supplement

Supporting Online Material: Material and Methods, and Fig. S1
1. fMRI experiments (Fig. 1A; Fig. S1)
Individual somatosensory thresholds were determined using the method of limits applying
electrical constant current pulses to either index finger (7 Hz, single pulse duration 0.2 ms). fMRI
scans were acquired at 1.5 Tesla using a T2*-weighted EPI-sequence (TR 2 s, TE 60 ms, flip
angle 90°, voxels 4×4×5 mm3; 16 slices covering the cerebrum down to the diencephalon
including the thalamus). In the first experiment (9 subjects, 28.6 yrs ± 4.6 yrs; Fig. 1A) the
stimulus intensity (mean 2.04 mA ± 0.65 mA) was set significantly lower than the sensory
threshold (mean 2.48 mA ± 0.73 mA) in order to guarantee a perceptually 'subliminal'
stimulation. During the subsequent fMRI scans the left index finger was stimulated at 7 Hz in
fifty blocks (each of 20 s duration) interleaved with fifty blocks without subliminal stimulation.
In the second experiment (7 subjects, 27.3 yrs ± 4.2 yrs; Fig. S1) near-threshold supraliminal
stimuli (mean 2.43 mA ± 0.77 mA; mean sensory threshold 1.99 mA ± 0.64 mA) were applied to
the left index finger at 7/3 Hz in fifty blocks (each of 20 s duration) half of which (arranged in a
randomised and balanced order) contained additional subliminal stimuli (mean 1.6 mA ± 0.57
mA; 7 Hz) with supraliminal stimuli applied 30 ms after the last preceding subliminal pulse.
These blocks alternated with periods (20 s) without any stimulation. Following standard
preprocessing, fMRI signal time courses were modelled by a delayed (4 s) squarewave-referencefunction using SPM99 (Wellcome Department of Cognitive Neurology, London, UK). The
realignment parameters were included as covariates into the model to improve the statistical
power. Since an additional regressor did not specify any parametric time-dependent signal
changes (p<0.001) for subliminal stimulation, trend effects (habituation or decreasing vigilance)
are negligible. Coordinates of activation sites are given in MNI space (Montreal Neurological
Institute).
2. Psychophysiological experiment (Fig. 1B)
Single electrical test pulses were delivered to the left index finger (0.5 Hz; six amplitude levels
with 0.2 mA inter-level difference, pulse duration 0.2 ms) in a randomised and balanced order
during blocks (20 s) with or without subliminal stimulation (7 Hz, single pulse duration 0.2 ms).
Each test pulse was applied 30 ms after the last preceding subliminal stimulus and detection was
confirmed by pressing a button with the right hand. The lowest test pulse amplitude was equal to
the subliminal current intensity (mean 1.91 mA ± 0.58 mA), i.e., 13.6 % below somatosensory
threshold (mean 2.19 mA ± 0.62 mA; method of limits for 7 Hz pulse trains).
Fig. S1. The T-contrast 'supraliminal stimulation' versus 'combined subliminal and supraliminal
stimulation' revealed a robust activation (random effects analysis, p<0.001 uncorrected for
multiple comparisons due to a strong a priori hypothesis) within contralateral S1 (X=44, Y=–36,
Z=62), and additionally to a lesser extent in ipsilateral S1 (X=–48 Y=–26 Z=44) and bilaterally in
S2. Thus, supraliminal stimulation alone evoked an activation stronger than when subliminal
stimulation was applied additionally.
Acknowledgements
This work was supported by the Deutsche Forschungsgemeinschaft (DFG, “Klinische
Forschergruppe”, Ei 207/2-3 and SFB 618-B4) and the Bundesministerium für Bildung und
Forschung (BMBF; “Berlin NeuroImaging Center”).