Supporting Artificial Social Intelligence With Theory of Mind.

In this paper, we discuss the development of artificial theory of mind as foundational to an agent's ability to collaborate with human team members. Agents imbued with artificial social intelligence will require various capabilities to gather the social data needed to inform an artificial theory of mind of their human counterparts. We draw from social signals theorizing and discuss a framework to guide consideration of core features of artificial social intelligence. We discuss how human social intelligence, and the development of theory of mind, can contribute to the development of artificial social intelligence by forming a foundation on which to help agents model, interpret and predict the behaviors and mental states of humans to support human-agent interaction. Artificial social intelligence will need the processing capabilities to perceive, interpret, and generate combinations of social cues to operate within a human-agent team. Artificial Theory of Mind affords a structure by which a socially intelligent agent could be imbued with the ability to model their human counterparts and engage in effective human-agent interaction. Further, modeling Artificial Theory of Mind can be used by an ASI to support transparent communication with humans, improving trust in agents, so that they may better predict future system behavior based on their understanding of and support trust in artificial socially intelligent agents.

Evaluation of a Measurement System to Assess ICU Team Performance.

OBJECTIVE: Measuring teamwork is essential in critical care, but limited observational measurement systems exist for this environment. The objective of this study was to evaluate the reliability and validity of a behavioral marker system for measuring teamwork in ICUs. DESIGN: Instances of teamwork were observed by two raters for three tasks: multidisciplinary rounds, nurse-to-nurse handoffs, and retrospective videos of medical students and instructors performing simulated codes. Intraclass correlation coefficients were calculated to assess interrater reliability. Generalizability theory was applied to estimate systematic sources of variance for the three observed team tasks that were associated with instances of teamwork, rater effects, competency effects, and task effects. SETTING: A 15-bed surgical ICU at a large academic hospital. SUBJECTS: One hundred thirty-eight instances of teamwork were observed. Specifically, we observed 88 multidisciplinary rounds, 25 nurse-to-nurse handoffs, and 25 simulated code exercises. INTERVENTIONS: No intervention was conducted for this study. MEASUREMENTS AND MAIN RESULTS: Rater reliability for each overall task ranged from good to excellent correlation (intraclass correlation coefficient, 0.64-0.81), although there were seven cases where reliability was fair and one case where it was poor for specific competencies. Findings from generalizability studies provided evidence that the marker system dependably distinguished among teamwork competencies, providing evidence of construct validity. CONCLUSIONS: Teamwork in critical care is complex, thereby complicating the judgment of behaviors. The marker system exhibited great potential for differentiating competencies, but findings also revealed that more context specific guidance may be needed to improve rater reliability.

Enhancing the effectiveness of human-robot teaming with a closed-loop system.

With technological developments in robotics and their increasing deployment, human-robot teams are set to be a mainstay in the future. To develop robots that possess teaming capabilities, such as being able to communicate implicitly, the present study implemented a closed-loop system. This system enabled the robot to provide adaptive aid without the need for explicit commands from the human teammate, through the use of multiple physiological workload measures. Such measures of workload vary in sensitivity and there is large inter-individual variability in physiological responses to imposed taskload. Workload models enacted via closed-loop system should accommodate such individual variability. The present research investigated the effects of the adaptive robot aid vs. imposed aid on performance and workload. Results showed that adaptive robot aid driven by an individualized workload model for physiological response resulted in greater improvements in performance compared to aid that was simply imposed by the system.

Olfactory Stimuli Increase Presence in Virtual Environments.

