Silencing KCs within the α/β lobes of the MBs results in premature habit formation. Furthermore, formation of premature habits can be achieved by silencing MBON-02. Based on the connectome reconstruction, MBON-02 gets input from β2/β’2a KCs. They, in turn, form the little-studied lateral (lACA) and dorsal (dACA) accessory calyx regions of the MBs (Li et al., 2020). We will prove which KC subpopulations are sufficient and/or necessary to trigger the inhibition. As β2 and β’2 KCs receive different types of input and the signals converge at the level of MBON-02, MBON-02 can be assumed as a site of coincidence detection. For the first time, we will address the possibility that such coincidence detection within learning events takes place not at the level of the KCs but at the level of their downstream output neurons.
KCs within the accessory calyces receive different types of sensory input. The dorsal accessory calyx (dACA) consist of 60 α/βp KC. They receive visual input from the lobula and the medulla. Interestingly, visual input is mainly processed through the ventral accessory calyx (vACA), which was shown to be involved in specific Pavlovian color learning, but not in operant conditioning. Therefore, it is conceivable that visual predictors during the world-learning phase are processed by another subset of KCs. We aim to functionally dissect upstream neurons that convey visual information to the dACA and, by that, trigger the inhibition of self-learning at early stage. Next, we will investigate neurons upstream of the lateral accessory calyx (lACA). The α’/β’ap1 KCs of the lACA get their input from thermosensory projection neurons (PNs): cooling air-responsive PN, VP3+ vPN and predicted warming air-responsive VP2 adPN. VP2 adPN receives in turn input from the thermosensory Gr28b.d neurons at the arista (Li et al., 2020, Marin et al., 2020). As during operant learning in the flight simulator, flies are punished by a heat beam, we speculate that the thermosensory information is processed through the KCs of the lACA to the MBON-02 and, by that, coincides with the visual input from the dACA.
Further, the MBs are strongly innervated by dopaminergic neurons (DANs). Anatomically, DANs were identified, down- and upstream of KCs as well as down- and upstream of MBONs. MBON-02 shows connections two high number of DANs. However, two of them, PAM-09 and PAM-10, form a prominent feedback loop from the MBON-02 back to the KCs within the dACA, and PAM-09 targets KCs within the lACA.Further, these DANs receive input from the fan-shaped body and send projections to the SLP (Li et al. 2020), and by that might play a role in interaction of different type memories stored in different brain/VNC areas. Additionally, dopaminergic modulation enhances the signaling within the MB circuitry to stabilize the memory. We will study the overall effect of dopamine on the transition from goal-directed behavior to habits. Thus, we aim to identify a modulatory site of interaction between memory systems and potential regulators of the activity level based temporal control on the transition between the world- and the self-learning systems.
