Bin-packing problem (BPP) is a typical combinatorial optimization problem whose decision-making process is NP-hard. This article examines BPPs in varying environments, where random number and shape of items are to be packed in different instances. The objective is to find a unified model to derive optimal decision process that maximizes the utilization of bins. To this end, by mimicking the experience-based reasoning process of humans, this article proposes a novel brain-inspired experience reinforcement model, which takes advantage of both biological and engineering systems. By learning experience from similar situations, the model is adaptive, such as the human brain for sophisticated scenarios and varying environments. The proposed model mimics the functional coordination among brain regions by knowledge representation and knowledge extraction modules. The former one corresponds to the part of information processing and experience storage. The latter one includes two parts that can train reasoning strategies and improve the decision performance. The proposed model is applied to instances of random number and shape of items of BPP. The obtained results outperform the state-of-the-art methods for BPPs in varying environments.