Statistical Analysis of Micro-error Occurrence Probability for the Fitts’ Law-Based Pointing Task

Abstract

Identifying mild cognitive impairment (MCI) at an early stage and preventing its progression to dementia has become an important task. In order to solve this problem, we focused on micro-errors (MEs), including stagnation of behavior, as a criterion for discriminating between healthy subjects and MCI patients. According to the naturalistic action test (NAT), when the difficulty of the task (index of difficulty: ID) is increased, the occurrence frequency increases drastically. In this research, we aimed to develop a model that simplifies the virtual kitchen challenge (VKC), which reproduces the NAT task on a tablet terminal, and estimates the ME occurrence probability based on a learning difficulty model that considers shape similarity. In this study, 20 university students were asked to perform a shape task. Using the generalized linear model showing the relationship between the occurrence probability of the ME and the result of the shape task, we confirmed that the ME occurrence probability increases with the difficulty level. Moreover, as future work, it is necessary to investigate the influence of handedness and gaze and the relation between color similarity and planning with regard to the ME occurrence probability.