Usefulness of blinking duration variability (BDV) in discriminating emotional states. (August 2021)

• Record Type: Journal Article
• Authors: Sarma, Parthana; Barma, Shovan
• Is Part Of: Biomedical signal processing and control. Volume 69(2021)
• Publication Date: 2021-08
• Subjects: Eye blinking -- Blinking rate variability -- Blinking duration -- Blink and emotion; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2021.102883
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Usefulness of blinking duration variability (BDV) in discriminating emotional states. Different time series techniques have been explored in BDV analysis. Histogram analysis, Poincare plot, sample entropy, recurrence plot are found useful. Above methods are suitable for discriminating neutral, positive, negative emotions. Abstract: Numerous studies show that various measures of eye blinking event including blinking rate, duration, blinking rate variability etc. has a strong correlation with mental processes. However, the blinking duration variability (BDV) in connection with different emotional states has not been explored in details. Therefore, this present study investigates the usefulness of BDV to discriminate different emotional states. In this purpose, SEED-IV database has been considered, in which 15 participants were presented with four types of emotional (neutral, sad, fear and happy) video clips in three different sessions and blinking number along with duration of each blinking (in ms) have been recorded simultaneously. Next, for each subject, the BDV signals have been generated for each type of emotional states and considered them as time series data. Now, keeping the analogy with the heart rate variability (HRV) or blinking rate variability (BRV) analysis, several time series analysis techniques, such as statistical, geometrical, sample entropy and recurrence plot have been employed. The results show that the sample frequency from histogram analysis, short-term BDV from Poincare plot, sample entropy and recurrence rate from recurrence plot are very much suitable to discriminate different emotional states, especially among neutral, negative (sad and fear) and positive (happy). Such findings reveal that the BDV could be a suitable measure to differentiate different basic emotional states. Hence, this study could enrich the domain of ocular event based emotion recognition and analysis.