Visual acuity thresholds for human face identity recognition

Purpose : Visual function of patients is usually assessed in clinical settings with behavioral tests such as the visual acuity grid of Snellen or the visual field evaluation with standard automated perimetry. However, the sight abilities evaluated have little in common with the challenging day-to-day complex visual recognition functions. Here we use fast periodic visual stimulation (FPVS) with electroencephalography (EEG) to objectively assess the impact of visual acuity (VA) on an automatic daily visual task: facial identity recognition (FIR). Methods : We tested automated discrimination of unfamiliar face identity with a well-validated FPVS-EEG paradigm in 12 normally sighted adult volunteers with a 68-electrode EEG system. Visual acuity was artificially decreased using a blur technique to obtain a linear scale of LogMar VA from 1,3 to 0. Results : At the group level, a right occipito-temporal FIR emerges already at a monocular VA threshold of 1,0 LogMar (z-score= 2,7259, p=.0032), with half of the sample showing a significant response. Neural amplitude reaches saturation at 0,7 LogMar (z-score= 7,798; p ≈ 0). Influence of monocular VA on FIR response can be separated in two trends: A first increase in neural response from 1,3 LogMar to 0,7 LogMar and a second flat trend from 0,7 LogMar to 0 LogMar. Conclusions : FIR in normally sighted participants is extremely well preserved at low VA, with a threshold around 1,0 LogMar of monocular VA.