With the increasing levels of automation and intelligence in nuclear power plants, the cognitive processes of operators and their interactions with plant systems are undergoing significant transformations. This paper aims to explore how different levels of automation technology impact the cognitive processes of operators and provides a comprehensive analysis of the characteristics and monitoring information requirements of human–machine interfaces (HMIs) based on a cognitive model. The paper begins by reviewing the characteristics of various automation technologies and categorizing them into three levels of automation based on their complexity: simple automation, complex automation, and intelligent automation. It discusses the features of each automation level and their respective impacts on the cognitive processes of operators. Building on this foundation, the study integrates multiple cognitive theories to construct a cognitive model applicable to operators under automated conditions. This model divides the operator’s cognitive process into four stages: perception, analysis, decision-making, and prediction, each encompassing distinct cognitive functions. Based on this cognitive model, the paper analyzes the monitoring information requirement of operators. It classifies monitoring information elements according to the characteristics of each cognitive stage, providing references for HMI information display. Finally, combining the resources of the main control room’s HMI with the cognitive behaviors of operators, the paper maps HMI resources to each stage of the cognitive model from both “human” and “machine” perspectives. It proposes HMI characteristic requirements from a cognitive perspective, offering theoretical support for the design of HMIs in nuclear power plant main control rooms.