If BC subtypes that feed RGCs were dominant in determining and shaping RGC response kinetics, then one would expect that all RGCs maintain comparable response transience when the same single BC type dominates their inputs. In contrast to photopic signals delivered by 5–6 bipolar cells to ON RGCs, low scotopic signals reach RGC targets mostly via the primary rod pathway that operates with the single rod BC type. Therefore, the same inputs are delivered to all RGC subtypes under scotopic conditions. In contrast, the unpublished data show that the scotopic RGC responses are just as diverse in terms of their kinetics as their photopic counterparts (Figure 3). This suggests that the observed kinetic variety of RGC responses is neither the result of the differential kinetics of the phototransduction cascade of rods and cones or the postsynaptic glutamate receptor on BC dendrites. In addition, different experiments demonstrated that outputs of a single BC can produce widely different RGC response patterns (including differences in kinetics), depending on the RGC type, and that this diversity is mediated by retinal ACs [97,98]. This suggests that irrespective of the kinetic nature of BC inputs, RGC response kinetics are likely reestablished in the inner retina both by passive and active membrane properties of RGCs. While circuit elements (mostly inhibitory) have been described in the previous section, there are a few more aspects to cover in this regard. These include the potential role of desensitization of postsynaptic glutamate receptors on RGC dendrites that may play a role in the generation of transient RGC responses [144], and also other voltage-gated currents that shape RGC responses [145].

Apart from RGC membrane properties, there is another crucial but often neglected factor that determines RGC response transience, namely the summation of all incoming inputs to the same RGC. One aspect of this issue, the feedforward inhibition from ACs to RGC, has already been discussed (see above in Section 3.3.2). However, RGCs may also integrate excitatory signals from more than one BC. Figure 3 provides one example, where an examined RGC clearly displays relative transient responses to both scotopic and photopic stimuli (small PSTHτ values, for the calculation of PSTHτ, refer to [146]), but the full-field light response of the same cell appeared more sustained (larger PSTHτ values) in the mesopic/low photopic range when both scotopic and photopic signaling streams were active. This is obviously the result of the summation of signals carried via a separate cohort of BCs serving at least two different signaling pathways. Under scotopic and photopic conditions, both the PSTH peak and τ values were influenced by only one signal component (the slow and sensitive component in scotopic conditions and the fast but less sensitive component in photopic conditions), whereas mid-range illumination brought about an intermingling of the fast signal component peak and a shoulder of the slow response component, thereby resulting in an increased PSTHτ value. The summation of these signals evidently caused the spiking pattern of this otherwise transient cell to appear more sustained. A similar signal summation should occur as well when two or more photopic signals are conveyed parallel to each other to the same RGC targets. In this scheme, a certain RGC subtype receives inputs from two or more signaling streams (BCs). This, in fact, is the case for ON alpha RGCs of the mouse retina that receive a mixture of inputs from type 6, 7, and 8 bipolar cells [147]. In addition, sustained OFF alpha cells are postsynaptic to type 2 bipolar and GluMI cells, whereas transient OFF alpha cells are targeted by type 3a and type 4 bipolar cells [148,149]. Therefore, the mixing of inputs originating from several parallel signaling streams seems to be a general feature for most RGCs in the mammalian retina. Furthermore, it appears that the summation of inputs with slightly different kinetics (delay and decay) and sensitivity plays a crucial role in determining the ultimate kinetics of RGC responses. Again, these results are not necessarily in conflict with other reports regarding mechanisms that determine BC response kinetics, however BC response characteristics are not simply inherited by postsynaptic RGCs but greatly altered via signal summation. The results of such signal summation can be exemplified for some RGC subtypes that have been studied more deeply, including the sustained and the transient OFF alpha cell populations. Sustained OFF alpha RGC responses are generated by the summation of excitatory inputs from the transient type 2 bipolar cells and GluMI cells [43,148,150], whereas the synaptic partners transient OFF alpha cells are the transient type 3a and sustained type 4 bipolar cells [149,150].

Clearly, rather than simply inheriting response kinetics of presynaptic BCs, these two RGC subtypes performed a transformation of incoming signals. One issue remains: while the summation of excitatory signals can provide an example of how transient inputs are transformed into intermediary or sustained signals by RGCs, the reverse (sustained-to-transient transformation) cannot be explained by the same mechanism. It is known that in addition to excitatory BC inputs to RGCs, some signaling routes also provide inhibitory signals via intermediary ACs. In fact, the sections above detailed how GABAergic inhibition may alter RGC response transience. In this scheme, the early onset of inhibition truncates the response of the RGC target, thus making it more sustained. GABAergic AC input alters RGC response kinetics and functionality, as a fast inhibition is known to be capable of truncating excitatory signals [33,151], whereas delayed inhibitory inputs can shift the signal towards the more transient domain of the spectrum (Figure 4) [49,121,152].