The OHL pixels consist of yellow organic dye (Dyenamo yellow) sensitized mesoporous TiO2 photoanodes infiltrated by a liquid methoxypropionitrile electrolyte containing 0.1 M Tris(2-(1H-pyrazol-1-yl)-4-tert-butylpyridine)cobalt(II)di(bis(trifluoromethylsulfonyl)imide), or in short (FK209)2+\cite{theory}, as the reducing agent (Figure 1c). No oxidizing species were added in order to minimize the presence of recombination centers to which the TiO2 electrons would readily transfer, resulting in high recombination rates and hence potential loss of memory feature due to a rapid VOC decay\cite{theory,w2013,shuttles}. To probe the learning behaviour of this sensor, the VOC response of a single OHL pixel was measured when switching from light to dark at various intensities (Figure 2a). It is evident that the VOC does not rapidly drop to zero but decays at a slow rate, showing that learning can be achieved. More precisely, for all probed intensities, the VOC retains more than 90% of the initial value for a period of 10 s in the dark. To further investigate the importance of recombination rate on the learning behaviour, DSSC pixels were fabricated containing both (FK209)2+ as the reducing agent and its corresponding oxidizing species Tris(2-(1H-pyrazol-1-yl)-4-tert-butylpyridine)cobalt(III)tris(bis(trifluoromethylsulfonyl)imide), in short (FK209)3+, with the latter acting as additional recombination centers to enhance recombination. As Figure 2b reveals, this device’s VOC declines much faster upon switching from light to dark, clearly indicating a much shorter memory capacity. That is, this pixel loses more than 10% of its initial VOC value after a significantly shorter period of time (around 1 s) for all probed intensities. The rate of VOC exponential decay in the absence of light is a method for quantifying the recombination rate since, at open circuit in the dark, all the electrons residing in the TiO2 photoanode have to transfer back to the oxidizing species in the electrolyte to reach an energetically equilibrated state\cite{cells}. Using this open circuit voltage decay method, the TiO2 electron lifetimes τ (the reciprocal of charge recombination rate) with respect to VOC were obtained for both types of pixels (Figure 2d and Table 1)\cite{ivn2004,t2010,shuttles}. Evidently, the TiO2 electron lifetimes in the (FK209)2+ only pixel are much longer than in the device containing the oxidizing species (FK209)3+, proving that the period of memory retention is critically linked to the TiO2 charge recombination rate. However, one has to keep in mind that charge recombination can be dependent on the dye regeneration rate as well. In especially, when a reducing agent regenerates the dye less efficiently, fewer oxidizing shuttles will be formed that can pose as charge recombination centers. To verify this point, DSSC pixels containing 0.1 M Tris-(2,2'-bipyridine)cobalt(II)di(bis(trifluoromethane)sulfonimide), in short Co(II)(bpy)3, were fabricated since this cobalt complex yields faster regeneration of the Dyenamo Yellow dye than (FK209)2+ \cite{anders2011}. As illustrated in Figure 2d, the TiO2 electron lifetime is indeed shorter than the (FK209)2+ pixels, triggering a much faster VOC drop when transitioning from light to dark (Figure 2c as compared to Figure 2a). In fact, the Co(II)(bpy)3 based pixel exhibits similar VOC and TiO2 electron lifetimes as the (FK209)2+ & (FK209)3+ device (Figure 2d and Table 1), despite the lack of initial oxidizing species Co(III)(bpy)3 acting as additional recombination centers in the former cell. This observation implies that, due to the faster dye regeneration rate of Co(II)(bpy)3, the amount of the oxidizing shuttle Co(III)(bpy)3 being generated during illumination is comparable to the initial concentration of oxidizing species in the (FK209)2+ & (FK209)3+ device, ultimately leading to the observed similar recombination rates. It is to be noted that, for all three types of pixels discussed this far, the VOC decay in the dark is most retarded after exposure to low light intensities (1.6k Lux to 7.1k Lux in Figure 2a, b, and c). The reason lies in the reduced number of charge carriers injected into and residing in the TiO2 photoanode at low intensities, leading to diminished VOC and a decreased driving force for recombination\cite{gerrit2006,gerko2006}. Furthermore, fewer dyes are active at dimmer conditions, resulting in more suppressed dye regeneration which in turn curtails the number of oxidizing species being created. Consequently, less recombination centers are present, yielding prolonged TiO2 electron lifetimes (slower recombination rates) at lower VOC (Figure 2d).