(A) A graphic representation of the olfactory pathways in male moth brain. The morphological features of MGC-PNs is displayed on the hemisphere to the left. Multiglomerular medial-tract MGC-PNs (gray) and uniglomerular PNs arborizing in the dmp (magenta) or dma (blue) project to the calyces (Ca), anteroventral LH, and VLP, while the uniglomerular PNs with dendrites in the cumulus (Cu; black) target the Ca, VLP, SLP, and posterior SIP. The multiglomerular mediolateral-tract PNs (dashed green lines) target primarily the VLP, whereas the uniglomerular lateral-tract PNs innervating the cumulus run directly to the column in the anterior SIP (Chu et al., 2020a). In the hemisphere to the right, a plethora of different PN types arborizing in the ordinary glomeruli (OG) and innervating several lateral protocerebral neuropils are indicated (Ian et al., 2016; Kymre et al., 2021). (B) Scheme illustrating projections of MGC-PNs versus OG-PNs in the higher brain regions. The three main antennal lobe tracts (ALTs) are illustrated: Solid line, medial antennal-lobe tract (mALT); dashed line, mediolateral ALT (mlALT); double line, lateral ALT (lALT). (C) An overview of temporal response properties implied by the mean firing rate during odor stimulation (MZIFR; mean Z-scored instantaneous firing rate) of the assembly of PNs projecting to the same neuropil. The PNs included in this framework are shown in (E). Note that these PNs were sampled from the antennal lobe, while the length of axons and action potential transmission rates have not been included, the actual timing of synaptic output onto the protocerebral neuropils is thereby not represented. (D) Correspondence between the proportion of sampled neurons originating from each of the MGC units and that of the volume of each MGC unit (volume data from Zhao et al., 2016) shows that our sampling strategy for the proposed framework (C) reflects the composition of the actual MGC-PN population. (E) Summary of recorded and labelled PNs’ morphologies across the three main tracts, including 10 lALT MGC-PNs from our previous study (Chu et al., 2020a). (F) Computational experiment with shuffled electrophysiological data indicating correlation between PN responses to attractive versus non-attractive stimuli and separated output areas. The data shuffling was repeated five times, and each time two-thirds of the PNs projecting to the same neuropil were randomly recruited into the data assembly. The cross correlations between each of the data assemblies showed that neuronal response profiles were consistent with the neuropils associated with different behavioral valences. (G) Cross-correlations of the firing traces between every two PNs. The neuron types are indicated by the color-coded boxes, in correspondence with (E). AL, antennal lobe; Ca, calyces; CB, central body; LH, lateral horn; OG, ordinary glomeruli; SIP, superior intermediate protocerebrum; SLP, superior lateral protocerebrum; VLP, ventrolateral protocerebrum. A, anterior; L, lateral; P, posterior.