Adaptive function allocation has been proposed to allow the advantages of task automation to be realized without some of the disadvantages of "static" automation. However, few empirical studies of the efficacy of adaptive allocation have been reported. The effects of adaptive function allocation on monitoring for automation failure during multi-task flight simulation were examined in two experiments. The first study examined the use of two methods of adaptive function allocation as a possible counter-measure to automation-induced monitoring problems. Subjects were required to perform a tracking and fuel management task while monitoring an automated system monitoring task for possible failures. For the "model-based" and "performance-based" adaptive groups, a single 10-minute block of fully manual performance on the monitoring task was allocated to subjects in the middle of a session. For the control group system monitoring was automated throughout all sessions. All three groups had low probabilities of detection of automation failures for the first 40 minutes spent with automation, before the adaptive function allocation change. However, detection probabilities were higher for both adaptive groups than for the control group following the function allocation change. The second study found that multiple adaptive changes (repeated function allocation) sustained these performance benefits over a longer automation period. These results clearly indicate that adaptive function allocation can improve detection of automation failures and that the improvement can be sustained over long periods of automation cycles.
Manufacturing automation has eliminated many traditional risks of injury, but new types of risks have appeared. In the United States only a limited amount of data is available on automated manufacturing-related accidents, or these data cannot be distinguished from general accident data. The purpose of this study was to collect data on accidents related to advanced manufacturing systems (CIMS). A questionnaire requesting information on one serious computer-integrated accident that occurred in the respondent's plant was used to collect the data. The presented results are based on the analysis of 85 cases. The cases included six fatal injuries, as well as several severe injuries that led to amputation. The results confirm the need to pay more attention to prevention of production disturbances in CIMS, especially in the case of material handling equipment. A relatively large amount of cases involving defeated safety devices calls for the design of intelligent safety systems that protect the human operator, but do not interfere with the work to be done, even during the disturbance situations.
Current evidence suggests that human error or impaired human performance is a significant causal factor in environmental accidents and incidents. Some data from studies and analyses, accident investigations, and research is used to emphasize the extent of the problems. However, there are existing laws, regulations, or policy that require consideration of human factors in prevention of environmental incidents. Some specific requirements related to human factors are discussed. Finally, some areas for human factors contribution to environmental protection are discussed. These include such things as human error/ accident audit reviews, human reliability analyses as a part of risk assessment, root cause analyses of accidents, fitness-for-duty assessment, health and safety training, and human factors in the design of future hazardous or toxic process systems.
The U.S. Nuclear Regulatory Commission (NRC) staff is presently reviewing human factors designs and development proposals for "advanced" reactors; plants that may be built in the future. One or more of these reactor designs will likely include some form of computerized emergency operating procedures systems (CEOPs) for use by plant operations personnel during transients and accidents. Twelve different CEOPs are known to be under development at the present time. These vary in terms of their philosophy (e.g. rule vs knowledge-based), their user interface, and their approach to function allocation and degree of automation. This study was undertaken to identify the salient characteristics of these systems in terms of their potential effects on control room crew performance and safe plant operation, as a first step in the development of guidelines by which the NRC can review such CEOPs prior to their implementation.
Government- and public-sponsored groups are demanding greater accountability by the Department of Energy's weapons complex. Many demands have focused on the development of a positive safety climate, one that not only protects workers ornate, but also the surrounding populace and environment as well. These demands are in part a response to findings which demonstrate a close linkage between actual organizational safety performance and the organization's safety climate, i.e., the collective attitudes employees hold concerning the level of safety in their organization. This paper describes the approach taken in systematically assessing the safety climate at EG&G Rocky Flats Plant (RFP).
This paper discusses the application of the methodologies of Hierarchical Task Analysis (HTA) and Tabular Task Analysis (TTA), incorporating Human Error Analysis during the detailed design phase of a new reactor. The MAPLE-X10 reactor, currently being developed by AECL Research, is a dedicated isotope production facility, scheduled for commissioning at Chalk River Laboratories, Ontario in 1994. The task analysis process developed for MAPLE-X10 consists of the representation and analysis of task data. The aim is to ensure compatibility between the design of MAPLE-X10 and the characteristics and capabilities of the diverse users of the system. Compatibility will lead to enhanced safety, operability and maintainability. Each stage of the task analysis process is described and discussed, emphasizing the practical application of Hierarchical Task Analysis for task representation, and Tabular Task Analysis for detailed analysis of tasks during which human error may have safety or production related consequences. The benefits of applying the methods of Hierarchical Task Analysis and Tabular Task Analysis to the MAPLE-X10 project are highlighted. These include clear representation of the organization of tasks and interactions between systems in the HTA, continuous feedback to design and operations personnel regarding identified mismatches between existing procedures and the design intent, and cost effectiveness. The multiple uses of the information elicited during the task analysis process are also discussed in this paper. These include design verification, the identification of training requirements and the development / verification of operating procedures. In addition the task analyses provide a framework for other assessments to be completed for the project, such as Human Reliability Analysis, Workload Assessment, Communications Analysis and Training Needs Analysis for Mature Operations.
A testbed was developed for studying the effects of different computer system designs on human-computer team problem-solving, using the real-world task of antibody identification. The computer interface was designed so that practitioners could solve antibody identification cases using the computer as they normally would using paper and pencil. A rule-base was then encoded into the computer such that it had knowledge for applying a heuristic strategy that is often helpful for solving cases. With this testbed, studies have been run comparing different computer system designs. A critiquing system was found to be better than a partially automated system on cases where the computer's knowledge is incompetent.
One of the most important questions facing developers of "human-centered" automation in air traffic control (ATC) is what and how much information does an ATC controller need to take an appropriate action or to make an appropriate decision, especially those with potentially catastrophic failure modes. To obtain such information one must develop a methodology that can be used to evaluate different information attributes and thereby collect data regarding what attributes the most experienced ATC controllers consider most/least important for decision making. A combination of methodologies for evaluating a set of information attributes required to identify and resolve aircraft conflicts and most importantly, specifying the interrelationships between these attributes for ATC problem solving is described. Eighteen attributes were chosen which covered those elements of information deemed most important in conflict identification and resolution and which reflected the thinking of a cross section of experienced ATC controllers. Multidimensional scaling (MDS) and rank-ordering techniques were used to determine the relative importance and interrelationships between attributes. The MDS and ordering results were then integrated into a unified conceptual framework. This framework reflected the importance of each attribute and thus, specified the way ATC controllers think about identifying and resolving conflicts. These results represent a step toward developing an empirical knowledge base for understanding the cognitive processes brought into action by controllers and also becomes a reference point for designing effective displays and decision support systems for ATC automation and to predict their value.
The 21st century Navy combatant ship will experience exponential increases in shipboard information to be processed, disseminated and integrated. High Definition System (HDS) technology will provide for the convergence of text, graphics, digital video, imagery, and complex computing to allow for a new range of advanced capabilities that exceed those of currently available workstations. These capabilities could result in unmanageable and overwhelming cognitive workloads for Navy tactical operators in CIC (Combat Information Center). For this reason, a prototype user interface was designed using future combat system requirements, proposed HDS capabilities, and human-computer interface design standards and principles. Usability testing of the prototype user interface was conducted as part of an effort to identify integrated information management technologies which reduce operator workload, increase human performance, and improve combat system effectiveness. This demonstration will focus on explanation and demonstration of future concepts envisioned for the AEGIS operational environment; organization and functionality of the menu structures and window contents; the usability testing methods utilized; results from usability testing; and plans for utilization of the prototype shell in other operational environments.
A system has been developed that graphically re-creates the radar data recorded at En Route air traffic control facilities. Each facility records data sent to the radar displays associated with the airspace under its control on a System Analysis Report (SAR) tape. SATORI (Situation Assessment Through Re-creation of Incidents) overlays the SAR data on the appropriate sector maps using map data from the Adaptation Control Environmental System (ACES) database. Voice records from the audiotape recording of the communications between the controller and pilots, and interphone communications are digitized and replayed in synchrony with the events displayed on the screen. Figure 1 details the SATORI data processing flow. The analog switch display settings of the PVD are not recorded, however subroutines have been written for SATORI that allow the display to be set up with the settings reported to have been used by a given controller. In addition to the above, SATORI has the capability to display the high and low weather intensity that was displayed on a given PVD. All software routines written for SATORI use Open Systems Foundation (OSF) technology. Similar data to those available from En Route facilities are recorded at TRACON facilities and should allow for the development of a re-creation tool much like the one discussed in this report. Once SATORI is evaluated, it will be possible to accomplish the goals of evaluating system designs, over-the-shoulder appraisals, training outcomes, and measuring controller performance.
The proliferation of digital avionic information presented to pilots has produced a critical need for intelligent avionic information management, particularly in the area of Caution, Warning, and Advisory (CWA) systems. This demonstration illustrates the role of an Active Man Machine Interface (AMMI) in the context of CWA systems in rotorcraft of the future. The basis of the AMMI's intelligence demonstrated here is provided by a cognitive model that 1) prunes the alert stream to only those messages that have meaning to the pilot depending on the tactical context, and 2) provide context-sensitive advice on the basis of the tactical context. The CWA AMMI is currently being designed using COGNET, an cognitive modelling methodology (Zachary, 1989), and implemented using BATON, a set of software tools used to implement and embed COGNET models into existing systems (Zachary et al, 1991).
Integrated Decision/Engineering Aid (IDEA) incorporates a standard process and a set of automated tools to support the application of the Department of Defense's Human/System Integration (HSI) program, the Army's Manpower and Personnel Integration (MANPRINT) initiative and Human Factors Engineering (HFE) throughout the materiel development process. IDEA provides the HSI/HFE analyst with guidelines, data and tools to integrate HSI/HFE into the acquisition of: (a) non-developmental items (NDI), (b) product improvements, and (c) new system developments, focusing on the activities and products at each phase of the materiel acquisition process. The purpose of the session is to demonstrate how IDEA is utilized in the definition of the HSI/MANPRINT requirements in support of the system development/acquisition process. Emphasis will be placed on a review of the overall architecture, arrangement and organization of the modules, and on the recent additions and modifications.
