be made in a common representational space estimated via hyperalignment. Results show that a DNN trained for action recognition can be used to accurately predict how dorsal stream responds to natural movies, revealing a correspondence in representations of DNN layers and dorsal stream areas. It is also demonstrated that models operating in a common representational space can generalize to responses of multiple or even unseen individual subjects to novel spatio-temporal stimuli in both encoding and decoding settings, suggesting that a common representational space underlies dorsal stream responses across multiple subjects.

Deep Neural Networks Reveal a Gradient in the Complexity of Neural Representations across the Ventral Stream.

Converging evidence suggests that the primate ventral visual pathway encodes increasingly complex stimulus features in downstream areas. We quantitatively show that there indeed exists an explicit gradient for feature complexity in the ventral pathway of the human brain. This was achieved by mapping thousands of stimulus features of increasing complexity across the cortical sheet using a deep neural network. Our approach also revealed a fine-grained functional specialization of downstream areas of the ventral stream. Furthermore, it allowed decoding of representations from human brain activity at an unsurpassed degree of accuracy, confirming the quality of the developed approach. Stimulus features that successfully explained neural responses indicate that population receptive fields were explicitly tuned for object categorization. This provides strong support for the hypothesis that object categorization is a guiding principle in the functional organization of the primate ventral stream.

Unsupervised feature learning improves prediction of human brain activity in response to natural images.

Encoding and decoding in functional magnetic resonance imaging has recently emerged as an area of research to noninvasively characterize the relationship between stimulus features and human brain activity. To overcome the challenge of formalizing what stimulus features should modulate single voxel responses, we introduce a general approach for making directly testable predictions of single voxel responses to statistically adapted representations of ecologically valid stimuli. These representations are learned from unlabeled data without supervision. Our approach is validated using a parsimonious computational model of (i) how early visual cortical representations are adapted to statistical regularities in natural images and (ii) how populations of these representations are pooled by single voxels. This computational model is used to predict single voxel responses to natural images and identify natural images from stimulus-evoked multiple voxel responses. We show that statistically adapted low-level sparse and invariant representations of natural images better span the space of early visual cortical representations and can be more effectively exploited in stimulus identification than hand-designed Gabor wavelets. Our results demonstrate the potential of our approach to better probe unknown cortical representations.

Gaussian mixture models and semantic gating improve reconstructions from human brain activity.

Better acquisition protocols and analysis techniques are making it possible to use fMRI to obtain highly detailed visualizations of brain processes. In particular we focus on the reconstruction of natural images from BOLD responses in visual cortex. We expand our linear Gaussian framework for percept decoding with Gaussian mixture models to better represent the prior distribution of natural images. Reconstruction of such images then boils down to probabilistic inference in a hybrid Bayesian network. In our set-up, different mixture components correspond to different character categories. Our framework can automatically infer higher-order semantic categories from lower-level brain areas. Furthermore, the framework can gate semantic information from higher-order brain areas to enforce the correct category during reconstruction. When categorical information is not available, we show that automatically learned clusters in the data give a similar improvement in reconstruction. The hybrid Bayesian network leads to highly accurate reconstructions in both supervised and unsupervised settings.