|
|
|
-- Nakahira, K.T., Harada, M., and Kitajima, M. (2023)Nakahira, K.T., Harada, M., and Kitajima, M. (2023). Local-Global Reaction Map: Classification of Listeners by Pupil Response Characteristics when Listening to Sentences Including Emotion Induction Words - Toward Adaptive Design of Auditory Information -. COGNITIVE 2023 : The Fifteenth International Conference on Advanced Cognitive Technologies and Applications, 79-85.
Local-Global Reaction Map: Classification of Listeners by Pupil Response Characteristics when Listening to Sentences Including Emotion Induction WordsWhen a person acquires a text as auditory information and derives the meaning of the text, he or she may simultaneously generate an emotion in response to the content of the text. Emotions are said to have a certain relationship with decision-making and memory. Therefore, it is expected that even sentences with the same meaning will be remembered differently depending on the emotion evoked. This study aims to clarify the relationship between the emotions that arise when listening to a text and the memory of the presented text. The classification of emotional states held by people is performed by a method based on subjective quantities by impression rating or by a method based on objective quantities by biometric information. In this study, we focus on pupil response, which is biological information that has been suggested to change with emotion. Based on this, this paper proposes the Local-Global Reaction Map (LGR-Map) as a classification method for pupil changes accompanying emotional changes, as a basic research for the construction of adaptive content design methods that utilize the degree of human emotional arousal. The LGR-Map is generated by capturing the emotional changes during listening to a text from the following two perspectives; Those generated by words in a specific region of a sentence (Local reaction); those generated by the context of the entire sentence (Global reaction). The total pupil diameter change within a certain time period is obtained as the characteristic quantity for each response. Error ellipses are defined for the distribution of listeners in the LR-GR for the presented text (LGR-Map), and classified into five types based on the rotation angle and flattening ratio of the error ellipses. The basic properties of the LGR-Map were investigated by using auditory stimuli presented in short sentences containing Affective Norm for English Words (ANEW). Download
|