Bioinspired multisensory neural network with crossmodal integration and recognition

TLDR

The human multisensory neural network integrates vision, touch, hearing, smell, and taste to enable high‑level cognitive functions such as cross‑modal integration, recognition, and imagination, providing accurate evaluation and comprehensive understanding of the multimodal world. This work reports a bioinspired multisensory neural network that integrates artificial optic, afferent, auditory, and simulated olfactory and gustatory sensory nerves. Using distributed multiple sensors and biomimetic hierarchical architectures, the system senses, processes, and memorizes multimodal information while fusing data at both hardware and software levels. Cross‑modal learning enables the network to recognize and imagine multimodal content—visualizing alphabet letters from handwritten input, integrating visual/smell/taste cues, and generating unseen images from auditory descriptions—showing promise for robotic sensing and perception.

Abstract

Abstract The integration and interaction of vision, touch, hearing, smell, and taste in the human multisensory neural network facilitate high-level cognitive functionalities, such as crossmodal integration, recognition, and imagination for accurate evaluation and comprehensive understanding of the multimodal world. Here, we report a bioinspired multisensory neural network that integrates artificial optic, afferent, auditory, and simulated olfactory and gustatory sensory nerves. With distributed multiple sensors and biomimetic hierarchical architectures, our system can not only sense, process, and memorize multimodal information, but also fuse multisensory data at hardware and software level. Using crossmodal learning, the system is capable of crossmodally recognizing and imagining multimodal information, such as visualizing alphabet letters upon handwritten input, recognizing multimodal visual/smell/taste information or imagining a never-seen picture when hearing its description. Our multisensory neural network provides a promising approach towards robotic sensing and perception.