Adaptation maximizes information and minimizes dissipation across biological scales

Biological and living systems possess the inherent capacity to process information across spatiotemporal scales, displaying the hallmark ability to constantly adapt their responses to ever-changing and complex environments. In this work, we show that adaptation is a fundamental mechanism that spontaneously emerges when the system seeks to both maximize its information on external environments and minimize the unavoidable dissipation of its receptors. Adaptive responses arise in a narrow optimal regime characterized by such information-dissipation trade-off and a maximal information gain over time. Surprisingly, in systems close to optimality, the dynamics of adaptation reveals that the information gain is necessarily tangled with a reduction of the entropy produced by microscopic processes and an increase in the efficacy of feedback mechanisms. Moreover, we demonstrate that adaptation necessarily requires the simultaneous presence of negative feedback and a slow mechanism of information storage, while being independent of biological details. Finally, we employ our framework to investigate large-scale neural adaptation in zebrafish larvae subjected to repeated visual stimulation, paving the way for the understanding of the essential mechanisms that underlie adaptive behaviors and information processing in biological and living systems.