Blink-writer is a communications system created to help restore the ability to communicate to the severely neurologically impaired. The system has been used successfully on patients with amyotrophic lateral sclerosis (ALS), on stroke victims, on patients with high spinal cord injury, and on a patient who was paralyzed as a result of having contracted bulbar polio.
This paper discusses some of the more important issues concerning the development of interactive multimedia systems. Specifically, the design of a multimedia tour or interactive presentation of human factors and ergonomics at Virginia Tech is demonstrated.
The Image Access and Management System (IAMS) is being developed to provide easy access to photographic images taken by Space Shuttle astronauts. The photographs, coupled with a textual catalogue of descriptive characteristics, forms a significant database of information about the Earth's surface. The IAMS incorporates a video laser disc containing digitized images, an electronic database containing descriptive information about each photo, and software allowing the database to be searched and the images to be viewed. The demonstration presents the IAMS and the unique interface development issues involved with a photographic image/database system.
A demonstration of a new graphic notation system which expands the existing symbol set associated with operational sequence diagrams. The notation system is being developed as part of a computer-aided instructional program created for undergraduate students involved in human factor studies. The basic premise of the notation system is to enhance the practitioners understanding of human-machine processes during task analysis and operational sequence diagraming.
Despite the wide acceptance of standard user interface platforms, style guides, and software tools, many human-system interfaces are still difficult to use. An user interface design guidance system (UIDGS) is any collection of tools that supports a software developer or designer in creating human-system interfaces. Current UIDGSs are insufficient. Some criticisms of design guideline documents, a popular form of UIDGS, are that they do little to ensure task consistency (Potter et al., 1990), are not well-suited for use during development (Thovtrup and Nielsen, 1991), are often too general (Chapanis, 1990) and are perceived as limiting creativity. These factors have inhibited the adoption of UIDGSs within the mainstream of software development. The current research addresses this problem through demonstration of a prototype UIDGS.
The nation-wide and global need for environmental restoration and waste remediation (ER & WR) presents significant challenges to the analytical chemistry laboratory. The expansion of ER & WR programs forces an increase in the volume of samples processed and the demand for analysis data. To handle this expanding volume, productivity must be increased. However, the need for significantly increased productivity faces a contaminant analysis process which is costly in time, labor, equipment, and safety protection. Laboratory automation offers a cost effective approach to meeting current and future contaminant analytical laboratory needs. The proposed demonstration will present a proof-of-concept automated laboratory conducting varied sample preparations. This automated process also highlights a graphical user interface that provides supervisory control and monitoring of the automated process. The demonstration provides affirming answers to the following questions about laboratory automation: * Can preparation of contaminants be successfully automated? * Can a full-scale working proof-of-concept automated laboratory be developed that is capable of preparing contaminant and hazardous chemical samples? * Can the automated processes be seamlessly integrated and controlled? * Can the automated laboratory be customized through readily convertible design? * Can automated sample preparation concepts be extended to the other phases of the sample analysis process? To fully reap the benefits of automation, four human factors areas should be studied and the outputs used to increase the efficiency of laboratory automation. These areas include: 1) laboratory configuration, 2) procedures, 3) receptacles and fixtures, and 4) human-computer interface for the full automated system and complex laboratory information management systems.
The High Definition Systems Usability Test Tool (HUTS) was developed to demonstrate and assess different human-computer interface (HCI) concepts for generic operator console positions within an advanced naval tactical display environment. To facilitate a "rapid prototyping" approach to interface development, HUTT was developed as a general purpose tool. The HUTT can be used by the human factors engineer as a rapid prototyping tool enabling quick construction and evaluation of alternative interface concepts. Changes to the organization and operation of a prototype can be made quickly, ensuring that more time can be spent on evaluation as compared to development. The demonstration will focus on the uses and operation of the HUTT. Demonstration participants will be shown how to develop, install and modify the OSF/Motif compliant graphics (pull-down menu structure and window contents) and the functionality of the HUTT to reflect changes in interface concepts.
InterView is a software tool for designing, specifying and visualizing human-machine interfaces. The design of the tool began with two premises. The first is that, in order to adequately specify an interface, a designer must consider the inputs, outputs, and states of the system for which the interface is being designed. In this context, inputs refer to user-to-system communications, as mediated through buttons, keyboards, etc. Outputs refer to system-to-user communications, as mediated by displays, or other feedback. States refer to discrete conditions, either real or conceptual, in which the system may be at a given time. The second premise is that good, clear diagrams play an essential part in designing complex systems. Given appropriate diagramming techniques, it is much easier to describe complex activities and procedures in diagrams than in text. A picture can be worth more than a thousand words, because it is concise, precise, and clear. It does not allow the sloppiness and possibilities for misinterpretation that are common in text specifications.
Recording transcripts of human-computer interaction can be a very time-consuming activity. This demonstration presents a new technology to automatically capture such transcripts in Open Systems environments (e.g., from graphical user interfaces running on the X Window System). This technology forms an infrastructure for performing distributed usability testing and human-computer interaction research, by providing integrated data capture, storage, browsing, retrieval, and export capabilities. It may lead to evaluation cost reductions throughout the software development life cycle.
This session demonstrates a computer program, called MacISOKIN, that reveals where and how the physical environment constrains the movements of inhabitants. An associated research paper explains how ISOKIN measures of environmental constraint are related to the environmental satisfaction of office workers. MacISOKIN was recently developed to facilitate a research project sponsored by Steelcase, Inc. at the University of Michigan. The demonstration emphasizes how this computer-aided approach can assist designers and planners with the critical design tradeoffs inherent in limited space scenarios. MacISOKIN provides color graphics and numerical reports to identify areas in a floorplan where furnishings or the activities of other inhabitants are a likely source of interference. The "object-oriented" user interface allows the user to quickly change numerous analysis variables. For example, the impact of alternative layouts may be tested by adding, deleting or moving a wall or furnishing. Alternative activity scenarios may be explored by cutting and pasting "body-motion envelopes" from a library of pre-analyzed activities.
To be a "player" in today's concurrent engineering design environment, engineers must provide reasonable estimates of system-level effects for their recommendations. Human factors engineers (HFEs), who formerly relied on quoting military standards or design guides, now must support their recommendations with consequences. Although human factors maintainability estimate methods have been employed, most are considered cumbersome, labor intensive, and not responsive to the quick-paced program or design office. CAMT proposes a new way to collect, organize, and estimate human consequences of design alternatives. This method is being demonstrated in a series of CAMT feasibility studies by the Armstrong Laboratory Logistics Research Division.
The Posture Video Analysis Tool (PVAT) has been developed to meet the special needs of ergonomist and human factors analyst at NASA Johnson Space Center. Often times these specialist must attempt to evaluate microgravity working posture from video footage not specifically recorded for the purposes of quantitative analysis. The purpose for developing PVAT was to provide a structured methodology in which these specialists could optimize the data collection technique. The PVAT is designed such that microgravity postures can be documented while systematically observing footage of astronauts working in a space environment. PVAT is an interactive Macintosh menu and button driven Supercard prototype. Users are provided with a set of input parameters related to the microgravity environment and human performance issues. The primary inputs are: subject code, body orientation, targeted body part camera view (given subject location), body movement and rating level. A secondary set of inputs is available for users wishing to document extraneous behaviors or activities such as bending, reaching, interruptions, etc. These secondary behaviors may be documented as part of the primary inputs or independently. Each entry is time stamped and stored automatically. Provisions are made that allow users to pause, tag incorrect selections, enter an "unsure" response and user comments. Data output is saved as a "text file" using tab delimiters for easy importation into programs such as Microsoft EXCEL. Future PVAT modifications will include adding more input parameters, data reduction capabilities, control of the video deck from the application, and an animated postural glossary.
As a result of Human Factors evaluations of procedures associated with incidents at Rocky Flats Plant (RFP) it was determined that the existing procedure format created significant opportunities for confusion in their attempt to convey information about a work process. For instance, there was no mechanism to clearly identify the participants and their roles during the Instructions portion of the procedure. In addition, procedure authors frequently used complex logic to convey a series of contingent actions within steps. It was also difficult to discern the actual procedure steps from other types of information in the procedure. These and other inadequacies prompted the Human Factors Engineering (HFE) department to propose solutions to these problems that followed well-researched principles of cognitive psychology, dealing with how humans process information. Format and style contribute to procedure usability, and therefore to safety and efficiency in operations governed by the procedures. Since it was difficult to tie specific performance failures to specific format and style characteristics and thereby clearly define costs and benefits, it was difficult on that basis to sell the idea that changes in procedure format and style were really necessary to improve safety and efficiency. In addition, we found that the socio-political systems governing this process, particularly at the subprocess interface level, were not functioning efficiently. Both the technological aspects of the process and the socio-political aspects were contributing to waste and considerable re-work. Fixing the customer feedback loop to the process owners not only minimized re-work and waste, but also provided the data to persuade subprocess owners to make the necessary changes that heretofore were being met with great resistance.
The development of an operability model analysis tool to support design requirement definition for advanced ASW combat systems has evolved over the last four years. Previous NUWC ASW Combat System Operability Modeling efforts using SLAM II have produced comparisons between various ASW combat systems to examine the relative benefits of equipment features and operating concepts. Development of these operability models enabled the representation of contact data handling processes, ASW combat system operators, and the man-to-man, man-to-machine, and machine-to-machine interfaces that occur during an ASW mission. Early operability model studies (Stretton, Swiontek, Morris, Conway, and Wachter, 1991) investigated data handling bottlenecks and contact data throughput for ASW systems by focusing on the sensor operators, supervisory personnel, and command decision makers. Earlier models did not incorporate variations in team member productivity as might be found under high-workload conditions. Also, ASW team member skill levels were not varied as might be found due to latency from training-to-shipboard activities or to individual differences in operators and their training. As a follow-on effort, this study examined how variations in operator proficiency on critical tasks affect overall ASW combat system performance. Results of this effort appear to have wide-reaching operational impacts in the area of team performance and operator variability.
