Does Saccade Response Amplitude Gain Vary with the Amplitude of Target Demand?

Abstract

Purpose : To assess the effect of saccade length on response accuracy as measured by %overshoot and %undershoot in normal controls and in concussed subjects. Studies report that it is common to undershoot larger saccades(20⁰) and to overshoot smaller saccades(5⁰). Is there a saccade length which maximizes response accuracy and minimizes position error? Understanding typical patterns of saccade response accuracy is beneficial when comparing normal controls to concussed subjects.

Methods : In accordance with human research subject protocol and as part of an IRB approved study, 5 control & 5 post-concussion subjects viewed 720 custom generated (VisualEyes software, VNEL, NJIT) Landolt-C targets in 5, 10, 15 & 20⁰ horizontal pseudorandom pro-saccades (gap-0 paradigm) on a monitor @40cm with chin and head stabilization. A binocular infrared ISCAN eye tracker (Woburn, MA, USA) recorded saccades. MATLAB (MathWorks inc. Waltham,MA) and SPSS software(IBM inc. Armonk,NY) were used to analyze saccade data. Blinks, wrong turns, outliers, anticipatory and non-movements were excluded. Saccade response amplitude gain (SRAG) is calculated as +%overshoot or -%undershoot. Mean absolute% position error is compared between the 2 subject groups.

Results : Control subjects: 45% of the 5⁰ saccades undershoot, 60% of the 10⁰ saccades, 74% of the 15⁰ saccades and 81% of the 20⁰ saccades undershoot. Mean absolute% position error @5⁰= 9.18%(range 4.46-13.17), @10⁰= 6.3%(5.83-6.62), @15⁰= 9.81%(3.37-17.73), and @20⁰=11.08%(6.3-20.3).
Concussed subjects: 47.5% of the 5⁰ saccades undershoot, 69% of the 10⁰ saccades, 84% of the 15⁰ saccades and 92% of the 20⁰ saccades undershoot. Mean absolute% position error @5⁰=3.36%(range 2.34-4.87), @10⁰=10.05%(8.44-12.94), @15⁰= 20.04%(17.89-24.08), and @20⁰=26.23%(19.85-33.32).
Mean overall accuracy measures for controls=9.08% and concussed=14.9%. Significant internal variance exists.

Conclusions : Saccadic eye movement responses tend to undershoot the longer 20⁰ demands at a greater frequency than the shorter 5⁰ demands. Concussed subjects undershoot large saccade demands (20⁰) at a higher rate than control subjects. Additionally, concussed subjects show greater %position error than normal control subjects. Low-midrange saccade lengths of near 10⁰ yield maximum precision in SRAG with lower mean absolute %position error.

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.