Realizing the full potential of optogenetic techniques will require concurrent development of light delivery technologies that improve spatial and temporal control of neuronal stimulation to facilitate increasingly informative electrophysiology recordings. Excitation (or inhibition) by light-activation of channelrhodopsin (ChR) and its many variants has quickly emerged as a preferred methodology in the neurosciences by reason of the cellular specificity achievable through conditional genetic expression. However, despite the variety of cell types targetable through the use of promoter-driven constructs, the spatial distribution of cells of any given classification can still be quite homogeneous, a fact that complicates the targeting of neural circuits that exhibit a very specific structure-function relationship, as in the circuits underlying functional maps. In studying the emergence of cortical maps in vivo where response variations are closely tied to spatial locations on the order of a few hundred microns in width, new light delivery techniques are needed to improve on the dispersive, on/off light control of implanted fiber optics that are commonly used to drive ChR stimulation. Two specific advances would enhance the utility of optogenetic stimulation in such applications: 1.) use of a controllable light source capable of projecting both stationary and dynamic light patterns; and 2.) improved spatial and temporal resolution while sustaining power sufficient to drive ChR activation at depth in the intact living brain. We have developed an optogenetic stimulator based on epi-illumination microscope design and equipped with a high numerical aperture objective to stimulate and collect images through the same optical axis. The stimulator is coupled to an LCD-MLA projector light source to produce high-resolution spatiotemporally varying patterns in a very confined area. We evaluate the performance of this stimulator with respect to power, optical resolution and the quality of spatial and temporal control of single and multi-unit responses at a multichannel NeuroNexus electrode. We further assay the stimulator’s suitability for experimental paradigms calling for fine sequential control of horizontal cortical connections in vivo. Preliminary results indicate that retinotopically-constrained horizontal activation through ChR-mediated surface training in ferret V1 is sufficient to drive asymmetry in the orientation-tuning response to drifting gratings.