AT&T Video Ergonomics Evaluator System (AT&T-VEE) is a state-of-the-art computer analysis system used to analyze the hazards of a Manual Material Handling (MMH) task. This project was funded and managed by the American Telephone & Telegraph Company (AT&T) and designed and developed at Texas A&M University. The system is a lifting analyses tool for the 1992 NIOSH Lifting Equation to determine minimal user effort in lifting an object and to standardize the outputs. This system is now capable of providing the AT&T safety specialists with clear access to: complete NIOSH analyses, pictures of the lift environment, and background information on the MMH process. It was determined that a system, such as AT&T-VEE, was needed to reduce AT&T's MMH injuries and ensuing problems. The scope of the project was to develop a menu driven, user friendly system that would provide employees with the ability to determine the maximum weight of objects utilized at MMH tasks. In order to perform a lifting analysis, a video tape of a manual lift is first produced. The system user, in conjunction with a computer, enters employee's measurement landmarks and points of analyses, e.g., ankles, load center, weight, etc. AT&T-VEE produces a NIOSH Lifting Equation analyses, pictures and records of the MMH task.
To provide a forum to discuss how system development techniques and approaches which have proved successful for weapon systems can be applied to the design and development of non defense related systems.
From a human factors perspective, ATIS and CVO are key components of IVHS. But impressive technology does not guarantee consumer acceptance. Several issues concerning acceptance of new technology are discussed including product characteristics, consumer characteristics, and implications for CVO. Two experiments that evaluate driver acceptance are described.
This paper describes the outcome of a comprehensive literature review specifically conducted to assess research relevant to the design of advanced traveler information systems (ATIS) for automotive and commercial vehicle use. The goal of the literature review was to identify existing human factors guidelines applicable to the design of ATIS and gaps in the research precluding the development of comprehensive human factors guidelines for ATIS applications. A summary of ATIS research findings and research gaps is presented. Through presentation of both research findings and research gaps, it is hoped that interest will be sparked and focus will be provided for researchers in this growing application of human factors. Given the criticality of ATIS applications (both in terms of driver safety and public acceptance) and the visibility that will be provided to the human factors community from participation in ATIS design, human factors research leading to comprehensive and usable guidelines is extremely important.
This paper describes a recently completed survey conducted to determine the transportation community's perspective on Advanced Traveler Information System (ATIS) and Commercial Vehicle Operations (CVO) system development objectives and performance requirements. The paper describes the research methods, some general findings regarding development objectives, and a general discussion and examples of performance requirements.
Advanced Traveler Information Systems/Commercial Vehicle Operations (ATIS/CVO) are segments of IVHS currently being researched as a means of decreasing road congestion and increasing safety. Due to the complex information requirements for these systems, three surveys have been designed by University of Washington researchers and distributed nationwide to collect these requirements from the users: commercial drivers, dispatchers and private vehicle drivers. This paper discusses the methodology used to design the surveys and the effort to ensure that a representative sample was included on a nationwide basis. Approximately 8,300 surveys were distributed in person and 10,000 dispatcher surveys were distributed in a newsletter. Data estimation procedures will include modeling the influence of an in-vehicle system for route guidance and determining the significant impacts of an ATIS/CVO in terms of age, gender, income, and other socioeconomic characteristics.
Data were obtained in an operational air traffic control (ATC) setting from seven (7) FAA Air Route Traffic Control Centers (ARTCCs) for 65 air space sectors within those centers. Measures used for study analysis were obtained during 90 minute data collection sessions and included: (a) ATC system inputs in the form of time of day, facility location, air traffic volume and air traffic configurational complexity, (b) the presence of a single radar controller or a two-person controller team, (c) work activity parameters obtained as frequencies of occurrence for tasks involving communications, computer interactions, flight strip activities and handoffs, and (d) controller performance outcomes based on observed performance pressures exhibited by the controller. A hypothesized path (causal) model incorporating the above variables was constructed and tested for its explanatory value. Computations of direct effects within the model showed generally significant linkages between work activities and traffic volume, level of traffic complexity, and controller configuration (e.g. higher levels of traffic volume and complexity, and the presence of a controller team were associated with higher levels of task activity). Most consistently significant, however, in its causal linkages to work activity was air traffic complexity. This was the dominant predictor, by far, of rated controller performance pressure when all other variables -- including traffic volume -- were accounted for by the analysis method. The time of day and facility location variables were erratic in their effect and difficult to interpret. On the basis of the study results, it was recommended that further research focus on development of the application of air traffic complexity as a measurement construct of potentially unique importance; one which seems to have been little understood and generally neglected in the research literature.
The need to understand the design process in all its complexity is motivated by an interest in the development of tools and technologies that would be capable of aiding collaborative design teams. This development effort depends upon an understanding of design activities as they occur within a real world context. Observations of design activities that are made without direct communication with the design team members may fail to capture many of the subtler aspects of the process -- aspects that are best understood when described by the design team members themselves. In order to supplement observational studies, this paper presents a case study in which a dialog with members of a variety of collaborative design teams was established in order to elicit information about the nature of collaborative design. A knowledge acquisition technique, concept mapping, was used to achieve an understanding of the role of human factors specialists within the collaborative design process specific to the Air Force's system acquisition program. Results highlight various findings about the nature of design problem solving such as the way different organizational settings influence human factors input in the design process/product. The paper discusses the usefulness of concept mapping to capture in-depth design knowledge and how this type of knowledge complements other approaches to understanding design.
Product design is often viewed as being a heterarchical and iterative process, possessing both systematic and chaotic qualities. However, a common denominator across all design activities is the access and utilization of information. In today's computer-aided design market, most of the available tools are narrowly focused on specific computational details for individual stages of design. Aids are needed to support information access and utilization during all stages of the design process. The application of human engineering and ergonomics data by designers is an increasingly challenging problem. Locating and understanding relevant information so that it can be applied to specific design issues is difficult given the abundance of existing and new data available. This is further complicated, in that the data are typically written to communicate research results to other human factors specialists. A new software product, Computer Aided Systems Human Engineering: Performance Visualization Subsystem (CASHE:PVS), is described which will assist the designer during the decision making process maximizing creative and analytical abilities while minimizing costs due to design time and errors. The software contains several features to enhance the designer's ability to interpret and apply the human factors data available in the product. Phenomena descriptions in text, figures, and tables are combined with experiential information via simulations, animations, and audio. This provides the user a unique and rich understanding of human performance phenomena and how they relate to the design of new products.
There are many problem-solving tasks that are too complex to fully automate given the current state of technology. Nevertheless, significant improvements in overall system performance could result from the introduction of well-designed computer aids. A major concern in the introduction of such tools to support problem-solving, though, is the potential to introduce new errors due to the interaction of the person with these computer support tools. We have been studying the development of cognitive tools for one such problem-solving task, enroute flight path planning for commercial airlines. Our goal has been two-fold. First, we have been developing specific system designs to help with this important practical problem. Second, we have been using this context to explore general design concepts to guide in the development of cooperative problem-solving systems. These design concepts are described below, along with a discussion of two empirical studies.
This paper describes a recently completed project conducted to identify functional requirements and software features of a Procedure Software Tool (PST) to be used as an aid in the preparation of written maintenance procedures in nuclear power plants. Results discussed include an overview of functional requirements and software features that address problems in procedure use and the process of procedure development. An overview of the recommended PST development process is also provided, along with a summary of the estimated benefits of PST development. Finally, potential applications in the nuclear industry, as well as other industries, are discussed.
The purpose of this study was to determine the effects of three variables on target designation tasks using three-dimensional (3-D) maps. Variable 1: two continuous control techniques, Variable 2: two aiding techniques, and Variable 3: two target densities. The two continuous controllers used were an ultrasonic hand tracker and a joystick. The two aiding techniques were referred to as simple and enhanced aiding. Simple aiding consisted of a color change to the target when the cursor penetrated the target volume. Enhanced aiding consisted of an algorithm (Osga, 1991) that "pulled" the cursor to the closest target and then changed its color. The two target densities were high (16 targets) and low (8 targets). Results showed that the hand tracker provided the best performance with respect to total target designation time. Enhanced aiding provided the best performance in terms of both total target designation time and percent errors. For the density variable, the common speed-accuracy trade-off was present -- the low density condition had faster total target designation times, but the high density had less percent errors.
This symposium addresses the characterization of human decision making within a complex environment for the purpose of developing improved decision support systems. All of the work presented in this symposium was conducted under a Navy research program entitled "Tactical Decision Making Under Stress" (TADMUS). The overall objective of the TADMUS program is to improve tactical decision making of anti-air warfare (AAW) crew members within the Aegis cruiser's combat information center (CIC) under conditions of stress and uncertainty. The unique aspect of this effort is that each presentation addresses decision making behavior, within a single domain, from a different perspective. The goal of each effort is to characterize some aspect of expert decision making performance within the AAW task environment, and to make recommendations for the resulting decision support system design based upon these characterizations. The result is a multi-faceted, human-centered approach to information organization and interface display design for a decision support system.
Recent research on decision making has focused on complex real-world domains and experienced decision makers, in contrast to artificial tasks and novice subjects (e.g., Klein et al., 1993). This research has produced a revised view of decision making, one that stresses knowledge-based, recognitional processing rather than general-purpose, analytical methods. Nevertheless, a full understanding of recognitional processing and its implications for the design of decision aids has not yet been achieved. In particular, the active, controlled aspects of recognitional processing have not been fully accounted for in approaches that stress relatively automatic pattern recognition as the basis for responding.
This paper presents the methods and findings of a study designed to identify the decision requirements for anti-air warfare officers in the Combat Information Center of an AEGIS cruiser. Decision requirements include the decisions that systems operators make, the cognitive strategies they invoke to make these decisions, and the cues and factors essential for making these decisions. These requirements can be used to design training, human-computer interfaces, or decision supports. The researchers adopted a method based on Naturalistic Decision Making (NDM) research. NDM describes how people make decisions in real-world settings under conditions of time pressure, high risk, and ambiguity. This paper describes a process for obtaining data necessary for describing these decision processes. The central method is a semi-structured interview method, the Critical Decision method (CDM). CDM was used to interview 31 experienced AEGIS personnel resulting in 14 incidents that reflect real problems experienced by the operational fleet. Analysis of these incidents revealed 183 decisions. Of these, 103 concerned situation assessments (SA). The operators used feature matching and story building to make all SA decisions. The operators invoked recognitional strategies to generate 95% of the course of action (COA) options and compared multiple options in only 5% of the COA decisions. The findings reported here indicate that under conditions of time pressure and ambiguity: decision makers rarely use analytical decision strategies, they usually satisfice rather than optimize, they rely heavily on diagnostic decisions, and they invoke singular rather than comparative evaluations of courses of action.
