Feedback is crucial to motion perception in animals' visual systems where its spatial and temporal dynamics are often shaped by movement patterns of surrounding environments. However, such spatio-temporal feedback has not been deeply explored in designing neural networks to detect small moving targets that cover only one or a few pixels in image while presenting extremely limited visual features. In this paper, we address small target motion detection problem by developing a visual system with spatio-temporal feedback loop, and further reveal its important roles in suppressing false positive background movement while enhancing network responses to small targets. Specifically, the proposed visual system is composed of two complementary subnetworks. The first subnetwork is designed to extract spatial and temporal motion patterns of cluttered backgrounds by neuronal ensemble coding. The second subnetwork is developed to capture small target motion information and integrate the spatio-temporal feedback signal from the first subnetwork to inhibit background false positives. Experimental results demonstrate that the proposed spatio-temporal feedback visual system is more competitive than existing methods in discriminating small moving targets from complex dynamic environment.