BACKGROUND: Exposure therapy (EXP) is the most empirically supported treatment for anxiety and trauma-related disorders. EXP consists of repeated exposure to a feared object or situation in the absence of the feared outcome in order to extinguish associated anxiety. Key to the success of EXP is the need to present the feared object/event/situation in as much detail and utilizing as many sensory modalities as possible, in order to augment the sense of presence during exposure sessions. Various technologies used to augment the exposure therapy process by presenting multi-sensory cues (e.g., sights, smells, sounds). Studies have shown that scents can elicit emotionally charged memories, but no prior research has examined the effect of olfactory stimuli upon the patient's sense of presence during simulated exposure tasks. METHODS: 60 adult participants navigated a mildly anxiety-producing virtual environment (VE) similar to those used in the treatment of anxiety disorders. Participants had no autobiographical memory associated with the VE. State anxiety, Presence ratings, and electrodermal (EDA) activity were collected throughout the experiment. RESULTS: Utilizing a Bonferroni corrected Linear Mixed Model, our results showed statistically significant relationships between olfactory stimuli and presence as assessed by both the Igroup Presence Questionnaire (IPQ: R2 = 0.85, (F(3,52) = 6.625, p = 0.0007) and a single item visual-analogue scale (R2 = 0.85, (F(3,52) = 5.382, p = 0.0027). State anxiety was unaffected by the presence or absence of olfactory cues. EDA was unaffected by experimental condition. CONCLUSION: Olfactory stimuli increase presence in virtual environments that approximate those typical in exposure therapy, but did not increase EDA. Additionally, once administered, the removal of scents resulted in a disproportionate decrease in presence. Implications for incorporating the use of scents to increase the efficacy of exposure therapy is discussed.

The effects of video game experience and active stereoscopy on performance in combat identification tasks.

OBJECTIVE: We investigated the effects of active stereoscopic simulation-based training and individual differences in video game experience on multiple indices of combat identification (CID) performance. BACKGROUND: Fratricide is a major problem in combat operations involving military vehicles. In this research, we aimed to evaluate the effects of training on CID performance in order to reduce fratricide errors. METHOD: Individuals were trained on 12 combat vehicles in a simulation, which were presented via either a non-stereoscopic or active stereoscopic display using NVIDIA's GeForce shutter glass technology. Self-report was used to assess video game experience, leading to four between-subjects groups: high video game experience with stereoscopy, low video game experience with stereoscopy, high video game experience without stereoscopy, and low video game experience without stereoscopy. We then tested participants on their memory of each vehicle's alliance and name across multiple measures, including photographs and videos. RESULTS: There was a main effect for both video game experience and stereoscopy across many of the dependent measures. Further, we found interactions between video game experience and stereoscopic training, such that those individuals with high video game experience in the non-stereoscopic group had the highest performance outcomes in the sample on multiple dependent measures. CONCLUSION: This study suggests that individual differences in video game experience may be predictive of enhanced performance in CID tasks. APPLICATION: Selection based on video game experience in CID tasks may be a useful strategy for future military training. Future research should investigate the generalizability of these effects, such as identification through unmanned vehicle sensors.

Latent factor modeling of four schizotypy dimensions with theory of mind and empathy.

Preliminary evidence suggests that theory of mind and empathy relate differentially to factors of schizotypy. The current study assessed 686 undergraduate students and used structural equation modeling to examine links between a four-factor model of schizotypy with performance on measures of theory of mind (Reading the Mind in the Eyes Test [MIE]) and empathy (Interpersonal Reactivity Index [IRI]). Schizotypy was assessed using three self-report measures which were simultaneously entered into the model. Results revealed that the Negative factor of schizotypy showed a negative relationship with the Empathy factor, which was primarily driven by the Empathic Concern subscale of the IRI and the No Close Friends and Constricted Affect subscales of the Schizotypal Personality Questionnaire. These findings are consistent with a growing body of literature suggesting a relatively specific relationship between negative schizotypy and empathy, and are consistent with several previous studies that found no relationship between MIE performance and schizotypy.

Toward understanding social cues and signals in human-robot interaction: effects of robot gaze and proxemic behavior.

As robots are increasingly deployed in settings requiring social interaction, research is needed to examine the social signals perceived by humans when robots display certain social cues. In this paper, we report a study designed to examine how humans interpret social cues exhibited by robots. We first provide a brief overview of perspectives from social cognition in humans and how these processes are applicable to human-robot interaction (HRI). We then discuss the need to examine the relationship between social cues and signals as a function of the degree to which a robot is perceived as a socially present agent. We describe an experiment in which social cues were manipulated on an iRobot Ava(TM) mobile robotics platform in a hallway navigation scenario. Cues associated with the robot's proxemic behavior were found to significantly affect participant perceptions of the robot's social presence and emotional state while cues associated with the robot's gaze behavior were not found to be significant. Further, regardless of the proxemic behavior, participants attributed more social presence and emotional states to the robot over repeated interactions than when they first interacted with it. Generally, these results indicate the importance for HRI research to consider how social cues expressed by a robot can differentially affect perceptions of the robot's mental states and intentions. The discussion focuses on implications for the design of robotic systems and future directions for research on the relationship between social cues and signals.

