It has been known for several decades that defects are largely responsible for the catalytically active sites on metal and semiconductor surfaces. However, it is difficult to directly probe these active sites because the defects associated with them are often relatively rare with respect to the stoichiometric crystalline surface. In the work presented here, we demonstrate a method to selectively probe defect-mediated photocatalysis, through differential AC photocurrent (PC) measurements. In this approach, electrons are photoexcited from the valence band to a relatively narrow distribution of sub-bandgap states in the TiO 2 , and then subsequently to the ions in solution. Because of their limited number, these defect states fill up quickly resulting in Pauli blocking, and are thereby undetectable under DC or CW excitation. In the method demonstrated here, the incident light is modulated with an optical chopper while the photocurrent is measured with a lock-in amplifier. Thin (5nm) films of TiO 2 deposited by atomic layer deposition (ALD) on various metal films, including Au, Cu, and Al, exhibit the same wavelength-dependent photocurrent spectra, with a broad peak centered around 2.0eV corresponding to the band-to-defect transition associated with the hydrogen evolution reaction (HER). While the UV-vis absorption spectra of these films show no features at 2.0eV, photoluminescence (PL) spectra of these photoelectrodes show a similar wavelength dependence with a peak around 2.0eV, corresponding to the sub-band gap emission associated with these defect sites. As a control, alumina (Al 2 O 3) films exhibit no PL or PC over the visible wavelength range. The AC photocurrent plotted as a function of electrode potential, shows a peak around-0.4 to-0.1V vs. NHE, as the monoenergetic defect states are tuned through a resonance with the HER potential. This approach enables the direct photo-excitation of catalytically active defect sites to be studied selectively without the interference of the continuum interband transitions or 3 the effects of Pauli blocking, which is limited by the slow turnover time of the catalytically active sites, typically on the order of 1 µsec. We believe this general approach provides an important new way to study the role of defects in catalysis in an area where selective spectroscopic studies of these are few.