Increasing prevalence of opaque black-box AI has highlighted the need for explanations of their behaviours, for example, via explanation artefacts/proxy models. The current paper presents a paradigm for human-grounded experiments to evaluate the relationship between explanation fidelity, human learning performance, understanding and trust in a black-box AI by manipulating the complexity of an explanatory artefact. Decision trees were used in the current experiment as exemplar interpretable surrogate models, providing explanations approximating black-box behaviour, by means of explanation by simplification. Consistent with our hypotheses: 1) explanatory artefacts brought about better learning, while greater decision tree depths led to greater interpretability of the AI's performance and greater trust in the AI; and 2) explanatory artefacts facilitated learning and task performance even after they were withdrawn. Findings are discussed in terms of the interplay between human understanding, trust and AI system performance, highlighting the simplifying assumption of a monotonic relationship between explanation fidelity and interpretability.

Road accidents involving autonomous vehicles are inevitable and have the potential to damage the public's confidence in the technology and ultimately result in its disuse. It's important to understand how people react to such incidents and the influencing factors of blame attribution and trust restoration. Research in this field has started to grow but faces a huge methodological challenge, which is to develop high-fidelity experimental stimuli as realistic representations of accident scenarios in order to elicit valid reactions from human participants. The present paper reviews and evaluates several existing methods used in the research field before proposing an alternative method of generating animated accident sequence using driving simulation software. It is argued that this method strikes a good balance of fidelity, versatility and cost-effectiveness. We also present some preliminary evidence for the effectiveness of variable manipulation using such a methodology.

Smart home device usage is increasing, as is the diversity of users and range of devices. Additionally, it is becoming increasingly common to interconnect devices (e.g., via trigger-action rules) which, while bringing benefits, can bring unforeseen security and privacy risks. Developing strategies to protect users as well as understanding what biographical or attitudinal characteristics contribute to these risks is a critical step for ensuring empowered, but safe, interconnected smart device usage. Using narrative descriptions of domestic smart devices, two experiments explored how the prevailing security/privacy contexts—priming conditions—in which 20 trigger-action rules (developed via a Delphi Study) were presented influenced the adoption of rules favoring either security or privacy. Both experiments contrasted three priming conditions: no prime, security prime, privacy prime. Experiment 1 (n = 254) used explicit priming, giving direct instruction to maximize a security or privacy outcome while Experiment 2 (n = 325) used implicit priming, with an apparently unrelated security or privacy problem-solving puzzle. Across both experiments, priming promoted safer rule adoption, markedly so when explicit. Explicit priming produced an asymmetry however: privacy priming improved privacy scores with security scores unchanged and security primes improved security scores while worsening privacy scores. Across experiments, two dimensions of user attitudes shaped riskier rule choice: perceived benefits of technology and pre-existing trusting beliefs in online companies. Our novel findings reveal that implicit and explicit priming shape safe use of trigger-action rules in domestic settings and that age, perceived trust and perceived benefits should be considered when designing safety messaging.

Object manipulation in cluttered virtual environments (VEs) brings additional challenges to the design of interaction algorithms, when compared with open virtual spaces. As the complexity of the algorithms increases so does the flexibility with which users can interact, but this is at the expense of much greater difficulties in implementation for developers. Three rules that increase the realism and flexibility of interaction are outlined: collision response, order of control, and physical compatibility. The implementation of each is described, highlighting the substantial increase in algorithm complexity that arises. Data are reported from an experiment in which participants manipulated a bulky virtual object through parts of a virtual building (the piano movers' problem). These data illustrate the benefits to users that accrue from implementing flexible rules of interaction.

Cooperation between multiple users in a virtual environment (VE) can take place at one of three levels. These are defined as where users can perceive each other (Level 1), individually change the scene (Level 2), or simultaneously act on and manipulate the same object (Level 3). Despite representing the highest level of cooperation, multiuser object manipulation has rarely been studied. This paper describes a behavioral experiment in which the piano movers' problem (maneuvering a large object through a restricted space) was used to investigate object manipulation by pairs of participants in a VE. Participants' interactions with the object were integrated together either symmetrically or asymmetrically. The former only allowed the common component of participants' actions to take place, but the latter used the mean. Symmetric action integration was superior for sections of the task when both participants had to perform similar actions, but if participants had to move in different ways (e.g., one maneuvering him/herself through a narrow opening while the other traveled down a wide corridor) then asymmetric integration was superior. With both forms of integration, the extent to which participants coordinated their actions was poor and this led to a substantial cooperation overhead (the reduction in performance caused by having to cooperate with another person).

Interactive multimedia provide a platform for delivering instructional material in different modalities and styles. Experimental work on learning and memory is extensive, particularly for verbal learning. However, little attention has been paid to the relative effectiveness of auditory and visual presentation. We warn against simply transferring techniques developed in more traditional audiovisual systems, arguing instead for an approach based on experimental analysis. In this article we review studies that suggest a series of tentative guidelines on the use of auditory and visual presentation in multimedia learning systems. These guidelines are intended as an aid to both programme developers and users, but they also serve to highlight areas requiring further research.

Speech, as a medium for communicating with computers, has several potential advantages, and its introduction may have far‐reaching implications for the evolution of information technology. Yet, despite the availability of speech I/O devices over the past few years, their introduction into day‐today use has been relatively slow. A human factors perspective helps to explain the predominance of the visual/manual interface, and the reasons for its preference over speech. This view reflects fundamental differences between the two modes of communication. However, human factors also suggest that much can be done to improve the usability of speech systems. Broad guidelines for the use of speech‐output devices are detailed, and include criteria for the assignment of output either to the visual or auditory channel. The human factors of synthetic speech are discussed, in terms of enhancing both its intelligibility and its acceptability to users. Performance measures of automatic speech recognition (ASR) technology must account for the sensitivity of such devices to extraneous factors. A description of contemporary ASR technology serves as a basis for the explanation of this sensitivity, and of the limited performance of such devices. Guidelines for the implementation of ASR technology are described that address both user behavior and the interaction of the device with its environment. Modality effects upon user behavior in a man‐machine dialogue are described.

Subjects recalled items from short‐term memory by speaking into a speech recognizer. Two experiments examined effects of the type of feedback provided by the device during this data entry task. Three types of feedback were compared, varying in: modality, either auditory or visual; timing, either concurrent or terminal, and specificity, either verbal or nonverbal. Recognizer performance was better with concurrent feedback than with terminal feedback and better with nonverbal feedback than with verbal feedback. In terms of the efficiency of memory (the number of errors and the rate of data throughput), performance was more impaired by concurrent verbal feedback than by nonverbal feedback. Two main functional features of feedback in automatic speech recognition were identified: (1) the degree of similarity between the feedback and the phonologically‐coded information held in short‐term memory, which pointed to the dangers of spoken feedback and to a lesser extent the use of verbal visual feedback, and (2) the extent to which prompting is required to establish the timeliness of data input, a feature which is especially important with isolated‐word speech recognition.