The　objective　of　this　study　is　to　examine　a　swarm　intelligence-based　learning　method　for　brain　effect　connected　networks.　Brain　effect　connected　networks　have　exhibited　approximation　properties　that　parallel　multi-layer　perception　properties.　The　brain　effect　connected　network　family　is　sufficiently　wide　to　be　able　to　uniformly　approximate　any　continuous　function　on　a　compact　set　and　perform　satisfactory　generalization　in　solving　different　classification　and　prediction　problems.　After　the　network　is　formulated　as　linear　systems,　the　system　equations　of　the　brain　effect　connected　networks　have　a　form　similar　to　that　of　the　RBFL　network,　and　both　networks　share　the　same　＂flat＂　architecture.　Thus,　the　ELM　algorithm　can　be　successfully　applied　to　brain　effect　connected　networks　similar　to　that　of　RBF.　Some　experimental　results　of　benchmarking,　as　well　of　real-world　approximation　and　classification　problems,　are　reported　to　show　the　superiority　of　the　presented　network　over　existing　brain　effect　connected　models　in　terms　of　generalization　performance.