Analyzing Affective Responses to Virtual Spaces Using Physiological Sensors and Verbal Descriptions

Abstract

Architects design spaces with assumptions about how their designs will affect users emotionally. These assumptions primarily rely on professional intuitions and subjective experiences. This thesis utilizes wearable sensors to collect and analyze human responses, especially emotions, when experiencing virtual spaces. It tests if the collected data can be used to predict users’ emotional responses to a spatial design in VR and to help architects design in a more informed and data-driven way.

To achieve this, data from 86 individuals in four experiments, each placed in a simulated environment of the synthetic model or scanned model, were analyzed. Collected data include verbal descriptions, recorded visual targets, electroencephalography (EEG), galvanic skin response (EDAs), and heart rates. The study consists of three parts: (1) design and build a VR environment with wearable sensors to collect data; (2) conduct experiments to collect participants’ physiological responses, verbal descriptions, and visual target data in the VR spaces; and (3) analyze the collected data to confirm that they relate to spatial design.

Experiments have demonstrated the existence of a certain relationship between physiological data and spatial parameters, such as EEG calm state and spatial height, and higher vigilance wandering from a relatively tall space to a relatively short space. In addition, by using verbal description analysis, we found an association between the physiological data and the spatial sequences and sounds. This evidence of correlations between physiological data and spatial or verbal description is a small step toward the development of a toolkit to assist designers in measuring user experience in VR environments.

This methodology offers a useful emotion measurement for a virtual architectural design using multiphysiological sensors and verbal descriptions. It leads to a potential future application that combines physiological metrics and AI methods and informs designers of users’ emotional experiences before a design is finished.