The raw data obtained from the PR included noisy and physiologically implausible samples that may be attributed to, for example, blinks, aberrations from the tear film, eccentric gaze, and instrument limitations. A filtering algorithm was used to remove these outliers before further analysis, with further visual inspection used to confirm that the filtering did not introduce bias. The algorithm used known characteristics of the measurement technique and the physiology of the human visual system.
11,30 Only data that met the following combination of criteria, implemented in the instrument or applied in filtering, were included: (1) The accommodation data were within the linear operating range of the instrument (+4 to −6 D),
42 (2) pupil diameters were between 3.5 and 8 mm,
42,47 (3) accommodation velocity was <10 D/s,
48 and (4) vergence velocity was <175 pd/s.
49 Very few data points were excluded based on the requirement that the refractive data fall within the operating range of the instrument. These points were considered to be the result of measurement error, as only one subject (described below) had hyperopia capable of reaching the criterion level of refractive state and the excluded data were not sustained for a long enough period to be considered a plausible accommodative response.
A video recording collected during the experiment was used to determine the video frames at which there was a change in stimulus, for comparison with the PR data. The data analyses were performed using MATLAB (Mathworks, Natick, MA, USA), SPSS (SPSS, Inc., Chicago, IL, USA), GraphPad Prism (GraphPad Software, Inc., La Jolla, CA, USA), MacSHAPA (University of Illinois, Urbana-Champaign, IL, USA), and Microsoft Excel (Microsoft, Redmond, WA, USA).