Perceptual training for visual search.

People are better at visual search than the best fully automated methods. Despite this, visual search remains a difficult perceptual task. The goal of this investigation was to experimentally test the ways in which visual search performance could be improved through two categories of training interventions: perceptual training and conceptual training. To determine the effects of each training on a later performance task, the two types of trainings were manipulated using a between-subjects design (conceptual vs. perceptual × training present vs. training absent). Perceptual training led to speed and accuracy improvements in visual search. Issues with the design and administration of the conceptual training limited conclusions on its effectiveness but provided useful lessons for conceptual training design. The results suggest that when the visual search task involves detecting heterogeneous or otherwise unpredictable stimuli, perceptual training can improve visual search performance. Similarly, careful consideration of the performance task and training design is required to evaluate the effectiveness of conceptual training. PRACTITIONER SUMMARY: Visual search is a difficult, yet critical, task in industries such as baggage screening and radiology. This study investigated the effectiveness of perceptual training for visual search. The results suggest that when visual search involves detecting heterogeneous or otherwise unpredictable stimuli, perceptual training may improve the speed and accuracy of visual search.

The influence of camouflage, obstruction, familiarity and spatial ability on target identification from an unmanned ground vehicle.

The purpose of this study was to examine the effects of environmental and cognitive factors on the identification of targets from an unmanned ground vehicle (UGV). This was accomplished by manipulating obstruction, camouflage and familiarity of objects in the environment, while also measuring spatial ability. The effects of these variables on target identification were studied by measuring performance of participants that observed pre-recorded video from a 1:35 scaled military operations in urban terrain facility. Analyses indicated that a combination of camouflage and obstruction caused the most detrimental effects on performance, and that there were differences in the recognition of familiar and unfamiliar targets. Further analysis indicated that these detrimental effects could only be overcome with a combination of target familiarity and spatial ability. The findings highlight the degree to which environmental factors hinder performance and the need for a multidimensional approach for improving performance under these conditions. Areas in need of future research are also discussed. PRACTITIONER SUMMARY: Cognitive theory is applied to the problem of perception from UGVs. Results from an experimental study indicate that a combination of camouflage and obstruction caused the most detrimental effects on performance, with differences in the recognition of both familiar and unfamiliar targets. Familiarity and spatial ability interacted to predict the performance.

The Play Experience Scale: development and validation of a measure of play.

OBJECTIVE: A measure of play experience in video games was developed through literature review and two empirical validation studies. BACKGROUND: Despite the considerable attention given to games in the behavioral sciences, play experience remains empirically underexamined. One reason for this gap is the absence of a scale that measures play experience. METHOD: In Study 1, the initial Play Experience Scale (PES) was tested through an online validation that featured three different games (N = 203). In Study 2, a revised PES was assessed with a serious game in the laboratory (N = 77). RESULTS: Through principal component analysis of the Study 1 data, the initial 20-item PES was revised, resulting in the 16-item PES-16. Study 2 showed the PES-16 to be a robust instrument with the same patterns of correlations as in Study 1 via (a) internal consistency estimates, (b) correlations with established scales of motivation, (c) distributions of PES-16 scores in different game conditions, and (d) examination of the average variance extracted of the PES and the Intrinsic Motivation Scale. CONCLUSION: We suggest that the PES is appropriate for use in further validation studies. Additional examinations of the scale are required to determine its applicability to other contexts and its relationship with other constructs. APPLICATION: The PES is potentially relevant to human factors undertakings involving video games, including basic research into play, games, and learning; prototype testing; and exploratory learning studies.

TPL-KATS-card sort: a tool for assessing structural knowledge.

The study of how individuals organize knowledge has been a popular endeavor for several decades. As a result, techniques have been developed to assess how individuals represent and organize knowledge internally. Although several conceptual knowledge elicitation methods have been developed and used to assess the organization of knowledge, their use is often labor intensive and time consuming. Presented here is a software tool that was developed to reduce the problems associated with manually administering the conceptual knowledge elicitation technique, or card sorting. The TPL-KATS-card sort software not only simplifies the administration of the task, but also adds features to the card-sorting task such as media insertion, time stamping, and instructorless administration. In the present article, an introduction to the card-sorting technique is provided, the new software tool is described, and the advantages of the software are detailed.