Dopamine signals reward in animal brains. A single presentation of a sugar reward to Drosophila activates distinct subsets of dopamine neurons that independently induce short- and long-term olfactory memories (STM and LTM, respectively). In this study, we show that a recurrent reward circuit underlies the formation and consolidation of LTM. This feedback circuit is composed of a single class of reward-signaling dopamine neurons (PAM-α1) projecting to a restricted region of the mushroom body (MB), and a specific MB output cell type, MBON-α1, whose dendrites arborize that same MB compartment. Both MBON-α1 and PAM-α1 neurons are required during the acquisition and consolidation of appetitive LTM. MBON-α1 additionally mediates the retrieval of LTM, which is dependent on the dopamine receptor signaling in the MB α/β neurons. Our results suggest that a reward signal transforms a nascent memory trace into a stable LTM using a feedback circuit at the cost of memory specificity. DOI: http://dx.doi.org/10.7554/eLife.10719.001, eLife digest An animal that finds particularly nutritious and palatable food will often develop a long-lasting memory—even if they experience that event only once. One example of this is the ability of the fruit fly Drosophila to form a long-term association between a sugar reward and a specific odor that was present when they received the reward. The consumption of sugar triggers the release of a chemical called dopamine on specific compartments of a brain structure called the mushroom body. Dopamine then acts to modify the connection between cells called “Kenyon cells”, which encode specific odors, and the neurons that send signals out from the mushroom body (called MBONs). The result is the formation of a memory that links the odor with the reward. However, little is known about how this process differs for long-term vs. short-term memories, and how it can occur when the fly has experienced the odor and reward together on only a single occasion. To find out, Ichinose et al. combined behavioral testing of fruit flies with genetics. The results confirmed that the dopamine neurons and the MBONs that project to a single compartment of the mushroom body, called α1, are both required for the formation of long-term odor-reward memories, but not their short-term equivalents. These neurons are called PAM-α1 and MBON-α1, respectively. Unexpectedly, anatomical data revealed that PAM-α1 dopamine neurons receive input from MBON-α1; that is, long-term memory formation involves a feedback circuit: from PAM-α1 to Kenyon cells, then to MBON-α1 and back to PAM-α1. Blocking feedback from the MBON-α1 onto the PAM-α1 neurons shortly after odor-reward training disrupted long-term memory formation. Conversely, blocking feedback at a later stage did not. This suggests that prolonged activation of PAM-α1 by MBON-α1 helps to strengthen newly established memories, converting them into memories that will last for a long time. The discovery of a specific circuit that supports long-term, but not short-term, memory formation in fruit flies is consistent with evidence of distinct mechanisms underlying these processes in mammals. Further work is now required to determine whether feedback circuits similar to those in fruit flies also contribute to reward-based learning in other animals. DOI: http://dx.doi.org/10.7554/eLife.10719.002