Convolutional neural networks (CNNs) have been shown to be state-of-the-art models for visual cortical neurons. Cortical neurons in the primary visual cortex are sensitive to contextual information mediated by extensive horizontal and feedback connections. Standard CNNs integrate global contextual information to model contextual modulation via two mechanisms: successive convolutions and a fully connected readout layer. In this paper, we find that self-attention (SA), an implementation of non-local network mechanisms, can improve neural response predictions over parameter-matched CNNs in two key metrics: tuning curve correlation and peak tuning. We introduce peak tuning as a metric to evaluate a model's ability to capture a neuron's top feature preference. We factorize networks to assess each context mechanism, revealing that information in the local receptive field is most important for modeling overall tuning, but surround information is critically necessary for characterizing the tuning peak. We find that self-attention can replace posterior spatial-integration convolutions when learned incrementally, and is further enhanced in the presence of a fully connected readout layer, suggesting that the two context mechanisms are complementary. Finally, we find that decomposing receptive field learning and contextual modulation learning in an incremental manner may be an effective and robust mechanism for learning surround-center interactions.

Images of visual scenes comprise essential features important for visual cognition of the brain. The complexity of visual features lies at different levels, from simple artificial patterns to natural images with different scenes. It has been a focus of using stimulus images to predict neural responses. However, it remains unclear how to extract features from neuronal responses. Here we address this question by leveraging two-photon calcium neural data recorded from the visual cortex of awake macaque monkeys. With stimuli including various categories of artificial patterns and diverse scenes of natural images, we employed a deep neural network decoder inspired by image segmentation technique. Consistent with the notation of sparse coding for natural images, a few neurons with stronger responses dominated the decoding performance, whereas decoding of ar tificial patterns needs a large number of neurons. When natural images using the model pretrained on artificial patterns are decoded, salient features of natural scenes can be extracted, as well as the conventional category information. Altogether, our results give a new perspective on studying neural encoding principles using reverse-engineering decoding strategies.

System identification techniques-projection pursuit regression models (PPRs) and convolutional neural networks (CNNs)-provide state-of-the-art performance in predicting visual cortical neurons' responses to arbitrary input stimuli. However, the constituent kernels recovered by these methods are often noisy and lack coherent structure, making it difficult to understand the underlying component features of a neuron's receptive field. In this paper, we show that using a dictionary of diverse kernels with complex shapes learned from natural scenes based on efficient coding theory, as the front-end for PPRs and CNNs can improve their performance in neuronal response prediction as well as algorithmic data efficiency and convergence speed. Extensive experimental results also indicate that these sparse-code kernels provide important information on the component features of a neuron's receptive field. In addition, we find that models with the complex-shaped sparse code front-end are significantly better than models with a standard orientation-selective Gabor filter front-end for modeling V1 neurons that have been found to exhibit complex pattern selectivity. We show that the relative performance difference due to these two front-ends can be used to produce a sensitive metric for detecting complex selectivity in V1 neurons.