Human tactical decision making in Naval Anti-Air Warfare (AAW) is time-critical and is performed in a multiple-task, team-based environment. These aspects make this domain extremely challenging for traditional cognitive modeling techniques. The COGNET (COGnition as a NEtwork of Tasks) framework, however, is inherently designed for real-time, multi-tasking work, and, with extensions to accommodate team decision processes, proved suitable for modeling AAW decision making in the Navy's Tactical Decision Making Under Stress (TADMUS) program. A COGNET model of AAW domain expertise is described, along with Decision-Support System (DSS) design principles derived from the COGNET AAW model and the underlying COGNET framework.
This presentation summarizes the results of an empirical study examining human judgment bias under conditions of uncertainty and time pressure in surface Anti-Air Warfare (AAW). A substantial body of research has demonstrated that humans apply a limited set of heuristics to simplify decision making in complex and ambiguous situations. Most of this research, however, has used college students making logical, but unfamiliar judgments. This study was designed to assess whether Naval personnel, trained and experienced in AAW operations, exhibit these biases when performing their normal duties. Specifically, we studied whether the judgments of Naval tactical action officers in a realistic task simulation exhibit characteristics of the biases of availability, representativeness, anchoring-contrast, and confirmation. Our prediction that experienced subjects would disregard lack of reliability in otherwise representative data was only partially supported by the study. On the other hand, each of our other predictions was strongly supported. Our subjects ignored baseline trends when other case-specific information was available (representativeness and availability). They were significantly influenced by the order they received evidence, showing a recency effect characteristic of contrast. Additionally, as is characteristic of confirmation bias, they recalled much more of the information that was consistent with their final hypothesis and evaluated it as more informative than the inconsistent data, regardless of which hypothesis they had adopted. Implications for Naval decision support systems information and display are discussed.
In the commercial design of human computer interfaces, two of the most important goals for development of a new or enhanced business system or product is to increase the company's profit and market share. The most practical way to accomplish these goals is to reduce the project development cycle to yield a higher return on investment. A balance must be maintained to ensure that the reduction in cycle time does not compromise users' needs. Reducing development time and cost will have minimal effect if users will not accept the final product. During the past decades, new life cycle development techniques and tools have evolved to reduce the development cycle and increase user acceptance. Since the mid 1980's Joint Application Development (JAD) techniques, Decision Support Center (DSC), rapid prototyping and CASE tools have been introduced. These techniques and tools have focused on improving the front-end development process to understand the requirements and design prior to implementation.
This study examined the effect of concurrent and retrospective thinking-out-loud (TOL) on the frequency and value of user verbalizations during a software usability test. Three groups of users first learned to use an off-the-shelf database management package by means of a short tutorial and then engaged in six structured tasks. Users in the Concurrent condition thought-out-loud while performing the six tasks, whereas those in Retrospective-Immediate and Retrospective-Delayed conditions thought-out-loud while watching the videotape of their interaction with the software. Results indicated that there were no significant differences among the three conditions in performance or subjective evaluation of the software. More importantly, a verbal protocol analysis revealed that users in the Retrospective conditions spent more time making statements which had high value for designers than in the Concurrent condition. The value of verbalizations generated by the Retrospective conditions were not impacted by the 24 hour delay. The results were interpreted within context of differences in workload and in terms of the trade-off between increased value gained by using the retrospective paradigm versus increased cost of additional time to conduct the usability test.
While pilots have an intuitive understanding of situational awareness (SA), researchers have had difficulty defining and measuring SA for practical ends such as pilot selection, training and cockpit design evaluation. To achieve the latter end a model of SA is proposed and a measurement technique is described. Specifically, SA appears to involve the development and maintenance of a highly dynamic mental representation of critical aspects of the flying environment. Using this system the pilot makes judicious decisions in a timely manner with little conscious effort. To render SA measurable, SA may be modeled as a computer system with elements and processes which serve to develop and maintain an extremely fast, efficient database. Computer systems are assessed via benchmark tests. To measure SA, the pilot flies a benchmark test which consists of a standardized mission in a medium fidelity simulator under progressively increasing assigned airspeeds. The pilot's SA is measured via techniques including objective metrics such as flying performance and responses to "unexpected" events, as well as subjective metrics such as ratings and knowledge elicitation. All must of necessity be indirect measures of SA for the pilot's actions reflect not only SA but the adequacy of the data initially available as well as the quality of training, talent, and the vehicle itself. A statistically weighted combination of these measures is used in order to improve sensitivity and minimize their individual limitations.
Heuristic Evaluation has been shown to be a quick cost-effective methodology that can lead to early identification of many of the same user interface errors as laboratory usability studies. In this paper, we describe a method designed to enhance the performance of expert, system developer, and non-expert evaluators. The evaluators most proficient at Heuristic Evaluation are Human-Factors Experts (Desurvire, Lawrence and Atwood 1991; Desurvire, Kondziela and Atwood, 1992; Jeffries, Miller, Wharton and Uyeda, 199i) and most notably, "double experts" (Nielsen, 1992). Similar results were obtained for the Cognitive Walkthrough developed by Lewis, Polson, Wharton and Rieman, 1990 (Desurvire, et al., 1992; Jeffries, et al., 1991). We were interested in whether a non-empirical method could be developed in which evaluators other than Human Factors Experts can perform nearly as well as Experts. Desurvire, et al. (1992) found that Heuristic Evaluation and Cognitive Walkthroughs not only predicted problems observed in laboratory studies but also encouraged evaluators to suggest improvements. In addition, non-empirical methods stimulated evaluators to point out problems that would be likely to occur in actual use, but would not be observed in laboratory studies. We were interested in expanding this finding by developing a method that encouraged a broader scope of thinking, and thus a broader evaluation. In this paper, we describe the method Programmed Amplification of Valuable Experts (PAVE) and how it enhanced the performance of System Developers and Non-Human-Factors-Expert evaluators. Future work is discussed in which real users in the field will be compared to these results.
A model-based method for assessing the usability of graphical, direct manipulation style interfaces was developed and applied to a military airspace scheduling system. The method involves collecting and integrating verbal protocol data and mouse/keystroke files, and having an analyst familiar with the task, the data, and Norman's (1986) user activity model review the data and make determinations on what the data mean in terms of the model. A hierarchical encoding scheme based on the model is then applied to the integrated data to structure the human-computer interaction (HCI) process at a detailed interaction level. Meaningful patterns can be identified, frequency of events per task, and number of actions per intention can be calculated at various levels in the hierarchical breakdown, highlighting potential usability problems or instances of indirectness. Repetitious sequences, for example, could imply a missing high-level task domain object or an inability to group objects for application of a single action. Detailed model-based error encodings reflect user-system interface difficulties not only in the execution stage of HCI but in the psychological stages as well. The types of usability problems identified and the advantages of the method arc discussed. Based on these results, we have begun developing a multi-media tool to support application of the method.
The purpose of the present study was to test and evaluate three on-screen keyboard arrangements with indirect input devices. Studies conducted for hard keyboard arrangements have considered various factors affecting typing; however, differences between the nature of the hard and on-screen keyboards tasks preclude extrapolation from hard keyboard studies to on-screen keyboard designs. In this study, Finger Placement and Non-Finger Placement typists provided data for Stimuli (Word vs. Non-Words), Devices (Mouse vs. Arrow Keys), and Keyboards (1-Row Alphabetical, 3-Row Alphabetical, and QWERTY arrangements). The primary data collection tasks were two movement tasks and a typing task. The typing task consisted of having the user type a given Stimulus utilizing one of the on-screen keyboard arrangements and input devices. The movement task served as a control for movement time in the typing task. At the conclusion of the study, users were asked to rank order their preference for keyboard arrangement and input device. The QWERTY keyboard resulted in the fastest overall input times, and was the most preferred arrangement overall. Interaction between Device and Keyboard showed that with the mouse, input times for the 1-Row Alphabetical were slower than the QWERTY; whereas, with the arrow keys, input times were equivalent. However, this change in relative performance under the 1-Row Alphabetical arrangement for the two devices can be simply attributed to movement time. After statistically removing the effects of movement time from the typing task, the 1-Row Alphabetical arrangement was equivalent to the QWERTY for both input devices. Conclusions suggest potential incompatibility between the mouse interface and the 1-Row Alphabetical arrangement used in this study.
Meta-analytic procedures were used to cumulate research findings across studies of multiple-task performances published between 1965 and 1985 (inclusive). Exhaustive literature search yielded 94 reports of studies of multiple-task performances that were sufficiently detailed to be statistically combined using the Hunter-Schmidt-Jackson procedure. These studies yielded a total of 202 statistical values and a total subject N of 3091. Meta-analysis was performed overall (resulting in a significant test statistic) and then as a function of secondary task (13 types), primary task (18 types), and specific secondary-primary task combination. (The secondary tasks were Choice Reaction Time (CRT), Classification, Detection, Memory, Mental Math, Monitoring, Motor Response, Problem Solving, Reaction Time, Shadowing, Speech, Task Battery (Multiple Tasks), and Tracking; the primary tasks were Association, CRT, Classification, Decision Making, Detection, Driving, Flight (Simulation), Memory, Mental Math, Monitoring, Motor Response, Problem Solving, Reading, Shadowing, Sketching, Speech, Tapping, and Tracking.) Ten external variables (Carbon Monoxide, Control-Display Compatibility, Drug Use, Feedback, Intelligence Tests, Neuroticism, Noise, Rating of Job Performance, Sinus Arrhythmia, and Temperature) as correlates of multiple-task performances were also examined. A generally positive relationship between secondary task performance alone and in combination with the primary task was found, along with moderating effects of secondary task type and primary task type. No substantial variation of results within a given combination of secondary and primary task types occurred in most cases, although four external variables (Carbon Monoxide, Feedback, Noise, and Temperature) displayed within-factor variation of results. All results were discussed in terms of integration of the past literature and directions for future research.
