Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2017. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 67-72). / This is me, my hair, my lip, my smell, the way my head moves and the body I live in. The sense of self is not fixed but continuously updates in response to the present information. When the body senses itself internally and localizes its actions, it provides the basis for a material sense of self existence. At the same time, the mind registers the sense of an agency with free will, the sense of being, the cause of voluntary action. The present self continuously becomes the past, and by the time we look into it we are in another present, consumed with planning the future. This thesis describes my theory and practice that concerns the sense of self. Two projects, Being A Tree and Masque, are included as examples of altering the perception of self through multisensory stimulation of exteroceptive signals and false feedback of interoceptive signals. Through these two systems, I aim to provide the audience a new, though temporary, relationship with themselves. This thesis incorporates my technical contribution in human computer interaction as well as my artistic inquiry. The goal is to reconfigure the tools of technology, not for exploitation but for the recovery of human feelings, affects and emotions. I hope the thesis delivers more than just the systems and study data. This thesis is about the moments of self-(re)organization and creating ripples in the fabric of self. It pays close attention to our shared psychological, emotional, cultural, and perceptual approaches to the inner and outer world and tries to bring light back to the sensitivity of self. / by Xin Liu. / S.M.

Thesis: Ph. D., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2017. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 161-172). / This dissertation introduces a controller for a context-aware office space that seamlessly and continuously transforms itself to support the activities of its occupants. Properties of the workspace, such as light, sound, textures and objects, have a remarkable influence on the human body, with measurable effects on physiology, cognition, and emotion. New control capabilities, e.g. wireless-controlled lighting and digital media displays, offer an opportunity to dynamically enhance the workspace for recreation, creativity, and productivity, in unison with the occupant's activities. Besides benefits for work engagement, personalized control could optimize for energy efficiency and foster Loose Fit planning to maximize flexible use of limited space. Accordingly, this controller incorporates sensing and computation to mediate between occupants' actions and the adaptive environment in a closed-loop fashion, producing what I call Mediated Atmospheres - seamless transitions of harmonious compositions of (1) light and (2) multimodal media. Through a series of studies, I evaluated this controller and demonstrated its potential to improve the workspace. Improvements were observed in energy savings (52% estimated energy savings compared to static illumination), significant (p < 0.05) increases of perceived fitness for focus and stress restoration, and significant (p < 0.05) physiological changes towards preferred conditions, e.g. in heart rate variability and facial expression. Furthermore, this work discusses the effects of adaptation as a form of behavioral feedback and raises broadly applicable questions about how to best balance automatic control and manual preference setting. The controller, in its core, builds on two contextual control dimensions, Focus and Restoration, which were experimentally discovered. They establish a lower dimensional representation (control map) that facilitates continuous mapping of the sensing and adaptive capabilities. Each unique space produces a new control map, which I computed using either subjective ratings, image analysis or physiological monitoring of the occupant. I performed an in-depth comparison of the image and rating approach for lighting and found that image analysis of either photographs or 3D renderings can accelerate the mapping process by reducing the amount of required human input. This finding allows us to generalize the subjective rating approach, and for the first time enables practicing designers to quickly link the positioning of lighting instruments to perceptual models of space. / by Nan Zhao. / Ph. D.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2018. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 75-84). / Learning language and literacy at a young age is important, as children's early language ability can impact their later educational success [1][2]. However, one of the major barriers to early language and literacy learning for many children around the globe is a lack of resources in homes and schools. A variety of technological interventions, such as TV series and educational apps, were designed to help overcome such barriers and support children's learning. However, not all of them necessarily provide children with conversational experiences, which have been found to significantly impact the children's language-related neural development [3]. Among a variety of educational media, embodied interactive agents (e.g., social robots) seem to be an effective yet resource-efficient tool that can enable children to learn through conversational turn taking. Specifically, embodied interactive agents can serve as learning companions for young children and provide more interactive and immersive learning experience. I explored how social robots could help promote children's language and literacy learning. More specifically, I designed and computationally created a collaborative, engaging learning interaction between a robot and a child who play as peers. First, I designed a tablet-based literacy learning game called WordQuest using the design principles for educational games. Second, I developed a reinforcement learning model that enabled the robot to adaptively switch its collaborative roles (e.g., expert and novice roles) in a way that promoted children's best learning. Third, I conducted an experiment with three conditions, which were fixed expert robot, fixed novice robot, and adaptive role switching robot, and tested on 60 children recruited from a local primary school in Boston. Last, I evaluated how the robot's collaborative roles differentially affected children's learning performance, engagement, and perception of the learning experiences. I found out that children across the three conditions all learned new words and had a very positive experience of playing WordQuest with the robot. In addition, children interacting with the adaptive robot consistently outperformed children from the other two conditions in terms of vocabulary acquisition and retention. / by Huili Chen. / S.M.