Animals benefit from a wide variety of sensory information while controlling their behavior. This seemingly simple, inherent process is one of the most challenging problems in neuroscience as animals regulate their behavior in closed loop. Thus, examining multisensory integration (MSI) necessitates the use of control-theoretic approaches and custom experimental systems that may shed light on their underlying mechanisms. Here, we built a custom refuge-tracking system for Eigenmannia virescens, a species of weakly electric fish. These fish hide within refuges and track their movement to remain hidden inside. To achieve this, Eigenmannia combines visual and electrosensory signals perceived from the environment. Classically, the movements of the refuge concomitantly generate electrosensory and visual cues, which makes it impractical to distinguish their contributions to the behavior. As a remedy, we built a nested refuge structure; an outer opaque refuge placed on top of an inner transparent refuge. In this design, the movements of the inner refuge generate electrosensory cues while the movements of the outer refuge create the visual cues. This independent stimulation scheme allows us to generate various sensory conflict scenarios that may reveal the contribution of these two sensory systems on the behavior. We experimented with N=3 under different sensory conditions and fit various models to explain how MSI occurs in the brain. Our results suggest that the classical minimum-variance models are not sufficient to explain MSI when there is a sensory conflict. In the future, we plan to adopt control-theoretic models to capture the characteristics of MSI in these fish.