Human factors practitioners are often concerned with defining and evaluating expertise in complex domains where there may be no agreed-upon expertise levels, no single right answers to problems, and where the observation and measurement of real-world expert performance is difficult. This paper reports the results of an experiment in which expertise was assessed in an extremely complex and demanding domain -- military command decision making in tactical warfare. The hypotheses of the experiment were: 1) command decision-making expertise can be recognized in practice by domain experts; 2) differences in the command decision-making expertise of individuals can be identified even under conditions that do not fully replicate the real world; and 3) observers who are not domain experts can recognize the expert behaviors predicted by a mental-model theory about the nature of expertise. In the experiment, the expertise of military officers in developing tactical plans was assessed independently by three "super-expert" judges, and these expertise-level ratings were correlated with independent theory-based measures used by observers who were not domain experts. The results suggest that experts in a domain have a shared underlying concept of expertise in that domain even if they cannot articulate that concept, that this expertise can be elicited and measured in situations that do not completely mimic the real world, and that expertise measures based on a mental-model theory can be used effectively by observers who are not experts in the domain.
The present study develops a methodology for design and usability evaluation of work control documentation for aircraft inspection based on the application of human factors research in the areas of information design and aircraft inspection. A taxonomy for design of usable documentation was developed using four basic categories: Information Readability; Information Content; Information Organization; and Physical Handling and Environmental Factors. Within the framework of this taxonomy the existing work control documentation for two extreme representative conditions of aircraft inspection tasks, the A-check and the C-check was evaluated for usability. Issues for workcard redesign were identified within this taxonomy using data from usability evaluation and these were then formulated into a set of 49 guidelines for design. These guidelines were then used to develop alternate design solutions offering improved usability. The increase in usability of the redesigned documentation was measured in an on-site empirical evaluation and proved significant. This methodology is currently being extended to the design of usable information for the design and evaluation of portable computer based documentation for aircraft inspection tasks.
Drug abuse is prevalent and impacts all aspects of society, including the workplace. Testing for impairment due to alcohol or other drug use has become commonplace in organizational settings. Despite potential problems with accuracy, the most commonly used method of screening is urinalysis. As drug testing becomes more common, concern over employees' rights to privacy and fairness of drug testing has also increased. Performance testing has been proposed as a solution to the problems associated with urinalysis, for jobs where eye-hand coordination is critical to job performance. Performance testing is based on the compensatory tracking task and involves correcting for the unpredictable movement of a visual stimulus on a computer screen. Although research suggests that urinalysis testing is perceived as fair in some situations (e.g., where need is high such as in safety-sensitive jobs), no research has been conducted which examines the perceived fairness of performance testing. The purpose of the current research was to examine the effects of type of testing (urinalysis versus performance testing) and need for testing (high versus low accident history) on perceived fairness and invasion of privacy. In addition, the effects of fairness and invasiveness on the willingness to apply for a job in the company and acceptance of the screening policy was examined. One-hundred adults participated in the study. Results indicated that urinalysis screening was perceived as more fair than performance testing. No effects of need (accident rate) were obtained for either fairness or invasion of privacy. Multiple regression analysis indicated that both fairness perceptions and invasion of privacy predicted willingness to apply for a job and acceptance of the policy. Implications of the results and suggestions for future research are discussed.
The issue of verifying and validating complex systems based on human factors criteria is becoming widely recognized. The need has become particularity significant with the development on the highly automated systems currently being developed for the new air traffic control systems around the world. This panel discusses several issues that have evolved from an international working meeting on the topic.
Recent reviews of team training and team performance have drawn at least one common conclusion: Much more research needs to be conducted on the process of and methods for measuring and evaluating teamwork. The current absence of reliable, valid, and robust measures of teamwork hinders not only the development of optimal collective training strategies, but also team-oriented research, human resource management efforts, and organizational and systems development programs. The objectives of this symposium are to (a) present an overview of some key theoretical and methodological issues which confront researchers and practitioners concerned with teamwork, and (b) introduce several innovative concepts, methods, and procedures which can aid in solving some of the problems associated with the measurement and evaluation of teamwork. These objectives will be met, in part, through presentations which describe and discuss (a) a conceptual framework for analyzing and interpreting measures of collective behavior, (b) the application of complexity theory-based measures to team and organizational performance, (c) a measure of teamwork based on the confidence team members have in their individual and collective capabilities, and (d) a system which supports the measurement and evaluation of unit collective performance in a simulation network environment. (The essence of these presentations are captured in the four papers which follow this abstract.) The discussant will provide his summary of the key issues addressed, as well as some that were not addresses. Finally, inputs and active participation of the audience will be solicited. The desired impact of this symposium is a heightened awareness and appreciation of the need for more research on teamwork measurement methods and procedures. The long-term goal of the symposium is to stimulate that research.
The objective of performance measurement is to assess the operational effectiveness of an individual or unit: use of resources to accomplish operational objectives. A continuum of collectives from groups to teams on the basis of organizational attributes is proposed. A shared plan of operation is proposed as the organizing agent of collective performance. Tactical decision making is defined in the context of adaptive problem solving. Although decisions are implemented as procedures, the procedures must be improvised in the situation and adapted to local conditions. Tactical Procedure Scripts are proposed as a structure for organizing performance data into operationally meaning segments for analysis, evaluation, and after action review. Operational and functional processes of GNUTS are differentiated and a three level process model is outlined. A domain model for tactical decision making and the use of behavioral objectives in a test plan are discussed.
The development of techniques for measuring complex task performance continues to be a problem area both in government and industrial contexts. This problem becomes even more critical as emphasis moves from individual, to group and/or team performance. This paper suggests a variety of measurement techniques that have been applied to the assessment of individual complex performance, and comments on the use of these (and similar) techniques with respect to the measurement of team performance. The measures are derived from the study of "Complexity Theory" (cf. Streufert and Swezey, 1986), and include consideration of both strategic and general planning activity, as well as the diversity of action required in complex team performance situations.
A fully portable quantitative EEG assessment system was used to evaluate workload in an advanced technology aircraft simulator. Air refueling and landing approach tasks were each performed at two difficulty levels in 15 Air Force pilots. Averaged and trended EEG spectral data were compared in the 8-12 Hz band to identify functional requirements for increased workload within and between tasks. A progressive suppression of 8-12 Hz activity at medial and right parietal sites accompanied increased workload in the air refueling task, while a sustained suppression at right and left temporal sites was associated with increased workload in the landing task. These findings suggest a potential electrophysiological index for workload. They also identify specific and differential cortical responses to visual integration in air refueling and working memory in ILS approach as primary correlates of the cognitive requirements for these tasks in these subjects.
The perceptions that team members have of each other's capabilities and the accuracy of those perceptions are very important in determining how well a team will perform. This paper describes a method for assessing, and the consequences of, team members' confidence in the capabilities of their peers. The mutual judgments of confidence are derived from knowledge each team member has of the capabilities of others to perform their respective tasks and to coordinate their actions with those of other team members. Implications of the accuracy of this knowledge and its impact on collective team training and performance are discussed.
The Unit Performance Assessment System (UPAS) was developed to support collective performance measurement in the simulation networking (SIMNET) environment. The UPAS collects network data and integrates these data with data on unit plans, terrain features, and communications to provide a more complete picture of performance than is possible with network data alone. This paper illustrates how UPAS data displays can be used to assess the performance of units in accordance with standards from the U.S. Army Mission Training Plan document for tank platoons.
This paper describes an ongoing program of real-time, interactive air traffic control simulations to test and evaluate the feasibility of conducting closely-spaced, independent, instrument approaches to dual, triple, and quadruple parallel runways. Real-time simulations are used to test and evaluate the human, technological, and systemic issues critical to the implementation of independent multiple parallel approaches. These simulations quantitatively evaluate controller performance, pilot/aircraft performance, and airport capacity issues. Simulation data are also used in the assessment of risk associated with the operation. To address human capabilities and limitations, qualitative analyses are used to assess the opinions of controllers, flight crew members, simulation technical observers, and recognized subject matter experts. This paper describes the planning, implementation, data collection, analysis, and related issues involved in real-time simulation for the development of national standards for independent multiple parallel approaches.
The Federal Aviation Administration (FAA) is currently involved in a tremendous effort to upgrade the nation's Air Traffic Control (ATC) system. Included in this effort is a new communications system called the Voice Switching and Control System (VSCS) for use primarily in the Air Route Traffic Control Centers (ARTCCs). The critical nature of the communications system, the complex Computer Human Interface (CHI) and functionality, and technology differences from the existing communications system caused the FAA to involve an Air Traffic Controller user group early in the design process. This group of subject matter experts was intended to help ensure adequate operational testing which would help secure user acceptance when VSCS was fielded. This paper discusses the methodology developed for the operational evaluation of the prototype VSCS in the Design Competition Phase (DCP) of the program. The methodology exposed problems in the areas of specification compliance, system stability, voice quality, CHI and operational suitability beyond those identified by the formal development test program called Factory Acceptance Testing (FAT). It was discovered that individual system functions that worked properly when exercised in isolation during FAT did not always work properly when exercised in realistic operational scenarios by controllers. The comparison of results between FAT and the operational evaluations show that controller evaluations should be included throughout the development process of FAA programs, and that the methodology used plays a fundamental role in uncovering issues.
The standardization of test methods for the evaluation of aircraft cockpit displays is an area display designers need to investigate. Comparable simulation facilities, and experimental protocols including tasks, performance parameters, data analysis techniques, and subject pools, need to be employed across military and civilian research laboratories to ensure that the results of simulation efforts will be interpreted similarly by all researchers and designers. This paper reviews the types of tasks and data collection, reduction, and analysis techniques used by researchers during a five-year Air Force research program designed to: 1) develop a standard head-up display (HUD) symbology set for use as a primary flight reference during instrument flight, and 2) develop a standard symbology set to be used as a baseline for comparing other HUDs. The program objective was met. The symbology set will be included in a revision of Military Standard (MIL STD) 1787, Aircraft Display Symbology.
