The three types of non-parametric permutation Fisher tests have been applied to inter-individual group studies and further to intra-individual multiple EEG recording sequences, providing computations of EEG probability maps testing two ordinal hypotheses. Two examples of previous group studies with "EEG local cerebral activation" are given: mental computation in a group of 20 controls and caffeine effects versus placebo in a group of 10 controls. For the intra-individual study, two successive recordings of 2.3 min eyes closed (EC1 and EC2), obtained at 50 min intervals, were compared by paired exact permutation Fisher tests (over 15 or 42 synchronous EEG sequences). These tests were applied to descriptive spectral parameters: RMS and % amplitudes, mean frequencies, resonance coefficient, for raw unfiltered EEG and delta, theta, alpha, alpha 1, alpha 2, beta 1, beta 2 frequency bands. Two hypotheses were tested for each of the computed 31 parameters, providing two probability maps indicating if the parameter was greater or lower in the first EEG recording or in the second. The second EEG sequence, EC2, was "EEG activated" compared to the first sequence EC1 if the following were present: decreased amplitudes mainly in raw EEG, low activity and alpha bands; increased frequencies mainly, in raw EEG, delta and beta 1 fast activities; increased fast activity percentages; decreased coefficient of resonance. The effect of choice of reference was also evaluated: probability maps for a frontal reference were different than other probability maps obtained after computation of average reference or source derivation.(ABSTRACT TRUNCATED AT 250 WORDS)