Prior work has offered evidence for functional localization in the brain; different anatomical regions preferentially activate for certain types of visual input. For example, the fusiform face area preferentially activates for visual stimuli that include a face. However, the spectrum of visual semantics is extensive, and only a few semantically-tuned patches of cortex have so far been identified in the human brain. Using a multimodal (natural language and image) neural network architecture (CLIP, \cite{CLIP}, we train a highly accurate contrastive model that maps brain responses during naturalistic image viewing to CLIP embeddings. We then use a novel adaptation of the DBSCAN clustering algorithm to cluster the parameters of these participant-specific contrastive models. This reveals what we call Shared Decodable Concepts (SDCs): clusters in CLIP space that are decodable from common sets of voxels across multiple participants. Examining the images most and least associated with each SDC cluster gives us additional insight into the semantic properties of each SDC. We note SDCs for previously reported visual features (e.g. orientation tuning in early visual cortex) as well as visual semantic concepts such as faces, places and bodies. In cases where our method finds multiple clusters for a visuo-semantic concept, the least associated images allow us to dissociate between confounding factors. For example, we discovered two clusters of food images, one driven by color, the other by shape. We also uncover previously unreported areas with visuo-semantic sensitivity such as regions of extrastriate body area (EBA) tuned for legs/hands and sensitivity to numerosity in right intraparietal sulcus, sensitivity associated with visual perspective (close/far) and more. Thus, our contrastive-learning methodology better characterizes new and existing visuo-semantic representations in the brain by leveraging multimodal neural network representations and a novel adaptation of clustering algorithms.

Neural network pruning is a popular approach to reducing the computational costs of training and/or deploying a network and aims to do so while minimizing accuracy loss. Pruning methods that remove individual weights (fine granularity) can remove more total network parameters before reaching a given degree of accuracy loss, while methods that preserve some or all of a network's structure (coarser granularity, such as pruning channels from a CNN) take better advantage of hardware and software optimized for dense matrix computations. We compare intelligent iterative pruning using several different criteria sampled from the literature against random pruning at initialization across multiple granularities on two different architectures and three image classification tasks. Our work is the first direct and comprehensive investigation of the relationship between granularity and the efficacy of intelligent pruning relative to a random-pruning baseline. We find that the accuracy advantage of intelligent over random pruning decreases dramatically as granularity becomes coarser, with minimal advantage for intelligent pruning at granularity coarse enough to fully preserve network structure. For instance, at pruning rates where random pruning leaves ResNet-20 at 85.0% test accuracy on CIFAR-10 after 30,000 training iterations, intelligent weight pruning with the best-in-context criterion leaves it at about 90.0% accuracy (on par with the unpruned network), kernel pruning leaves it at about 86.5%, and channel pruning leaves it at about 85.5%. Our results suggest that compared to coarse pruning, fine pruning combined with efficient implementation of the resulting networks is a more promising direction for easing the trade-off between high accuracy and low computational cost.

Working memory requires information about external stimuli to be represented in the brain even after those stimuli go away. This information is encoded in the activities of neurons, and neural activities change over timescales of tens of milliseconds. Information in working memory, however, is retained for tens of seconds, suggesting the question of how time-varying neural activities maintain stable representations. Prior work shows that, if the neural dynamics are in the 'null space' of the representation - so that changes to neural activity do not affect the downstream read-out of stimulus information - then information can be retained for periods much longer than the time-scale of individual-neuronal activities. The prior work, however, requires precisely constructed synaptic connectivity matrices, without explaining how this would arise in a biological neural network. To identify mechanisms through which biological networks can self-organize to learn memory function, we derived biologically plausible synaptic plasticity rules that dynamically modify the connectivity matrix to enable information storing. Networks implementing this plasticity rule can successfully learn to form memory representations even if only 10% of the synapses are plastic, they are robust to synaptic noise, and they can represent information about multiple stimuli.

