Currently, many newly proposed evolutionary algorithms have been successfully employed in the field of feature selection with the aim of finding a single optimal feature subset for a given dataset. However, due to the complexity of real-world data, the optimal feature subset in a dataset is often not unique. To represent data information with a single feature subset will be biased, which has made multimodal feature selection become a research hotspot. Most of those proposed algorithms could not deal with multimodal feature selection or were easily trapped into the local optima. Therefore, inspired by the nature, this paper investigates a new evolutionary algorithm to deal with multimodal feature selection. We proposes to handle multimodal feature selection by utilizing the hybrid breeding optimization algorithm (HBO), which is a novel evolutionary algorithm derived from Heterosis theory and incorporated with dynamic niching technology. Furthermore, to improve the performance of the traditional HBO, neighborhood search and elite mutation strategies are introduced in the global search, and a neighborhood crossover strategy is applied to broaden the diversity of population. Experimental results demonstrate the effectiveness of the proposed method, which outperforms some latest algorithms such as grey wolf optimizer algorithm, whale optimization algorithm and Harris hawks optimization algorithm etc. for multimodal feature selection.