Precision-Recall-Gain Curves: PR Analysis Done Right.

Abstract

Woodworth’s Two-Component model (1899) partitioned speeded limb movements into two distinct phases: (1) a central ballistic open-loop mechanism and (2) a closed-loop feedback component. The present study investigated the implementation of multi-gain control configurations that utilized separate gain values optimized for each movement phase. A target acquisition task using Fitts’ Law (1954) was performed within a virtual environment using multiple control devices with three gain settings: (1) mono-gain, (2) dual-gain, and (3) continuous gain. It was found that dual-gain and continuous gain configurations yielded lower movement times and information-processing rates than the mono-gain configurations. The secondary gain values presented in the dual-gain and continuous gain configurations were reported to mitigate oscillations around smaller targets that were responsible for additive settling time. Therefore, implementation of multi-gain control logic could help improve performance when navigating through large spaces and acquiring small targets.