Paradoxes of the first-night effect: a quantitative analysis of antero-posterior EEG topography