Task-Optimized Convolutional Recurrent Networks Align with Tactile Processing in the Rodent Brain
This work addresses the challenge of developing effective tactile perception for embodied AI and neuroscience, providing quantitative insights into neural processing, but it is incremental as it builds on existing neural network architectures.
The study tackled the problem of improving tactile sensing in artificial systems by introducing an Encoder-Attender-Decoder framework, finding that convolutional recurrent neural networks (ConvRNNs) achieved superior tactile categorization and closely matched neural representations in rodent somatosensory cortex, with a linear relationship between performance and neural alignment.
Tactile sensing remains far less understood in neuroscience and less effective in artificial systems compared to more mature modalities such as vision and language. We bridge these gaps by introducing a novel Encoder-Attender-Decoder (EAD) framework to systematically explore the space of task-optimized temporal neural networks trained on realistic tactile input sequences from a customized rodent whisker-array simulator. We identify convolutional recurrent neural networks (ConvRNNs) as superior encoders to purely feedforward and state-space architectures for tactile categorization. Crucially, these ConvRNN-encoder-based EAD models achieve neural representations closely matching rodent somatosensory cortex, saturating the explainable neural variability and revealing a clear linear relationship between supervised categorization performance and neural alignment. Furthermore, contrastive self-supervised ConvRNN-encoder-based EADs, trained with tactile-specific augmentations, match supervised neural fits, serving as an ethologically-relevant, label-free proxy. For neuroscience, our findings highlight nonlinear recurrent processing as important for general-purpose tactile representations in somatosensory cortex, providing the first quantitative characterization of the underlying inductive biases in this system. For embodied AI, our results emphasize the importance of recurrent EAD architectures to handle realistic tactile inputs, along with tailored self-supervised learning methods for achieving robust tactile perception with the same type of sensors animals use to sense in unstructured environments.