From Environmental Noise to Computational Degrees of Freedom: Adaptive Neuromorphic Devices

Conventional electronic devices have been developed with a primary focus on suppressing environmental perturbations to ensure stable and reproducible operation. External stimuli such as light, temperature, humidity, and gas have been regarded as undesirable sources of noise, leading to extensive efforts in material, structural, and interface engineering to minimize their impact. However, as emerging applications demand more intelligent, adaptive, and energy-efficient systems, particularly in edge artificial intelligence and in-sensor computing, this conventional design paradigm is undergoing a fundamental shift. Here, we present a comprehensive review of environment-adaptive electronic and neuromorphic devices, with an emphasis on a paradigm shift from environment-blind to environment-aware and ultimately to environment-native systems. We first examined conventional strategies that suppress environmental influences in both two-terminal and three-terminal devices to achieve operational stability. Then, we discussed recent advances in environment-adaptive devices, where environmental stimuli are intentionally exploited as active control parameters to tune device characteristics and enable multimodal sensing. We further explored environment-adaptive neuromorphic devices, in which environmental factors directly govern synaptic plasticity, enabling dynamic modulation of short-term and long-term plasticity, as well as synaptic weighting and learning behaviors. We also provided perspectives on future directions, including material design, device architectures, and system-level integration, to further advance environment-native computing paradigms.