BGAM is acquired through DANs, and blocking DANs rescued the memory deficits caused by BGAM. (A) BGAM was completely diminished by blocking the synaptic output from DANs during the odor 2 presentation (Kruskal–Wallis, n = 8, 8, 10, 8). Blocking DANs did not cause any difference as compared to other controls. (B) Blocking synaptic outputs from DANs during the CS− presentation rescued the memory impairment caused by BGAM (Kruskal–Wallis, n = 9, 8, 9, 8). (C) Blocking synaptic outputs from DANs during the CS+ presentation impaired STM, regardless of the MBON‐γ1pedc output (ANOVA, n = 9, 10, 10, 7). (D) Activating MBON‐γ1pedc during CS+ presentation did not impair STM (ANOVA, n = 7, 7, 8). (A–D) All bar graphs are mean ± SEM, and dots represent individual trials. *: P < 0.05, ***: P < 0.001, n.s.: P > 0.05.