Figure 3
From: Improvement in reading performance through training with simulated thalamic visual prostheses
Reading Acuities Over Time. The progression of learning over time is shown for the population. (A) The mean performance at the start of the experiment in percent correctly read words at each font size is plotted according to viewing condition (Clear is shown in orange; P2000/Easy in green; P1000/Medium in blue, and P500/Hard in red). Filled symbols show the means of the measured values while colored vertical bars show the standard deviations. Curved lines are logistic fits, with the interpolated 50% threshold values highlighted with unfilled circles. The first five sessions (approximately one week) have been pooled in this plot. (B) Interpolated 50% thresholds are plotted over time using a sliding window of 5 sessions. Session number runs horizontally. Because logistic fits became quite poor when the underlying acuity, as identified by the 50% performance level, went below logMAR 0.8, the P1000 (blue) and P2000 (green) plots are quite noisy. Moreover the Clear (orange) plot is shown at the default value of zero, indicating no fit was possible, a result of the answers being all 100% correct. All three phosphene-view curves drop substantially over the training period indicating a striking improvement in functional acuity. Robust linear fits are shown as overlaid straight lines of decreasing slope from the Hard to Easy conditions, although the fit for the Easy condition (green) must be taken with due consideration given the low quality of the data. (C) The same plot as in A, for the final five sessions. While a reasonable fit was possible for the P500/Hard (red) viewing condition, the fit for P1000/Medium (blue) illustrates the incipient problems in fitting an accurate curve when overall subject performance is high on all cases within the tested range. An accurate fit was not possible for P2000/Easy (green), nor for Clear (orange, largely hidden by the other traces).
