Neuro-Participatory Design: Neuroarchitecture with Natural Language Processing & Eye Tracking

Immanuel Koh | Assistant Professor in Architecture & Sustainable Design (ASD) and Design & Artificial Intelligence (DAI), Singapore University of Technology and Design (SUTD)

Elissa Gowika Hartanto | MSc. in Urban Science, Policy and Planning, Singapore University of Technology and Design

This paper aims to explore issues of public urban spaces that have emerged from the gaps found in today’s participatory design. While it is evident that the public are more involved in the design thinking process to produce results that should cater to the community, their opinions are often overlooked as less important remarks from non-professionals. This research attempts to explore public responses in relation to their implicit perceptions and behaviours. The study provides an empirical understanding of users of public spaces in embedding greater value for public voices within participatory design. By applying machine learning methods drawn from Natural Language Processing and the new field of Neuroarchitecture, the traditional methodology of participatory design could incorporate public participation as an implicit feature during the design process. Traditional methods of data collection (e.g., online questionnaires and intercept surveys) are used to extract respondents’ sentiments regarding two popular locations in Singapore: Keong Saik Road and Geylang Road. The data collected is processed through sentiment analysis to determine the positive and negative polarities towards urban elements at both sites. The subjectivity scores derived are used to determine users’ biases. To obtain users’ subconscious data that can be projected with the polarity scores, participants were invited to experience both sites while wearing eye tracking glasses. The device recorded the users’ periphery with gaze patterns. The data collected was then projected as heatmaps and Areas of Interests (AOI), with metrics involving gaze fixations, saccades and graphs depicting peaked interests. The saturations in heatmaps and AOIs are then projected with the metrics against the results from the sentiment analysis. Using correlation analysis, the qualitative relationship between the respondents and their subconscious human behaviour is made quantifiable. The application of machine learning analysis facilitates the production of valuable evidence that can be implemented for a more collaborative design process. It also extends the versatility of public opinions, to allow communities to become a larger stakeholder as the main users of public spaces. Our neuro-participatory design approach aims to better align the design of cities with implicit humanistic goals.