1. Non-averaged scalp-recorded brain potentials were studied in humans during selective attention to randomly intermixed series of stimuli to fingers. Physiological tests were use for validating the presence or absence of the short-latency cognition-related P40 electrogeneses in parietal cortex in the response to a single-target stimulus (P40 signifies a positive polarity of about 40 ms peak latency). 2. To minimize interference from the electroencephalogram and noise we mapped single brain responses over the scalp and identified P40 topographies by an updated form of the numerical estimator Z for assessment of recorded potentials over time. We found that Z should exceed 0.96 for at least 15 ms for validation of the topographical congruity between the single P40 and an averaged P40 template. 3. Individual responses to 145 target finger stimuli correctly identified by the subject were analysed. P40 occurred only intermittently (34.5%) in a series of targets, but its voltage was unexpectedly large, exceeding the P40 voltage in averaged responses by a factor of about 10. 4. The usual assumption in the averaging method that the single brain responses combined in the average are stable but merely contaminated by unrelated noise was shown to be false for the cognition-related P40, which was considerably underestimated because of its intermittency in the averaged single trials. 5. The reaction time of the subject was on average 19% shorter in the trials in which a P40 was present, thus suggesting that P40 can influence subsequent perceptual processing by the brain in the same trial. 6. The feasibility of identifying specific cognition-related electrogeneses in single brain responses opens up the study of momentary shifts in brain processing strategies thereby allowing the neurophysiology of cognition to be based in real time.