Relationship Between Social Response To Virtual Avatar And Symptom Severity Of Patients With Schizophrenia

A virtual avatar has been used for various applications that require communication among persons, or to train or educate people by demonstrating human-like behavior. Recently, much research has shown an enhancement of virtual avatar technology, with the avatar perceived as a real person. As a result, the technology could begin to be used for observing human behavior to a virtual avatar. This paper concerns whether a virtual avatar could acquire patients' behavioral characteristics during experiencing a virtual task which is composed of approaching and conducting a short conversation. For this, we designed a virtual avatar that was standing in a virtual room, with eleven schizophrenic patients assigned the task of approaching the virtual avatar, initiating a conversation, and providing answers to the avatar's questions. To measure behavioral parameters in the virtual environment, we acquired the interpersonal distance and the verbal response time. In addition, we rated patients on the Positive and Negative Syndrome Scale (PANSS) in order to investigate a relationship between patients' symptomatic characteristic and behavioral parameters. Results of this study revealed that the interpersonal distance was negatively correlated with the negative syndrome scale, which is a subscale of PANSS (r=-0.687, p=0.02). By contrast, the verbal response time was not correlated with any other subscale of PANSS. However, after analyzing this variable with subitems of the negative syndrome of PANSS, two positive correlations were found: one with blunted affect (r=. 638, p=. 035) and the other with poor rapport (r=. 615, p=. 044). The negative correlation between the distance and negative symptom severity observed in this study is consistent with studies that investigated the relationship between schizophrenic patients' interpersonal distance and their symptoms. The positive correlation between the verbal response time acquired and subscales in PANSS in this study could be explained by the definition of these subscales. Therefore, this positive correlation means that the less a patient's emotional response, intimacy, and relationship making with a virtual avatar, the slower they answer the avatar's question. Inferring from these results, we conclude that the virtual avatar could be perceived as a real human by schizophrenic patients, the avatar could facilitate the schizophrenic patients' behavioral characteristics, and the avatar could be used as a tool for assessing the behavioral characteristics of patients with schizophrenia.