The fly eye contains different subtypes of unit eyes (ommatidia) with molecularly and morphologically specialized photoreceptors for comparing either between different wavelengths (color vision) or between different angles of the linearly polarized skylight (polarization vision). However, microcircuit differences between those parts of the columnar medulla neuropil computing color versus polarization remain largely unknown. There is virtually nothing known about the circuit elements immediately downstream of polarization-sensitive photoreceptors in the ‘dorsal rim area’ (DRA). In this work, I described the cellular and synaptic architecture of medulla columns that receive skylight polarization input from DRA photoreceptors. I showed that only in the DRA region, R7 and R8 photoreceptors resemble each other by targeting their axons to the same medulla layer. However, within this layer DRA R7 and R8 connect to morphologically distinct Dm target cells (called Dm-DRA1 and Dm-DRA2, respectively). Both Dm-DRA cell types are modality-specific by avoiding contact with color-sensitive photoreceptors. Using the genetic toolbox of Drosophila such as activity-dependent GFP-reconstitution across synaptic partners (GRASP) and the genetically inducible trans-synaptic tracer ‘trans-Tango’, I confirmed that Dm-DRA1 and Dm-DRA2 are the specific post-synaptic targets of DRA.R7 or DRA.R8, respectively. Neither Dm-DRAs overlap with the main synaptic targets of color-sensitive R7 cells (called Dm8 cells), revealing for the first time that skylight polarization is processed by separate modality-specific circuits in the early visual system. These modality-specific differences are not limited only Dm-DRA cells. I described modality-specific cellular and synaptic specializations in other optic lobe cell types in the DRA region of the medulla: the dendritic arbors of certain cell types (neuromodulatory cells and visual projection neurons) specifically avoid the DRA region. Furthermore, Transmedullary (Tm) cells that are post-synaptic to color-sensitive photoreceptors showed modality-specific differences in connectivity or were absent from the DRA. Finally, I contributed a study describing the cellular organization of the ‘anterior visual pathway’ that carries skylight information from the eye to the central brain. In this study, I showed that an optic glomerulus called the anterior optic tubercle (AOTU) receives direct information via different classes of medulla-to-tubercle (MeTu) neurons, terminating in different subdomains of the AOTU. Finally, we hypothesize that different classes of MeTu cells carry different types of skylight information to the central brain via parallel pathways.