This research examined the role of strategic and automatic processing in the acquisition of cognitive skill. These factors were dissociated by manipulating the predictability of Consistently Mapped (CM) and Variably Mapped (VM) stimuli in a memory search task. In blocked CM and VM conditions, subjects could choose different strategies for each condition. However, in mixed CM and VM conditions, the subjects had no basis for choosing differential strategies. Subjects who received mixed training exhibited less skilled performance than subjects who received blocked training. Moreover, transfer conditions revealed what appears to be a critical interval for learning to use differential strategies. The implications for part-task training are discussed.
This paper is based on results from an on-going effort sponsored by the US Army Research Institute (ARI) concerned with human performance and training issues in automated, near-real-time air defense command and control systems. Air defense command and control is the specific applications context, but the paper's implications extend to many contemporary process control settings. Topics that are addressed in the paper include: (1) human performance problems associated with automation, (2) a new look at human performance requirements in near-real-time process control, (3) and training and job performance support requirements for supervisory controllers. The "new look" portion describes a reasonable and evolving concept for human participation in automated processing, designated Rule-Based Supervisory Control. The paper is intended to introduce these topics to concept developers, system designers, and trainers dealing with automated process control technology.
An Active Interlocked Modeling (AIM) Dyadic protocol for training complex skills, an AIM Tetradic protocol, and an Individual Control protocol were tested alone and in combination with a Multiple Emphasis on Components (MEC) protocol creating 6 conditions for training a complex computer game. We randomly assigned 120 paid subjects to the six conditions. Total game score improved over 10-1 hr sessions for all conditions. Improvement rate replicated advantages previously reported for AIM Dyad, AIM Tetrad, and MEC over the Individual Control. The AIM Dyad with MEC was better than either the AIM Dyad or the Individual with MEC. The AIM Tetrad with MEC was worse than either the AIM Tetrad or the Individual with MEC. Similar patterns occurred on retention, transfer, and resistance to secondary task interference. We discuss implications for acquiring and automatizing attention control strategies through observational learning.
A cognitive model of instructor Crew Resource Management (CRM) assessment was developed as an integral component of a prototype CRM assessment expert system. The cognitive model provides an instructor-centered approach to CRM assessment in the multi-tasking, time-constrained environment of recurrent Line-Oriented Flight Training (LOFT). This cognitive modeling is based on systematic analysis of how experienced instructors are able to attend to key crew behaviors and derive consistent CRM assessments in the complex simulator-based training environment. Concept sorting data, collected from recurrent training instructors, was subjected to hierarchical cluster analysis. The analysis identified the mental organization or knowledge structures required to make efficient CRM assessments in the time-constrained LOFT environment. There were three primary assessment clusters that experienced instructors had in common. These clusters of assessment concepts were used to develop the architecture and content of the assessor modules for the prototype CRM assessment expert system. That expert system provides an efficient CRM assessment that is similar to the process and output of the experienced recurrent training instructor. These findings have operational implications for the feasibility of an instructor CRM assessment tool and for making the CRM assessment process more systematic.
The purpose of the present study was to examine the effects of providing two simple tactical aids designed to bolster cognitive ability while problem-solving. After receiving computer-based instruction, subjects used either an orientation aid, a short-term memory aid, both aids, or no aids to perform a computer-based electronic problem-solving task. A major goal was to determine whether the effect of each aid on performance conformed to either a compensatory model or a training model. Results were mixed. That is, some evidence supported the compensatory interpretation, while other evidence supported the training model interpretation. Specifically, an interaction between short-term memory scores and use of the short-term memory aid showed that the aid helped low scoring subjects solve problems quicker when they used the aid more. However, most of the evidence suggested that, in terms of the time needed to solve problems, the effects of both aids were characteristic of a training model.
There is no question that issues surrounding effective, coordinated team performance are considered topics of vital importance in today's organizations (Salas, Dickinson, Converse & Tannenbaum, 1992). This is particularly true for teams that must operate in stressful, unpredictable circumstances. Indeed, the increasing complexity of the workplace has created a dependence on teams who must often perform under stressful conditions (e.g., aircrews, firefighters, surgical teams). The effectiveness of such teams depends on the coordinated and cooperative activity of the individuals who comprise them. These individuals must often operate under hazardous conditions characterized by heavy workload and time pressure. Research has shown that these stressors severely restrict the ability of team members to communicate with each other. In effect, reducing the ability of the team to coordinate effectively (Kohn, Kleinman, & Serfaty, 1987).
Teams with records of superior performance have one common critical characteristic: they are extremely adaptive to varying task demands. These teams were observed to switch between several different coordination strategies and organizational structures, with different lines of authority, communication patterns, and task responsibilities, as they move between normal operations and high-tempo or emergency situations. Two questions are central to the issue: What are the effects of external stressors on team performance, and what are the mechanisms by which teams of decision-makers cope with stress? Our main hypothesis is that team coordination strategies evolve from explicit coordination under low workload conditions to implicit coordination as workload increases. To illustrate these ideas, this paper presents findings from an experimental study on the effects of stress on the performance of command teams. The computer-based experimental task simulates operations in a naval environment in which a hierarchical team of four decision-makers must coordinate complex and ambiguous information to make identifications on air targets. Three task-related stressors -- time-pressure, uncertainty, and ambiguity, -- and one information-structural variable were manipulated in a within-subject, full-factorial design. Results show some complex patters of the way the different stressors combine to generate stress and affect the team decision and coordination strategies. Implicit coordination patterns, anticipatory behavior, and redirection of the team communication strategy are evident under conditions of increased time-pressure. Discrepancy between the subordinates' and the team leader's mental model of the costs of errors generates non-trivial patterns of error-making in the teams. The team leader's periodic update had a stabilizing effect on the team communication strategy. Different implementations of team training interventions to enhance mutual anticipation, prevent inadequate adaptation to stress, and foster implicit coordination in command teams are proposed.
Recent empirical studies of decision making in teams demonstrate that team structure and workload significantly influence team performance. In many operational environments, however, it is impossible to change these factors, even as a mechanism for enhancing team performance. Therefore, it is necessary to create training interventions that will optimize performance within existing team structures and workload levels. Several studies suggest that team processes are the most likely target for this type of intervention. The current investigation sought to develop a laboratory analogue of a common team structure (i.e., the "product team") and to assess the effects of high and low workload on team performance processes within this structure. The results suggest that different communication behaviors facilitate effective performance under low and high workload.
Critical decisions are made every day by teams of individuals who must coordinate their activities to achieve effectiveness. Recently, researchers have suggested that shared mental models among team members may help them to make successful decisions. Several avenues for training shared mental models in teams exist, one of which is training in planning behaviors. The relationship between team planning, team shared mental models, and coordinated team decision making and performance is explored.
Understanding teams and team functioning is more important today than ever before. Recent economic conditions and technological developments have made team work increasingly critical in a wide array of settings. This increased level of dependence on the team as an operative unit brings new insight and in turn new questions, regarding the internal mechanisms and operational effectiveness of the team. In particular, how and why do different types of teams function differently and what kind of training is best suited for what type of teams? To a large extent, the nature of a team and its degree of role-related interdependence may regulate the amount of interactional requirements, knowledge, and therefore training, necessary for each member to perform optimally. Specifically, as team member roles become more highly interdependent, the level of each member's knowledge about his fellow team mates' roles becomes increasingly important. This knowledge becomes even more critical under stressful, fast-paced conditions. Although little attention has been given to workload in conjunction with team processes or team performance, early teamwork studies demonstrate that communication and overall performance decline as workload is increased (Briggs & Johnson, 1967; Hemphill & Rush, 1952).
The design of training for large groups or collectives demands a focus on group performance that differs from a mere aggregation of individual, crew, and smaller team performances. Emphasis on the training of large groups has broad military and non-military applicability, especially where success depends on complex interactions and multiple subgoals. Five panelists describe and discuss training system design factors and "lessons learned" as they relate to issues in training large groups. The session features a discussant with extensive experience in training research and development, and includes interactions between the discussant, the panel, and the audience.
Simulator sickness occurs in a large number of Army, Navy, and Marine Corps simulators, and is most prevalent in moving-base, rotary-wing devices which employ cathode ray tube (CRT) video displays as opposed to fixed-wing, dome-display trainers with no motion base. Based on data from a factor analysis of over 1000 Navy and Marine Corps pilot simulation exposures, a new scoring procedure was applied to two helicopter simulators with similar rates of simulator sickness incidence. Based on the factor analytic scoring key, the two simulators showed slightly different sickness profiles. Preliminary work was begun to record the visual scene by video frame-by-frame decomposition and automated scoring algorithms were developed. The findings are discussed from the standpoints of (1) recommendations for future design and use of simulators, and (2) the metric advantages and other merits of the "field experiment" methodology to address human factors problems with simulator sickness.
A backward transfer technique was used to identify component skills related to performance on a PC based flight simulator. Forty subjects were first trained on the simulator for eight hours and were then tested on several component skills tasks (backward transfer). A second group of 40 subjects was first given the component skills tasks and then was trained on the simulator task (forward transfer). Both groups were then divided into high- and low-ability groups based on whole-task performance and were compared on the component tasks. Backward transfer subjects showing proficiency on the simulator also showed proficiency on tasks related to pitch, altitude, and heading control, and tasks related to spatial orientation skills. Analysis of the forward transfer group suggested that many of the ability differences observed after whole-task training may have existed prior to whole-task training. Furthermore, some ability differences obtained for the forward transfer group were not obtained in the backward transfer group suggesting that these skills may be acquired through whole-task training. The implications of these results are discussed in the context of part-task training, adaptive training, and gender differences.
The skills approach to aircrew coordination focuses on the training of specific coordination behavior. While this approach has shown promise, careful consideration must be given to the manner in which these skills are trained. A growing body of literature suggests that training workload impacts the acquisition of skills. The present study examined the effectiveness of two training paradigms on the development of aircrew coordination. One regimen of training was characterized by consistent levels of task difficulty over a series of training sessions. The other regimen involved incrementally increasing task difficulty over training sessions. Twenty IFR rated pilots comprised 10 two-person crews. One-half of the crews (control group) completed nine training flights that were of consistent task difficulty. The remaining half completed nine training flights that increased incrementally in task difficulty: three low, three moderate, and three high task difficulty flights. Following the nine training flights, all crew completed five novel aerial reconnaissance flights during which they were instructed to map buildings within a specified area. Measures of subjective workload, flight performance, and secondary task performance were taken for all flights. An additional measure of building identification was also taken for the five reconnaissance-type flights. Measures of subjective workload validated the increasing workload associated with the experimental training regimen. Analysis of the workload data taken during novel task flights failed to reveal differences between training groups. Moreover, there were no significant differences in flight performance between groups. Measures of building identification, a task heavily dependent on crew coordination, revealed significant differences between groups. Experimental crews identified significantly more buildings than their control counterparts for two of the five flights. Implications for aircrew coordination training are discussed.
