, 2002) and transmit the major input signals to the motion detection circuitry ( Rister et al., 2007). In both neurons, onset and offset of histamine release cause transient hyperpolarizing and depolarizing dendritic responses, trans-isomer cost respectively, with a small sustained hyperpolarization

in between ( Laughlin and Hardie, 1978 and Laughlin et al., 1987). L1 and L2 relay their signals via long axons to separate layers in the second-order neuropil, the medulla. Here, information is picked up by mostly unidentified neurons that constitute the motion detection circuit and finally transmit their output to the third-order neuropil consisting of lobula and lobula plate. In the lobula plate, large directionally selective tangential cells extend their elaborate dendrites and spatially integrate see more the output of local presynaptic motion detectors ( Single and Borst, 1998 and Borst et al., 2010). Their responses to large-field motion in the preferred direction (PD) are positive (membrane depolarizations, or firing rate increases) and negative (hyperpolarizations, or firing rate decreases) in the

opposite, the so-called null direction (ND). In this study, we build on the recent discovery that the lamina neurons L1 and L2 constitute the input channels to the motion detection circuitry in Drosophila. Joesch et al. (2010) recorded from directionally selective tangential cells in the lobula plate while genetically blocking synaptic transmission from L1 and/or L2. Blocking both L1 and L2 removed motion-sensitive responses in lobula plate tangential cells. Importantly, blocking either L1 or L2 revealed that in flies, similar to vertebrates, the visual input is split into an ON and an OFF component. Here, we adapt the Reichardt Detector to incorporate these new findings, giving rise to two alternative models. Both models require a more elaborate internal structure of the detector to allow for an implementation of separate ON- and OFF-input signals. The first model, the “4-Quadrant-Detector” (Figure 1B) (Hassenstein and Reichardt, 1956) consists of four parallel detectors that cover all four possible combinations of input signals (ON-ON, ON-OFF, OFF-ON, and OFF-OFF). From

its input-output behavior, a 4-Quadrant-Detector mafosfamide is mathematically identical to the original Reichardt model. The second model, proposed by Franceschini et al. (1989), contains just two subunits, an ON-ON and an OFF-OFF detector (Figure 1C). Notably, this “2-Quadrant-Detector” is no longer equivalent to the original Reichardt Detector since input signals of opposite sign do not interact. These differences in response behavior should allow us to decide between the two models experimentally. We first presented apparent motion stimuli consisting of sequences of spatially displaced, persistent light increment (ON) and decrement (OFF) steps to two different fly species, Calliphora and Drosophila, while recording from lobula plate tangential cells.