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2018. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 73-77). / Storytelling is a fundamental way in which human beings make sense of the world. Animated movies tell stories that engage audience across culture and age groups. I designed and built My Personalized Movies (MPM), a novel system where animated stories are automatically created based on data provided by individuals. The data include self-tracked mood and behavior captured in quantitative measures and descriptive text. MPM is designed to engage viewers through an emotive narrative, induce self-reflection about their mood and behavior patterns, and to improve self-compassion and self-esteem, which mediates behavior change. I demonstrate with a few stages of studies, involving in total 107 participants, that viewers show strong emotional engagement with MPM and can explicitly connect animated characters' stories to one's past experiences. An analysis of 22 participants' facial expression data during MPM reveals that participants' change in implicit self-esteem is positively correlated with the happiness of their facial expression. Participants with higher depression severity, as measured by PHQ9, showed less positive facial expression at the happy moments in the animation. / by Fengjiao Peng. / S.M.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2018. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 93-96). / We live in an era of constant connectedness; we carry a smartphone in our pocket, headsets on our ears and enjoy limitless and regular access to almost any content we wish. However, the use of the personal computing devices that allow this connection with the virtual world damages our ability to connect with the physical world surrounding us; our eyes are focused on screens, our ears are covered by headphones and our attention jumps between apps. As a result, many of us are actually finding it harder to have face to face interactions with others than ever before. We are getting worse at communicating with the people around us, in the present, and tend to prefer virtual alternatives, as they are easier to operate, less stressful and fully under our control. This thesis proposes a perspective at wearable and personal computing devices and the role that their design may play in creating and fighting the epidemic of growing isolation. We hypothesize that the negative social trends that we witness as a result of using smartphones, headphones and other personal devices are not the purpose of these technologies, but rather an unwanted byproduct of their use. We propose to redesign ubiquitous personal technologies to reduce their isolating effect and use them to foster more physical interpersonal interactions and spatial awareness, by equipping them with additional modes of operation that force interpersonal interaction. We call this family of new interfaces IceBreakware. As a proof of concept, we present LeakyPhones, an instance of IceBreakware and a social version of the ubiquitous headphones. LeakyPhones is an interface that allows colocated and real time audio sharing between two or more people by coupling music sharing with a gaze. LeakyPhones encourages users to explore their surroundings with their eyes, and interact with the people around them. They also change the meaning of a previously private medium such as the headphones and turn it into public at will. By doing this, Leakyphones tries to overcome some of the limitations of normal headphones. This work explores corrective measures to standard personal devices that can possibly be implemented to existing technologies in order to encourage desired social behaviors. It demonstrates how gaze and music sharing may act as a social vehicle and help and encourage positive real-world interactions between people while not substituting them with virtual alternatives. / by Amos Golan. / S.M.

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2018 / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 75-81). / We face a future of delegating many important decision making tasks to artificial intelligence (AI) systems as we anticipate widespread adoption of autonomous systems such as autonomous vehicles (AV). However, recent string of fatal accidents involving AV reminds us that delegating certain decisions making tasks have deep ethical complications. As a result, building ethical AI agent that makes decisions in line with human moral values has surfaced as a key challenge for Al researchers. While recent advances in deep learning in many domains of human intelligence suggests that deep learning models will also pave the way for moral learning and ethical decision making, training a deep learning model usually encompasses use of large quantities of human-labeled training data. In contrast to deep learning models, research in human cognition of moral learning theorizes that the human mind is capable of learning moral values from a few, limited observations of moral judgments of other individuals and apply those values to make ethical decisions in a new and unique moral dilemma. How can we leverage the insights that we have about human moral learning to design AI agents that can rapidly infer moral values of human it interacts with? In this work, I explore three cognitive mechanisms - abstraction, society-individual dynamics, and response time analysis - to demonstrate how these mechanisms contribute to rapid inference of moral values from limited number of observed data. I propose two Bayesian cognitive models to express these mechanisms using hierarchical Bayesian modeling framework and use large-scale ethical judgments from Moral Machine to empirically demonstrate the contributions of these mechanisms to rapid inference of individual preferences and biases in ethical decision making. / by Richard Kim. / S.M. / S.M. Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2019 / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 65-71). / This thesis presents the design and implementation of TurboTrack, a 3D localization system for fine-grained robotic tasks. TurboTrack's unique capability is that it can localize backscatter nodes with sub-centimeter accuracy without any constraints on their locations or mobility. TurboTrack makes two key technical contributions. First, it presents a pipelined architecture that can extract a sensing bandwidth from every single backscatter packet that is three orders of magnitude larger than the backscatter communication bandwidth. Second, it introduces a Bayesian space-time superresolution algorithm that combines time series of the sensed bandwidth across multiple antennas to enable accurate positioning. Our experiments show that TurboTrack simultaneously achieves a median accuracy of sub-centimeter in each of the x/y/z dimensions and a 99th percentile latency less than 7.5 milliseconds in 3D localization. This enables TurboTrack's real-time prototype to achieve fine-grained positioning for agile robotic tasks, as we demonstrate in multiple collaborative applications with robotic arms and nanodrones including indoor tracking, packaging and assembly, and handover. / by Zhihong Luo. / S.M. / S.M. Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, 2019 / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 81-85). / The agonist-antagonist myoneural interface is a novel surgical construct that shows promise as a method of providing persons with amputation proprioceptive sensation of movement and force. This thesis aims to quantify the volitional coordination capabilities of the agonist- antagonist myoneural interface for applications related to control of active prostheses. In the first section, bilateral rhythmic coordination of ankle and subtalar joint movements is investigated in a control group of physically intact human subjects to characterize stereotypical kinematics of volitional lower limb movement. Subsequently, neuromusculoskeletal modeling techniques are developed to directly map estimated neural excitations from agonist-antagonist myoneural interface musculature to intended subtalar inversion and eversion kinematics. In a case study, the developed neuromusculoskeletal modeling techniques are applied to optimize a dynamic subtalar model for use by a unilateral subject with amputation possessing the agonist-antagonist myoneural interface. The subject's subsequent performance in bilateral rhythmic coordination utilizing the model and her own intact subtalar demonstrates the capacity of the agonist-antagonist myoneural interface to coordinate with intact anatomy in a biomimetic manner. / by Tony Shu. / S.M. / S.M. Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences

Thesis: S.M., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, May, 2020 / Cataloged from the official PDF of thesis. / Includes bibliographical references (pages 77-84). / Conventional imaging for health applications captures photons from the objects that are directly in the field of view of the camera. In this thesis, we develop computational frameworks to exploit scattered photons to image regions that are not directly visible to the camera. First, we will explore a new framework to model volumetric scattering with time-of- flight imaging to recover objects in scattering media with less need for calibrations. This technology can be applied to see under the skin. Second, we will exploit fluorescent tags, and quantum dots to image tagged objects around the corner for endoscopy with traditional cameras. We introduce a novel parametric approach to NLOS imaging for localizing tags around corners from radiometric measurements. The goals of the thesis are to develop novel approaches to model scattered light-transport and to demonstrate recovery of hidden objects, though scattering or around corners. The proposed technology can extend the scope of medical imaging. / by Tomohiro Maeda. / S.M. / S.M. Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences

Thesis: Ph. D., Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences, May, 2020 / Cataloged from the official PDF of thesis. / Includes bibliographical references (pages 171-186). / Individuals form network connections based on homophily; individuals' networks also shape their actions. Pervasive behavioral data provides opportunities for a richer view of the decisions on networks. Yet, the increasing volume, complex structures, and dynamics of behavioral data stretch the limit of conventional methods. I develop mathematical modeling (e.g., machine learning, game theory, and network science) and large-scale behavioral data to study collective behaviors over social networks. My dissertation will tackle this area in four directions, revolving around the intricate linkage between individuals' characteristics, actions, and their networks. First, I empirically investigate how social influence spreads over networks using two massive cell phone data, and theoretically model how do individuals aggregate information from local neighbors. Second, I study how to leverage influential nodes for selective network interventions (e.g., marketing and political campaigns), by proposing a centrality measure going beyond network structures. Third, I build a geometric deep learning model to infer individual preferences and make personalized recommendations to utilize noisy network information and nodal features effectively. Last, given that the network is essential, I develop a framework to infer the network connections based on observed actions, when networks are unavailable. My thesis provides building blocks for further network-based machine learning problems integrating nodal heterogeneity and network structures. Moreover, the findings on human behaviors and frameworks developed in my thesis shed light on marketing campaigns and population management. / by Yan Leng. / Ph. D. / Ph.D. Massachusetts Institute of Technology, School of Architecture and Planning, Program in Media Arts and Sciences