In real society, interactions between individuals are accompanied by emotional changes, which in turn affect their decisions. This paper proposes an emotion-strategy persistence mechanism that represents individuals’ value orientations through changes in emotions. Intelligent agents are classified into selfish, competitive individuals and mutually beneficial, non-competitive individuals. A positive-negative emotion threshold is introduced to define the emotional states of intelligent agents. A positive emotional state indicates that the intelligent agent is relatively satisfied with the current strategy, leading to persistence in maintaining it. Conversely, the agent is more inclined to change its current strategy. In addition, this paper introduces the concept of emotional sensitivity to measure the amplitude of emotional fluctuations in intelligent agents and introduces strategy duration sensitivity to measure the willingness of intelligent agents to change their current strategy. Extensive simulation experiments show that non-competitive individuals play a decisive role in promoting and emerging cooperative behavior within the group. When the positive-negative emotion threshold is higher and the emotional sensitivity is greater, the level of group cooperation is higher. However, the impact of strategy duration sensitivity on the group cooperation rate exhibits a non-monotonic trend, with an optimal strategy duration sensitivity that maximizes the level of group cooperation.