Direction-selective ganglion cells (DSGCs) have been studied since the 1960s, but the source of their selectivity was only more recently defined as coming from SACs. Now, the Arkansas group has followed the activity of SACs in retinal preparations. Light going out along a SAC dendrite centrifugally (away from the cell body) resulted in SAC activation; the activated SAC then inhibits the downstream DSGC. But light coming in along a SAC dendrite centripetally activated the SAC far less, so there is little or no inhibition of the DSGC.
The centripetal inhibition of a SAC started when light was up to 2 dendritic radii away. This evidence, plus the transmitters involved and the presence of reciprocal inhibition, suggest that the source must be the neighboring, overlapping SACs. The group suggest that, as light moves toward a given SAC (SAC0), it first encounters the neighboring SAC (SAC1). SAC1 sends an inhibitory signal to SAC0, so by the time the light reaches SAC0 it is inhibited and barely reacts to the excitatory light signal. Consistent with this, the group saw inhibition of SACs by centripetal light before they saw (lowered) excitation. By contrast, light moving centrifugally first excites SAC0, and by the time it reaches neighboring SAC2 the game is over; any inhibition from SAC2 to SAC0 is irrelevant (because it is late) and reduced (because SAC0 has already inhibited SAC2).
SAC dendrites radiate in all directions, but each dendrite is essentially an independent unit for computing directionality. Thus, a single cell can contribute to detection of light coming from all possible directions. DSGCs, pointed along SAC dendrites in one direction, are turned on only when the SACs are not sending them inhibitory signals, which is when light is moving centripetally along the SAC dendrite. The next question is how this precise architecture is established in the first place.