The benefits of ultra-high fMRI are maximized at high spatial resolutions with less physiological noise contribution. However, increasing the temporal resolution is often avoided at such high spatial resolutions due to limitations in the temporal signal-to-noise-ratio (tSNR). We investigated the effects of varying spatial and temporal resolutions on fMRI sensitivity measures at 7T, and their implications on an fMRI application, Brain-Computer-Interface (BCI). We demonstrate overall BOLD sensitivity increased with temporal resolution despite the tSNR decrease specifically at high-spatial resolution 7T-fMRI. We further demonstrate that this sensitivity increase also improved the accuracy of linear classifiers commonly used in BCIs.