Objective measurement of visual resolution using the P300 to self-facial images

David J. Marhöfer, Michael Bach, Sven P. Heinrich

To assess visual acuity objectively "beyond V1", the P300 event-related potential is a promising candidate and closely associated with conscious perception. However, the P300 can be willfully modulated, a disadvantage for objective visual acuity estimation. Faces are very salient stimuli and difficult to ignore. Here, we present a P300-type paradigm to assess visual acuity with faces. Gray-scale portraits of the respective subject served as oddball stimuli (probability 1/7), scrambled versions of these as the standard stimuli (probability 6/7). Furthermore, stimuli were spatially high-pass filtered (at 0, 2.2, 4.2 and 8.3 cpd), making them recognizable only with sufficient acuity. Acuity was systematically reduced by dioptric blur, chosen individually to render faces unrecognizable when high-passed at ≥ 4.2 cpd. EEG was recorded from 11 subjects at 32 scalp positions and re-referenced to the average of TP9 and TP10. One of the rare face variants was designated as target, for which a button had to be pressed. The event-related potential was dominated by the P300 at 300-800 ms. All subjects had a significant (P < 0.05) P300 for 0- to 8.3-cpd filtering. When vision was blurred, the fraction of significant P300 responses to 8.3-cpd filtered faces dropped to 18 %, but stayed at 100 % for 4.2 cpd. Another component, the vertex positive potential (VPP) at 170 ms, was undetectable in most participants with blur and all levels of filtering, even when the images were recognizable. The study demonstrates the feasibility of a face-based P300 approach to objectively assess visual acuity. The sensitivity to stimulus degradation was comparable to that of a grating-based approach as previously reported. An unexpected finding was the differing behavior of the P300 and the VPP. The VPP was quite sensitive to high-pass filtering, while the P300 sustained stronger filtering, although for its generation, the faces must also be discriminated from scrambled faces.