Sensory neurons give highly variable responses to stimulation, which can limit the amount of stimulus information available to downstream circuits. Much work has investigated the factors that affect the amount of information encoded in these population responses, leading to insights about the role of covariability among neurons, tuning curve shape, etc. However, the informativeness of neural responses is not the only relevant feature of population codes; of potentially equal importance is how robustly that information propagates to downstream structures. For instance, to quantify the retina's performance, one must consider not only the informativeness of the optic nerve responses, but also the amount of information that survives the spike-generating nonlinearity and noise corruption in the next stage of processing, the lateral geniculate nucleus. Our study identifies the set of covariance structures for the upstream cells that optimize the ability of information to propagate through noisy, nonlinear circuits. Within this optimal family are covariances with "differential correlations", which are known to reduce the information encoded in neural population activities. Thus, covariance structures that maximize information in neural population codes, and those that maximize the ability of this information to propagate, can be very different. Moreover, redundancy is neither necessary nor sufficient to make population codes robust against corruption by noise: redundant codes can be very fragile, and synergistic codes can-in some cases-optimize robustness against noise.

Over repeat presentations of the same stimulus, sensory neurons show variable responses. This "noise" is typically correlated between pairs of cells, and a question with rich history in neuroscience is how these noise correlations impact the population's ability to encode the stimulus. Here, we consider a very general setting for population coding, investigating how information varies as a function of noise correlations, with all other aspects of the problem - neural tuning curves, etc. - held fixed. This work yields unifying insights into the role of noise correlations. These are summarized in the form of theorems, and illustrated with numerical examples involving neurons with diverse tuning curves. Our main contributions are as follows. (1) We generalize previous results to prove a sign rule (SR) - if noise correlations between pairs of neurons have opposite signs vs. their signal correlations, then coding performance will improve compared to the independent case. This holds for three different metrics of coding performance, and for arbitrary tuning curves and levels of heterogeneity. This generality is true for our other results as well. (2) As also pointed out in the literature, the SR does not provide a necessary condition for good coding. We show that a diverse set of correlation structures can improve coding. Many of these violate the SR, as do experimentally observed correlations. There is structure to this diversity: we prove that the optimal correlation structures must lie on boundaries of the possible set of noise correlations. (3) We provide a novel set of necessary and sufficient conditions, under which the coding performance (in the presence of noise) will be as good as it would be if there were no noise present at all.

The sparse coding hypothesis has enjoyed much success in predicting response properties of simple cells in primary visual cortex (V1) based solely on the statistics of natural scenes. In typical sparse coding models, model neuron activities and receptive fields are optimized to accurately represent input stimuli using the least amount of neural activity. As these networks develop to represent a given class of stimulus, the receptive fields are refined so that they capture the most important stimulus features. Intuitively, this is expected to result in sparser network activity over time. Recent experiments, however, show that stimulus-evoked activity in ferret V1 becomes less sparse during development, presenting an apparent challenge to the sparse coding hypothesis. Here we demonstrate that some sparse coding models, such as those employing homeostatic mechanisms on neural firing rates, can exhibit decreasing sparseness during learning, while still achieving good agreement with mature V1 receptive field shapes and a reasonably sparse mature network state. We conclude that observed developmental trends do not rule out sparseness as a principle of neural coding per se: a mature network can perform sparse coding even if sparseness decreases somewhat during development. To make comparisons between model and physiological receptive fields, we introduce a new nonparametric method for comparing receptive field shapes using image registration techniques.

Sparse coding algorithms trained on natural images can accurately predict the features that excite visual cortical neurons, but it is not known whether such codes can be learned using biologically realistic plasticity rules. We have developed a biophysically motivated spiking network, relying solely on synaptically local information, that can predict the full diversity of V1 simple cell receptive field shapes when trained on natural images. This represents the first demonstration that sparse coding principles, operating within the constraints imposed by cortical architecture, can successfully reproduce these receptive fields. We further prove, mathematically, that sparseness and decorrelation are the key ingredients that allow for synaptically local plasticity rules to optimize a cooperative, linear generative image model formed by the neural representation. Finally, we discuss several interesting emergent properties of our network, with the intent of bridging the gap between theoretical and experimental studies of visual cortex.