# Unwrapping the mirror tracing task

## TL;DR

This paper introduces a new method to analyze the Mirror Tracing Task, offering detailed insights into specific regions of difficulty and how they vary with age.

## Contribution

A novel angle-based analysis method for the Mirror Tracing Task, introducing residuals and density as new variables for detailed performance evaluation.

## Key findings

- Residuals and density variables identified age-sensitive error regions in the star drawing.
- The new method enabled clustering of drawings and quantifying their similarity through a time series-like approach.
- Results from the new method were comparable to traditional metrics when summarized as single values.

## Abstract

The Mirror Tracing Task (MTT) is a method used to study visuomotor skills learning. It is traditionally evaluated by counting the number of times a person draws outside of the borders of a figure, typically a star, while looking at its mirror reflection. While insightful for overall performance, this metric lacks a precise analysis of the tracing, such as details on errors in specific regions. We propose a new MTT analysis method that studies the drawing as a function of the angle around the figure’s center. Two new variables are introduced: residuals, which measure deviation from the ideal drawing, and density, which measures how often a specific path is retraced. These variables are defined per angle or region, allowing a more detailed analysis, highlighting the most challenging parts of the drawing for each person, and enabling comparison across trials or finding common patterns between individuals. We applied this approach to the first MTT trial of 210 participants using age as a variable of interest. Residuals and density were summarized as a single value and compared with the traditional approach, providing similar results. When analyzed as a function of the angle, these variables enabled the identification of specific regions of the star where the errors are age-sensitive. Additionally, a time series-like approach enabled us to cluster drawings and quantify their similarity. The code used for this new method has been made openly accessible to make it easier for its applications in new research or the reanalysis of previous projects.

The online version contains supplementary material available at 10.3758/s13428-025-02845-6.