We aim to improve stimulation sequences used to induce steady-state visual evoked potentials (SSVEPs) in brain-computer interfaces: The user fixates a flickering stimulus with the phase of the flicker coding the to-be-transmitted information. Using electroencephalography (EEG), SSVEPs are recorded and the phase of the stimulus is decoded from the EEG signal by a machine-learning classifier. We use high-frequency flicker (above the flicker-fusion threshold) to alleviate undesirable side effects such as visual fatigue, or headaches. Nevertheless, when phase transitions are introduced these are perceived as flashes which still can distract the user. Therefore, we reduced the perceptibility of phase transitions by, for example, gradually converting a flickering black/white stimulus to the perceived mid-gray level, then shifting the phase by repeating the perceived mid-gray level stimulus accordingly, and then gradually converting back to the full black/white contrast. This attenuated the detection rate of phase transitions to acceptable levels (62.5% accuracy). Here, we will test whether with such improved stimuli it is still possible to decode the phases of the flicker and thereby transmit information. Participants (N=24) will perform a two-alternative forced choice discrimination task trying to detect phase transitions. In parallel we will record EEG and a machine-learning classifier will decode the stimuli’s phases.