Keith S. Jones

What Is an Affordance

What should and should not be considered an affordance is still an open issue. At the 2002 North American meeting of the International Society for Ecological Psychology, a symposium was held concerning this topic. The articles in this special issue are expanded versions of the papers presented at that symposium. To set the stage for those discussions, this article presents an analysis of the evolution of J. J. Gibsons thinking on this issue, from some of his earliest writings (i.e., J. J. Gibson & Crooks, 1938) through his most recent writings (i.e., J. J. Gibson, 1979/1986). It is hoped that this forum will stimulate thinking concerning this important ecological concept.

Human-Robot Interaction Toward Usable Personal Service Robots

The widespread adoption of personal service robots will likely depend on how well they interact with users. This chapter was motivated by a desire to facilitate the design of usable personal service robots. Toward that end, this chapter reviews the literature concerning people interacting with personal service robots. First, ongoing research related to the design of personal service robots is discussed. This material is organized around generic activities that would take place when a user initiates interaction with a future personal service robot, for example, understanding the robot’s affordances or its cognitive capabilities, as well as when a personal service robot initiates interaction with a user, for example, understanding the user’s intent or engaging and communicating with the user. Second, research areas that deserve more attention from the human-robot interaction community are discussed, for example, understanding when people do and do not treat robots as if they were people. Throughout the chap...

An Investigation of the Prevalence of Replication Research in Human Factors

Objective: The present studies investigated the nature of replication research within the human factors literature. Background: Many claim that researchers in certain fields do not replicate prior research. This is troubling because replications allow science to self-correct. A successful replication corroborates the original finding, whereas an unsuccessful replication falsifies it. To date, no one has assessed whether this issue affects the field of human factors. Method: In the first study, eight articles (parent articles) were selected from the 1991 issues of the journal Human Factors. Each article that had referenced one of the eight parent articles between 1991 and September 2006 (child articles) were also retrieved. Two investigators coded and compared each child article against its 1991 parent article to determine whether the child article replicated its parent article. The second study replicated these procedures. Results: Half or more of the parent articles in Study 1 and Study 2 (75% and 50%, respectively) were replicated at least once. Furthermore, human factors researchers conducted replications of their own work as well as the work of others. However, many researchers did not state that they replicated previous research. Conclusion: Replications seem to be common in the human factors literature. However, readers may not realize that a study replicated prior research. Thus, they may incorrectly assess the evidence concerning a given finding. Application: Human factors professionals should be taught how to identify replications and to be cautious of research that has not been replicated.

Teleoperation Through Apertures Passability Versus Driveability

Urban search and rescue (USAR) robots get stuck. Furthermore, USAR workers complained that it is difficult to judge whether a teleoperated robot can go through certain apertures. Two experiments tested teleoperators’ abilities to (a) judge whether a robot could fit through apertures (passability), (b) judge whether they could drive a robot through apertures (driveability), and (c) drive the robot through apertures. Experiment 1 examined teleoperators’ passability judgments and whether those same operators hit apertures that were wider than the robot. Experiment 2 replicated Experiment 1 and examined driveability judgments. Experiment 1 indicated that teleoperators made accurate passability judgments and routinely hit apertures that were wider than the robot. Experiment 2 successfully replicated Experiment 1 and demonstrated that teleoperators did not make accurate driveability judgments. Experiment 1 indicated that teleoperating a robot through an aperture is constrained by the robots’ physical dimensions plus a safety margin associated with how well the operators drive the robot. Thus, teleoperators should base decisions to enter an aperture on their ability to drive the robot. However, Experiment 2 indicated that teleoperators do not make accurate driveability judgments. These results have implications for teleoperator training and the design of robots for specific applications.

Why Does the Negative Impact of Inconsistent Knowledge on Web Navigation Persist

Farris (2003) discovered that users had greater difficulty finding information on a Web site when their prior knowledge was inconsistent with the Web sites content, relative to when their knowledge was consistent with it. In addition, he found that this difficulty was persistent over trials. To explain this persistence, Farris offered a schema-based account, which instantiated inconsistency in a single manner. These studies tested 2 predictions that were derived from Farriss account. Specifically, Experiment 1 assessed whether schema elaboration would be gradual, whereas Experiment 2 assessed whether task repetition would eliminate the negative impact of inconsistent knowledge. The results associated with navigation efficiency supported the predictions derived from Farriss account. The results associated with the choices made by participants as they navigated the site, however, contradicted the predictions. A new account, based on production-rules rather than schemata, is offered that considers both sets of results, because it instantiates inconsistency in more than 1 manner. This new account has implications for the design and redesign of Web sites.

Can Metric Feedback Training Hinder Actions Involving Distance

Objective: The present studies tested whether distance estimation training with metric feedback can degrade the performance of untrained primarily perceptual-motor tasks. Background: Training with metric feedback can improve distance estimations. However, previous research led to the conclusion that those improvements stemmed from changes in cognitive processing rather than in perception. If trainees applied their new cognitive strategies to primarily perceptual-motor tasks, then the performance of those tasks should degrade. The present studies tested that possibility. Method: Experiment 1 sought to replicate that training with metric feedback would improve metric distance estimations. Experiments 2 and 3 investigated whether such training would degrade the performance of a primarily perceptual-motor task. Experiment 4 investigated whether such training would affect a perceptual-motor task that required cognition. Results: Metric feedback improved metric distance estimation (Experiments 1—4) and throwing to a specified distance (Experiment 4). Metric feedback degraded throwing to a target (Experiments 2 and 3), although that effect was not evident when pretesting was omitted (Experiment 3). Conclusion: If distance estimation trainees apply what they learned from metric feedback to untrained primarily perceptual-motor tasks, then the performance of those tasks will suffer. However, if trainees apply what they learned to untrained tasks that require metric estimation, then the performance of those tasks will improve. Application: Distance estimation training with metric feedback may not generalize to other tasks and may even degrade performance on certain tasks. Future research must specify the conditions under which distance estimation training with metric feedback leads to performance improvements and decrements.

