Recognition of extraversion level based on handwriting and support vector machines.

Area of Science:

• Psychology
• Computational Linguistics
• Biometrics

Background:

• Extraversion, a key personality trait, is typically assessed via self-report questionnaires.
• Handwriting analysis has historically been explored for personality insights, though scientific validation remains ongoing.
• Machine learning offers novel computational approaches to analyze complex data patterns, including behavioral and biometric markers.

Purpose of the Study:

• To determine if machine learning algorithms can accurately discriminate levels of extraversion based on handwriting characteristics.
• To identify specific handwriting variables that correlate with extraversion in a large sample.
• To investigate potential sex differences in the relationship between handwriting and extraversion.

Main Methods:

• Utilized support vector machines (SVMs), a supervised machine learning algorithm, for classification tasks.
• Analyzed handwriting samples from 883 participants (404 males, 479 females).
• Extracted 48 distinct handwriting variables and employed 10-fold cross-validation for model training and testing, with separate analyses for each sex.

Main Results:

• Achieved good recognition accuracy (approximately 0.7) in discriminating extraversion levels.
• Identified a set of 10 key handwriting variables that were predictive of extraversion.
• Observed that the significant handwriting variables differed between male and female participants.
• Demonstrated the feasibility of using computational methods to link handwriting features to personality traits.

Conclusions:

• The study provides evidence supporting a link between specific handwriting elements and the personality trait of extraversion.
• Machine learning models, particularly SVMs, can be effectively trained to recognize personality traits from handwriting.
• Sex-specific handwriting features appear to be associated with extraversion, suggesting nuanced relationships that warrant further investigation.