The onset of a stimulus, whether sensory or non-sensory, quenches trial-to-trial neural variability. The universality of this phenomenon across brain areas and species implies that it is a fundamental strategy employed by the brain to boost the fidelity of stimulus representation. Multiple network models have attempted to explain this phenomenon via circuit interactions, but a recent alternative theory proposes a much simpler cell-intrinsic mechanism: stimulus onset increases membrane conductance which itself is enough to quench output variability. Although this model can explain all of the physiological data, experimentally testing this hypothesis in vivo is currently impossible---requiring a means to artificially increase the membrane conductance of a single neuron while also minimizing any coincident network effects. One way to achieve this would be using bidirectional, single-cell resolution optogenetic manipulation to inject a balanced combination of excitatory and inhibitory conductances; however, no such technology exists. To address this technological gap, we developed a novel multi-color two-photon (2p) holographic microscope with single cell resolution optogenetic manipulation. We paired this novel optical platform with the ‘somBiPOLES’ construct, which fuses two spectrally separable excitatory and inhibitory opsins: Chrimson and GtACR2. First, we demonstrate that this platform enables flexible bidirectional control of neural activity with single-neuron resolution. Next, we optogenetically inject large balanced excitatory and inhibitory conductances while keeping mean activity constant, demonstrating the fine-scale control of this system. Finally, we show that increasing total conductance to a neuron is entirely sufficient to reduce its trial-by-trail variability in the absence of network effect, strongly supporting a cell-intrinsic model for stimulus-induced quenching of neural variability. These results are the first demonstration of an in vivo bidirectional holographic microscope, and demonstrate its ability to elucidate the mechanism of a fundamental coding principle of neurons.