Analyzing the Impact of Cognitive Load in Evaluating Gaze-based Typing

Korok Sengupta, Jun Sun, Raphael Menges, Chandan Kumar, Steffen Staab

Abstract

Gaze-based virtual keyboards provide an effective interface for text entry by eye movements. The efficiency and usability of these keyboards have traditionally been evaluated with conventional text entry performance measures such as words per minute, keystrokes per character, backspace usage, etc. However, in comparison to the traditional text entry approaches, gaze-based typing involves natural eye movements that are highly correlated with human brain cognition. Employing eye gaze as an input could lead to excessive mental demand, and in this work we argue the need to include cognitive load as an eye typing evaluation measure. We evaluate three variations of gaze-based virtual keyboards, which implement variable designs in terms of word suggestion positioning. The conventional text entry metrics indicate no significant difference in the performance of the different keyboard designs. However, STFT (Short-time Fourier Transform) based analysis of EEG signals indicate variances in the mental workload of participants while interacting with these designs. Moreover, the EEG analysis provides insights into the user's cognition variation for different typing phases and intervals, which should be considered in order to improve eye typing usability.

Introduction

Gaze-based typing—using eye movements to select letters on a virtual keyboard—is a critical assistive technology for individuals with severe motor impairments, such as locked-in syndrome. Historically, researchers have evaluated these interfaces using "black box" performance metrics: Words Per Minute (WPM), Keystrokes Per Character (KPSC), and backspace usage.

However, these metrics fail to capture the user's internal cognitive experience. Two keyboards might facilitate the same typing speed, but one may impose a significantly higher mental strain, leading to quicker fatigue. Given that eye movements are deeply intertwined with brain cognition, this study argues that measuring cognitive load is essential to truly understand interface usability.

Methodology

The researchers compared three distinct virtual keyboard designs implemented via the eyeGUI library:

  • Keyboard A (Standard): Word suggestions appear in a single row at the top.
  • Keyboard B (Inter-spaced): Word suggestions are interspersed between the rows of keys.
  • Keyboard C (Per-key): Suggestions appear both at the top and directly on the keys themselves (the Gazethekey approach).

To measure mental effort, five able-bodied participants (ages 22–26) typed sentences from the Mackenzie and Soukoreff phrase set while wearing an Emotiv EPOC+ EEG headset. The researchers applied a Short-time Fourier Transform (STFT) to the raw EEG data to analyze the Beta band (≥14Hz), where a higher spectral power ratio indicates increased cognitive workload and concentration.

Results and Discussion

The Deception of Traditional Metrics

If evaluated solely by traditional metrics, the three designs appeared nearly identical. Average speeds were 9.20 WPM (A), 8.60 WPM (B), and 9.05 WPM (C), with ANOVA tests showing no statistically significant difference in speed or keystroke savings.

Cognitive Load by Typing Phase

The EEG data revealed significant differences that performance metrics missed. Analysis showed that cognitive load is not uniform across the typing process. The "INSERT" phase (typing individual letters) consistently demanded the highest mental effort across all designs due to the intensive foveal scanning and decision-making required for letter selection.

Comparison of Beta band spectral power ratios across typing phases (Backspace, Suggestion, and Insertion) for three keyboard designs.

Keyboard C (per-key suggestions) demonstrated the lowest overall cognitive load (mean ratio of 0.0824), significantly outperforming Keyboard A (0.0865) and Keyboard B (0.0860). By placing suggestions directly where the user is already looking, Keyboard C minimizes the mental "cost" of scanning a separate word list. Conversely, Keyboard B’s inter-spaced design proved to be a "mental trap"; despite the intention to reduce eye travel, the dynamic appearance of words between keys increased user frustration and cognitive load.

An interesting finding was the "Pre-test Spike": cognitive load was significantly higher during the initial instruction and setup phase than during the actual typing task. This suggests a period of initial adjustment to the eye-tracking environment before the user reaches a cognitive steady state.

Variation of cognitive load (Beta band ratio) over time, showing the initial 'Pre-test' spike followed by stabilization across sequential sentences.

Conclusion

This study demonstrates that traditional performance metrics are insufficient for evaluating gaze-based interfaces. EEG-based cognitive load analysis provides a more granular understanding of user experience, identifying specific design flaws and high-effort typing phases.

Future work points toward "cognition-aware" keyboards. Such systems could monitor EEG in real-time and dynamically adjust interface parameters—such as increasing dwell times or simplifying layouts—when a spike in cognitive load is detected, thereby preventing errors and reducing user exhaustion.

Summary updated

Abstract

Gaze-based virtual keyboards provide an effective interface for text entry by eye movements. The efficiency and usability of these keyboards have traditionally been evaluated with conventional text entry performance measures such as words per minute, keystrokes per character, backspace usage, etc. However, in comparison to the traditional text entry approaches, gaze-based typing involves natural eye movements that are highly correlated with human brain cognition. Employing eye gaze as an input could lead to excessive mental demand, and in this work we argue the need to include cognitive load as an eye typing evaluation measure. We evaluate three variations of gaze-based virtual keyboards, which implement variable designs in terms of word suggestion positioning. The conventional text entry metrics indicate no significant difference in the performance of the different keyboard designs. However, STFT (Short-time Fourier Transform) based analysis of EEG signals indicate variances in the mental workload of participants while interacting with these designs. Moreover, the EEG analysis provides insights into the user's cognition variation for different typing phases and intervals, which should be considered in order to improve eye typing usability.
BibTeX
@inproceedings{sengupta2017analyzing,
  author = {Korok Sengupta and Jun Sun and Raphael Menges and Chandan Kumar and Steffen Staab},
  title = {{Analyzing the Impact of Cognitive Load in Evaluating Gaze-based Typing}},
  booktitle = {CBMS},
  year = {2017},
  doi = {10.1109/cbms.2017.134},
}
Figure