Cognitive load examination via pen interactions

Cognitive load refers to the mental effort when a cognitive task is conducted. As an important mental construct affecting task performance and subjective experience, cognitive load is considered a key factor in human-computer interaction system design and user operation efficiency improvement. Among possible modalities for cognitive load examination, pen interaction-based assessment is just beginning to be investigated, and only very few empirical studies have been conducted on a small set of writing features. This thesis involves several novel examinations of the relationship between cognitive load and writing features, aiming at the first comprehensive understanding of cognitive load measurements and effects of cognitive load on writing. Three new handwriting datasets, each focused on a different category of writing content (text, sketch, digits respectively), have been specifically designed and collected. Analytical results have shown that writing pressure, velocity and pen orientation can be used for cognitive load estimation. Examination of smoothness-related features has revealed that the rapidly-varying component of writing velocity is suppressed during high cognitive load. Furthermore, cognitive load affects writing shapes, which implies that characters written under high cognitive load conditions are difficult to recognize with handwriting recognition methods. Features extracted from stroke-level, sub-stroke-level and point-level data have been compared, and an important finding is that stroke and sub-stroke level features are more suitable for cognitive load estimation than point level features. Furthermore, it is identified that long strokes are more sensitive to cognitive load changes, and stroke selection criteria based on pen orientation and stroke curvature can improve cognitive load classification accuracy. Finally, cognitive load is found to affect writing behaviour and writing shape: for straight strokes, cognitive load has a higher impact on horizontally-directed writing than vertical, meaning that finger movements are more sensitive to cognitive load variations than wrist. Cognitive load also contributes to inconsistencies in writing shape, especially for curved strokes, which may ultimately affect handwriting recognition and result in up to 17% difference in recognition rate. Knowledge of the link between cognitive load and writing behavior is critical to improve writing system design, and further to enhance user writing experience.

Grantor dc:publisher

UNSW, Sydney

Year dc:date

2015

Author dc:creator

• Yu, Kun

• Written strokes
• Cognitive load measurement
• Pen interaction
• Behaviour analysis

Statement dc:rights

• open access
• CC BY-NC-ND 3.0
• free_to_read

Licence

• https://purl.org/coar/access_right/c_abf2
• https://creativecommons.org/licenses/by-nc-nd/3.0/au/

Language dc:language

EN

Identifier

https://doi.org/10.26190/unsworks/18814

OAI identifier oai:identifier

oai:unsworks.library.unsw.edu.au:1959.4/55674