Research on civil aircraft inspection and maintenance has shown the potential for employing human factor interventions in improving performance. A series of training experiments was developed to understand the effects of different training interventions in the visual inspection domain. This paper reports on preliminary results obtained in applying a combined active and progressive part training scheme in improving the decision making performance for a visual inspection task. The task was a computer simulated airframe visual inspection task.
Established Crew Resource Management (CRM) research has concentrated on the analysis of actual crew behaviors and crew training. As CRM research and training matures, assessment issues take on increasing importance. Current research emphasis has broadened and shifted focus from crew training methodologies in recognition of the importance of the assessment process. The identification of the strategies that experienced instructors employ in efficient CRM assessment is essential to a better understanding of the assessment process. These learned strategies help experienced instructors focus on key crew behaviors in complex Line-Oriented Flight Training (LOFT). This present research was designed to identify these strategies employed by experienced instructors as part of a larger effort to develop a prototype CRM assessment expert system. Protocol analysis techniques were used to develop a set of assessment rules for the prototype expert system. The experienced instructor protocols were translated into a set of pseudo-code rules. The lower-level rules, derived from the experienced instructor protocols, were analyzed and abstracted into a set of general strategies. These are the strategies that the experienced instructor has learned to use to help focus on the key crew behaviors in the complex Line-Oriented Flight Training (LOFT) scenario environment where the instructor cannot observe all behaviors. These strategies have a number of implications for training new instructor/assessors, the standardization and calibration of LOFT assessors, and for achieving more systematic CRM assessments.
Two experiments on the learning and transfer of the hypothesis testing strategy of testing easy-to-test hypotheses first were conducted. The first experiment found that this strategy could be discovered and used in a very simple fault diagnosis task but not in a slightly more complicated task. Subjects who learned the strategy in the simple task were able to transfer it to the more complicated task. The second experiment showed far transfer: The learning of this principle of hypothesis testing transferred to a task sharing no surface features with the training task. It is concluded that it is worthwhile to train people on the use of fault diagnosis strategies.
If subjects are given a series of problems that all have the same type of solution, they often have great difficulty with a subsequent problem that would ordinarily be solved very easily. This phenomenon is referred to as psychic blindness or Einstellung. This study explored whether knowledge of the Einstellung phenomenon facilitates problem solving. Eight subjects in each of three experimental conditions were given set-inducing problems followed by an otherwise easy "critical" problem that could not be solved by the strategy used on the set-inducing problems. Subjects in one condition worked on the easy problem without interruption. Subjects in a second condition were interrupted by an unexpected event. Subjects in a third condition were interrupted by a message explaining how the ability to solve an otherwise easy problem could be inhibited after solving a series of problems with more difficult solutions. Subjects given this hint were three times more likely to solve the easy problem than were subjects in the other experimental conditions. Four subjects in the No-Set Control condition all solved the easy problem by the third trial. The implications for training electronics technicians, computer programmers and other problem solvers are discussed.
Part of the success of computerized intelligent tutoring systems will be associated with their ability to assess and diagnose students' knowledge in order to direct pedagogical interventions. What is needed is a methodology for identifying general relationships between on-line action patterns and patterns of knowledge derived off-line. Such a methodology would allow an assessment and diagnosis of knowledge, based only on student actions. The focus of this initial research is the development of a means of identifying meaningful action patterns in student-tutor interactions. Actions executed by subjects on a set of verbal troubleshooting tests (Nichols et al., 1989) were summarized using the Pathfinder network scaling procedure (Schvaneveldt, 1990). The results obtained from this work indicate that meaningful patterns of actions can be identified using the Pathfinder procedure. The network patterns are meaningful in the sense that they can differentiate high and low performers as defined by a previous scoring method. In addition, the networks reveal differences between high and low performers suggestive of targets for intervention.
A study is directed at a comparison of Virtual Reality as an educational tool in physics instruction with standard, teacher-organized, or computer-aided learning. Findings generally indicated that virtual reality-based learning is superior to lecture-based control conditions. The dependent variable was a residualized knowledge of physics measure obtained from subjects four weeks following termination of training. As a training method, virtual reality was superior to the control condition at the four-week retention period. Such a finding supports cognitive theorists who argue that the lack of opportunities for hands-on, manipulation of objects in the physical world is one of the reasons children are often poor at intuitive physics. Virtual reality provides them the opportunity to develop manipulational skills they did not previously possess.
Visual search is fundamental to many tasks that human operators perform in conjunction with engineered systems. Some obvious examples include industrial inspection, acquisition of information from aviation and weapon systems displays, and interaction with general purpose computer displays. Ease of visual search is a fundamental consideration in the design of computer-based menus and ironic displays (e.g., Andre and Wickens, 1991). Visual search efficiency affects performance variables such as the ease with which an interface or a piece of software can be learned, the time required to complete tasks, and user satisfaction. In weapon systems, visual search affects the probability of target detection, the time until detection, and the range at which detection occurs. These parameters can be critical for combat survival and mission effectiveness.
This paper provides a brief review of the Guided Search model of human visual search behavior. In the model, parallel processes analyze a limited number of basic visual features across large portions of the visual field. The output of these processes can be used to guide attention in the deployment of the limited-capacity processes that are capable of identifying more complex visual targets.
Previous research has demonstrated that search times are reduced when flicker is used to highlight color coded symbols, but that flicker is not distracting when subjects must search for non-highlighted symbols. This prompted an examination of flicker and other stimulus dimensions in a conjunctive search paradigm. In all experiments, at least 15 subjects completed a minimum of 330 trials in which they indicated the presence or absence of target stimuli on a CRT display that contained either 8, 16 or 32 items. In Experiment 1, subjects searched for blue-steady or red-flickering (5.6 Hz) circular targets among blue-flickering and red-steady distractors. Blue-steady targets produced a more efficient search rate (11.6 msec/item) than red-flickering targets (19.3 msec/item). In Experiment 2, a conjunction of flicker and size (large and small filled circles) yielded the opposite results; the search performance for large-flickering targets was unequivocally parallel. In Experiment 3, conjunctions of form and flicker yielded highly serial search performance. The findings are consistent with the response properties of parvo and magnocellular channels of the early visual system, and suggest that search is most efficient when one of these channels can be filtered completely.
The frequency distribution of eye fixations and fixation durations during a search and target acquisition task was examined to determine if the allocation of visual attention was related to target, scene, and/or observer characteristics. Ninety computer-generated scenes simulating infrared imagery and containing different levels of clutter and zero, one, two, or three targets were produced. Targets were embedded in these scenes counterbalancing for range and position. Global and local clutter were measured using both statistical variance and probability of edge metrics. Thirty-three aviators, tankers, and infantry soldiers were shown still video images of the 90 scenes and were instructed to search for targets. Results of multiple regression analyses of global clutter, local clutter, range, number of targets, target dimensions, target complexity, and group membership on eye fixations and fixation durations are given and discussed in terms of search strategies.
Work in progress at Georgia Tech to develop a model of human pattern perception, visual search, and detection is reviewed. The model's algorithms are based on research on low-level visual processes. Recent advances in that field have led to the development of computational models of the image processing performed by the visual system from the cornea to the striate cortex. The model also incorporates recent advances from research on visual search. The organization of the model is described, and the results of some preliminary tests are presented.
The bargraph has been described in several ways: as a separable display, as an integral display, and as a configural display with emergent features. The versatility of the bargraph may be in part due to the support it provides for different individual processing strategies. This research identifies two general types of strategies -- holistic and analytic -- which are developed by individuals to solve a classification problem on the bargraph. Multidimensional scaling (MDS), response times, and verbal reports are used to analyze individual strategies. Individuals who developed holistic strategies produced significantly faster reaction times, and reported simple, efficient strategies, with the emergent feature of bargraph shape as an important dimension. The results indicate that the bargraph provides perceptual features which can support several general types of processing strategy.
The present study examined pilot scheduling behavior in the context of simulated instrument flight. Over the course of the flight, pilots flew along specified routes while concurrently performing three different flight-related secondary tasks. Seven pilots flew the simulation with no preview of future workload conditions, while another seven received preview information in the form of both written instruction and practice. The results show evidence for both macro and micro scheduling strategies. Specifically, those pilots with preview of future workload demands adopted an efficient macro strategy of scheduling more of the difficult secondary tasks during the low workload phase of flight. Subjects in both groups engaged in micro scheduling strategies as a function of flight path workload and secondary task workload.
Stimulus-response compatibility effects have been shown to persist even after extended practice. In the present study, two experiments were conducted to see if the effects persist when knowledge of results that allows subjects to set performance goals is provided. In the first experiment, summary feedback about mean accuracy and mean reaction time was provided after each block of 40 trials of practice in a two-choice spatial compatibility task. Subjects practiced the task for 2,400 trials, yet the compatibility effect was not eliminated. Compared to previous experiments, reaction times were faster overall, but the degree of change was the same for both compatible and incompatible assignments. In the second experiment, a response deadline was imposed on each trial. If the subject did not respond within the time limit, which was reduced as the experiment progressed, auditory feedback was presented. Summary feedback was also presented as in Experiment 1. Again, 2,400 trials of practice reduced but did not eliminate the compatibility effect. The greater reduction in the difference in reaction times for compatible and incompatible assignments, relative to other experiments, could be attributed to speed-accuracy tradeoff. The results indicate that the persistence of stimulus-response compatibility effects with extended practice is not due to poorer motivation to perform with the incompatible assignment. The results suggest that training will be insufficient to overcome difficulties in performance resulting from spatially incompatible assignments.
