Automatic binding of basic sensory features requires consciousness
Zhili Han, Hao Zhu, Qian Chu, Yuchunzi Wu, Xu Chen, Yuanqing Wang, Sixian Li, Xiangbin Teng, Patrick C. M. Wong, Chen Yao, Xing Tian
bioRxiv Preprint Server December 10, 2025 preprint DOI: 10.64898/2025.12.10.693567 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Experimental study with intracranial stereo-electroencephalography (sEEG) recordings in awake and anesthetized states |
|---|---|
| Population | Human participants with intracranial electrodes |
| Intervention | anesthesia |
| Key findings | In the awake state, auditory feature binding occurs automatically without attention in a localized sensory cortical network, while in the anesthetized state, binding is abolished despite preserved single-attribute encoding. |
Abstract
Conscious awareness requires establishing coherent percepts. Yet, whether consciousness is necessary for initiating the integration of basic sensory features remains unclear. Competing theories implicate distinct functional regimes of consciousness in the process of feature binding and creating conscious percepts. We used a novel multi-feature oddball paradigm with intracranial stereo-electroencephalography (sEEG) recordings in awake and anesthetized states to investigate the functional boundary of consciousness. In the awake state, the auditory attributes of loudness and tone, as well as the binding of the two features, were automatically encoded without attention to the stimuli in a functionally localized sensory cortical network. In the anesthetized state, the cortical registration of single attributes was preserved, whereas the binding was abolished. Moreover, anesthesia mostly influenced later cortical processes after stimuli offset. These results reveal the borderline of exertion of consciousness between the encoding and manipulation of basic sensory features in local cortical circuits – the functional boundary of consciousness constrains the feedforward binding and recurrent process directly at local rather than global level computations.