Framework for Developing a Brief Interview to Understand Cyber Defense Work: An Experience Report

Cyber defense is increasingly important for the wellbeing of our economy and our national defense. Universities can help meet our growing cybersecurity needs by training the next generation of cyber defenders, and it is crucial that the curricula for such programs are designed to prepare students for the type of work that is performed in the field. Unfortunately, collecting data about cyber work is hindered in situations where cybersecurity professionals are uncomfortable with traditional human factors work analysis methods. Four potential constraints are 1) no naturalistic observations, 2) anonymity and safety, 3) short data collection time, and 4) no deep process questions. We developed a brief interview technique that allowed us to measure the importance of knowledge, skills, and abilities related to offensive and defensive cyber work. Based on our experience using this technique, it fits within the four potential constraints to cyber research and produces information that is directly applicable to the development of cybersecurity curricula. Our technique could potentially be used for other research purposes and personnel selection and by researchers interested in other high-security populations.

Tele-operators' judgments of their ability to drive through apertures

It has been suggested that operators should base decisions to enter apertures on their ability to control the robot, rather than its static dimensions. Doing so, however, assumes that operators know whether they can drive a robot through the aperture. The present study tested that assumption. Results indicated that judgments about control of the robot were not accurate. In contrast, judgments of static dimensions were accurate. Thus, operators will require support if they must base decisions to enter apertures on their ability to control that robot.

Width guidelines for rectangular objects with penetrable and impenetrable borders

Generally, selection times quicken when objects are placed against a displays edge. Experiment 1 investigated whether or not this continues to be true if the width of those objects, i.e. rectangular scrollbars, was manipulated. The results indicated that increasing width affected selection times for objects with penetrable borders, but not for those with impenetrable borders. A follow-up experiment examined whether or not selection times vary when participants selected very thin and wider scrollbars, each with impenetrable borders. The results indicated that width manipulations did not influence selection time, thus designers could use very thin objects with impenetrable borders without slowing selection time.

Formative Analysis of Aging in Place: Implications for the Design of Caregiver Robots

Many have conceptualized caregiver robots as consumer products and studied elders’ perceived needs afor and preferences about such products. For reviews, please see (Broadbent, Stafford, & MacDonald, 2009; Jones & Schmidlin, 2011). That approach, though, could create robots that cannot satisfy elders’ actual caregiving needs. Alternatively, one can conceptualize caregiver robots as workers in complex socio-technical systems. To do so, one would need a detailed account of the caregiving that takes place in elders’ homes. Unfortunately, as noted in a National Research Council (2011) report, such a detailed account of caregiving does not exist. Accordingly, we sought to develop such an account. There are many ways to analyze work (for a discussion of general approaches, see Vicente, 1999). They can be categorized into 3 general types: normative, descriptive, and formative approaches (Vicente, 1999). We adopted a formative approach because formative approaches are tailored to the analysis of complex socio-technical systems (Vicente, 1999). They capture work requirements without specifying how that work must be done or who must do it. For example, the constraint “must not lose track of time” captures a work requirement but allows the associated work to be accomplished in a number of different ways (e.g., by checking a clock, setting an alarm) and by a number of different entities (e.g., family member, caregiver robot). To conduct our analysis, researchers observed caregiving in elders’ homes, and interviewed caregivers about their work activities. Researchers then organized their findings into an Abstraction Hierarchy (AH; Vicente, 1999), that is, a detailed account of the aging in place socio-technical system. Our primary aim was to create an AH that describes means-ends relations between the complex socio-technical caregiving system’s overall objectives, work tasks, and physical resources. Such a description provides a detailed account of the caregiving work domain, and serves as the foundation for subsequent formative analyses of caregiving. To create the AH, research team members completed 4 steps: 1) analyzing existing caregiving documentation, 2) observing caregiving and interviewing caregivers, 3) drafting and/or refining the AH, and 4) validating the AH. Steps 2 and 3 were iterative. This process is consistent with Naikar, Hopcraft, and Moylan’s (2005) recommendations regarding formative analyses. The AH made clear that caregiving for those who age in place is a complex and nuanced activity. More specifically, our analysis confirmed existing research regarding categories of caregiving tasks and revealed aspects of caregiving that have not been detailed so far. The existing literature indicates that caregivers assist older adults with self-maintenance activities of daily life (ADLs), such as eating, toileting, and dressing (Lawton, 1990), instrumental activities of daily life (IADLs), such as cooking, cleaning, and shopping (Lawton, 1990), and enhanced activities of daily life (EADLs), such as participating in social activities and pursuing hobbies (Rogers, et al., 1998). Our analysis confirmed those findings, and our AH provides a more detailed account of those tasks than was previously available. Our analysis also revealed aspects of caregiving for those who are aging in place that have not been detailed thus far in the research literature. For example, our AH contains a purpose-related function called Counseling, which concerns ensuring that the elder does not experience psychological distress. To perform this function, the caregiver must understand the elder’s situation (e.g., a family conflict), use information about that situation (e.g., experience with relevant family members and/or past conflicts; the elders’ past choices), and offer the elder advice about how to proceed (e.g., which family member’s advice to follow). The main implication of our AH for the design of caregiver robots is that such robots cannot be designed to perform purpose-related functions in a one-size-fits-all way; rather, caregiver robots must exhibit context-conditioned variability (Vicente, 1999). Our AH has many other important implications for the design of caregiver robots, which unfortunately cannot be detailed here due to space constraints.