This study was conducted to compare the difference in User brain activity in response to the background types when constructing a metaverse for ecosystem education and proposes an effective method of implementing a metaverse for learning. The metaverse ecosystem was developed into three background types: [Type 1] realistic illustration, [Type 2] pixel graphic, and [Type 3] 2.5D graphic on the metaverse platform zep, modeled after Duung Wetland in Taean-gun, Chungcheongnam-do. fNIRS was used to neuroscientifically identify the difference in background types for middle and high school students, and one subject experienced the learning effect according to the pre-experienced background type in three types, and the background type was randomly presented to reduce the learning effect according to the order. After preprocessing the collected data, it was analyzed using 21 datasets that can fully present the measured values of all areas, and the results are as follows. First, in each background type, there was a decrease in activity in DLPFC and an increase in activity in OFC. Second, the activity of BA47 was shown in [Type 1] and [Type 2]. Third, there was a decrease in activity in the FP region in [Type 3]. Fourth, [Type 1] has higher activity in the right DLPFC region than [Type 2]. Fifth, [Type 1] has higher activity in the left DLPFC region and both FP regions than [Type 3]. Based on these results, we intend to present effective metaverse design guidelines based on brain activity.