Dithering is the process of intentionally adding artificially generated noise to an otherwise uncorrupted signal to actually improve the performance of an end overall system. This article demonstrates that a dithering procedure can be used to improve the performance of an EEG interictal spike detection algorithm. Using a previously reported algorithm, by adding varying amounts of artificially generated noise to the input EEG signals the effect on the algorithm detection performance is investigated. A new stochastic resonance result is found whereby the spike detection performance improves by up to 4.3% when small amounts of corrupting noise, below 20 $\mu$VRMS, are added to the input data. This result is of use for improving the detection performance of algorithms, and the result also affects the dynamic range required for the hardware implementation of such algorithms in low power, portable EEG systems.