This study examined the effects of display element arrangement on observers' performance in both Yes/No and Four-Alternative-Forced-Choice (4AFC) visual signal detection tasks. Observers were given four independent informational sources whose values were drawn from either a signal or noise distribution, depending on the task and type of trial. The information was displayed graphically in one of six formats constructed from a combination of two factors: 1) whether the display elements were arranged to produce an emergent feature, and 2) whether or not the magnitude of the emergent feature was monotonically related to the optimal decision statistic (for the Yes/No task). Arranging the line graph displays to produce an emergent feature improved Yes/No performance and impaired 4AFC performance. Due to the highly efficient performance produced by the angular element code, it was not possible to determine whether visual signal detection was affected by the relationship between the emergent feature and the optimal decision statistic.
The use of various dimensions of color to encode continuous data has become commonplace with the advent of sophisticated computing hardware and software. Applications users can choose from a variety of color pallets as well as create their own for viewing digitized data sets. A primary HCI question emerging from this expanded availability of color for data display is how best to map color dimensions to data dimensions for various applications. The current study examined a subset of the perceptual/cognitive processes underlying pattern recognition tasks, whose efficacy could be affected by the nature the color scale used to visualize the data being viewed. Three types of observers' judgments were examined: absolute discrimination of a value; relative judgment of the difference between two values; and a rank order judgment of 4 values. These values were expressed in the color of a specific region in images displayed using eight different color and gray scales. Preference ratings were collected for the color scales. Grey scales were best for rank ordering tasks, while a blue-green-yellow scale proved superior for an absolute independent task. Scale preferences did not necessarily agree with performance. Implications of the findings and future research are discussed.
Head-up displays (HUDs) represent the leading candidate display technology for inclusion in an enhanced or synthetic vision system (EVS or SVS) for commercial transport aircraft. One common EVS concept assumes the raster display of raw or processed sensor (radar or IR) data. However, experience with the use of raster rather than stroke display modes has been largely limited to the presentation of images captured by IR sensitive and image-intensified cameras during night flying conditions when the luminance of the forward scene over which the image will be superimposed is much lower than in daytime. The objective of this work is to generate a specification for minimum HUD raster image modulation assuming real-world luminance values typically found in low-visibility, daylight flight. Six Honeywell pilots rated the image quality and utility of flight video as presented through a military-style HUD in a transport cockpit mockup. Flight video came from daylight FLIR and daylight CCD cameras. The luminance of the forward scene against which the HUD image was superimposed was varied among nine levels ranging from 5 fL to 10,000 fL. The results indicate that HUD raster luminance must be approximately 50% external scene luminance to promote good pilot awareness of general terrain. To maintain good utility and visibility of standard, high-contrast runway markings, runway center line, and runway edges, HUD raster luminance must be approximately 15% of the forward scene luminance.
Previous research has shown that the presence of head-up display (HUD) symbology containing altitude information improves altitude performance at the cost of terrain path performance, implying that these information sources may not be available for concurrent cognitive processing. In two flight simulation experiments, the influence of attentive field size on this concurrent processing limitation was evaluated. In Experiment 1, a superimposed digital altitude (i.e., HUD) indicator was presented at three distances from a flight-relevant ground track. A control condition eliminated the digital altitude indicator. Altitude symbology improved performance on the altitude maintenance task, but impaired performance on the ground track task only when directly superimposed. Experiment 2 tested a visual masking explanation of the performance trade-off. Irrelevant HUD information yielded identical results to the HUD absent condition, ruling out effects of visual masking. An explanation in which visual/spatial attention cannot be directed to both HUD information and terrain information simultaneously is proposed. The absence of a performance tradeoff when the HUD and the terrain information are not directly superimposed is attributed to a breaking of attentional tunneling on the HUD, possibly due to eye movements.
Previous research (McCann, Foyle, & Johnston, 1993) has shown that in a simulated approach to a runway, performance of a choice reaction time task is faster when all relevant information is available on the HUD or in the world, compared to when information has to be acquired from both domains. The present experiment tested two attentional models of these results: attention switching and attention sharing. Removing differential motion cues from the display, so that both the HUD and the world were motionless, attenuated the domain effect. The attenuated difference reflected both slower responses on within-domain trials and faster responses on between-domain trials. We conclude that performance with Head-Up Displays is affected by both attention switching and the degree to which attention is shared between domains.
We examined whether the inclusion of a third head-slaved "roll" degree of freedom (dof) -- in addition to pitch and yaw dofs -- to control the orientation of a remotely-viewed or computer-synthesized scene can enhance spatial situation awareness. Six subjects were required to match the position and orientation of stationary target markers on a remote taskboard by manually placing response markers on an identical local taskboard. Subjects could only view the remote taskboard through images transmitted to a head mounted display (HMD) from a motorized pitch-yaw-roll camera platform; they could see neither the local taskboard nor their own limbs. Results show that, while systematic overshoot errors in azimuth judgment occurred regardless of the roll condition, the addition of the roll dof to the platform had no statistically discernible effect on the subjects' ability to match the position (i.e., azimuth and elevation) of the remote targets. Absence of the roll dof, however, did affect the subjects' judgment of target orientation when their heads were at maximum elevation (pitch) and azimuth (yaw) combinations.
The perceived spatial frequency of a visual pattern can vary with changes in contrast. Because size is inversely related to spatial frequency and because perceived size is an important distance cue, this has implications for task performance in a variable contrast environment. These environments are common in everyday situations, such as driving in the fog, and in the use of night vision devices. Understanding the underlying visual mechanisms of this effect would help us design systems that compensate for the effect. This understanding also could further develop models of the human low-level visual processes. However, most testing of perceived size and contrast has been done at relatively high contrast levels. This research is conducted at contrast levels near detection threshold. This range allows a more thorough testing of different models of contrast detection. We tested two versions of a multiple spatial-frequency channels model of contrast perception. One model assumes a single set of channels functioning throughout the dynamic range of contrasts used here, the other assumes two sets of channels based on the parvo and magnocellular systems. Our results support the hypothesis that there is a single set of channels at work in the contrast range tested.
The forced-choice (FC) format is used for stimulus presentation and performance assessment in a variety of tasks, including target detection and acquisition, tasks that involve primarily vigilance, monitoring, and search (VMS) behaviors. Laboratory experiments, comparison investigations, and system performance assessments have been reported in the literature with numerous claims about FC procedures, including that they lead to simplification of data handling and increased cost-effectiveness. The present report takes issue with these and other claims, as well as with the underlying assumption that FC procedures are interchangeable with continuous-search procedures, when applied to VMS tasks.
Our industrialized society places a premium on the ability to resolve visually fine-spatial detail in the environment. But the perceptual demands of new display systems may involve temporal acuity as much as spatial acuity. Inability to "switch" attention and fixation rapidly from one visual display to another may be a major factor in the "human error" component. We hypothesize that individuals differ in their temporal visual acuity and, if so, then scores on tests which tap this capability would be predictive of productivity on jobs and activities with these demand characteristics. A battery of five temporal factors tests implemented on a portable lap-top microcomputer was administered over several trials to 44 subjects. The tests exhibited satisfactory metric properties (stability, reliability), and did not appear to relate to commonly available global mental abilities. An off-the-shelf portable battery of temporal factors which stabilizes rapidly could have several important applications. First, individual differences in temporal acuity could be employed to improve job productivity to the extent that such abilities are at the basis of certain jobs and activities. Second, since time-course changes were evident with the tests, they can be used to study training implications of these temporal factors. Third, the results could be used to identify temporal aspects of visual displays for advancing understanding in engineering design.
Thermal imaging (TI) systems, transform the distribution of relative temperatures in a scene into a visible TV image. TIs differ significantly from regular TV images. Most TI systems allow their operators to select preferred polarity which determines the way in which gray shades represent different temperatures. Polarity may be set to either black hot (BH) or white hot (WH). The present experiments were designed to investigate the effects of polarity on object recognition performance in TI and to compare object recognition performance of experts and novices. In the first experiment, twenty flight candidates were asked to recognize target objects in 60 dynamic TI recordings taken from two different TI systems. The targets included a variety of human placed and natural objects. Each subject viewed half the targets in BH and the other half in WH polarity in a balanced experimental design. For 24 out of the 60 targets one direction of polarity produced better performance than the other. Although the direction of superior polarity (BH or WH better) was not consistent, the preferred representation of the target object was very consistent. For example, vegetation was more readily recognized when presented as dark objects on a brighter background. The results are discussed in terms of importance of surface determinants versus edge determinants in the recognition of TI objects. In the second experiment, the performance of 10 expert TI users was found to be significantly more accurate but not much faster than the performance of 20 novice subjects.
An analysis methodology and associated experiment were developed to assess whether definable and repeatable signatures of eye-gaze characteristics are evident, preceding a decision to zoom-in, zoom-out, or not to zoom at a computer interface. This user intent discrimination procedure can have broad application in disability aids and telerobotic control. Eye-gaze was collected from 10 subjects in a controlled experiment, requiring zoom decisions. The eye-gaze data were clustered, then fed into a multiple discriminant analysis (MDA) for optimal definition of heuristics separating the zoom-in, zoom-out, and no-zoom conditions. Confusion matrix analyses showed that a number of variable combinations classified at a statistically significant level, but practical significance was more difficult to establish. Composite contour plots demonstrated the regions in parameter space consistently assigned by the MDA to unique zoom conditions. Peak classification occurred at about 1200-1600 msec. Improvements in the methodology to achieve practical real-time zoom control are considered.
Since Gibson's pioneering work in the 1950s, there has been increasing interest in describing the dynamic visual cues operators extract from the "out-the-window" scene to utilize in vehicular control. Despite this interest, we are still a long way from an adequate understanding of what optical information is utilized, and how the information is integrated into an active control strategy. There are a number of reasons for this apparent shortfalling. First, it is difficult to isolate a candidate optical cue; geometry dictates that several candidate cues will co-vary in any natural scene (e.g., edge rate and flow rate). The experimental isolation of an optical cue often results in visual scenes which are quite unnatural, creating the possibility that strategies used in the experimental setting will not generalize to operational settings. Also, much of the laboratory work has focused on demonstrating people's sensitivity to optical variables, utilizing passive verbal judgments rather than active control paradigms. Whereas the demonstration of sensitivity to an optical cue is a logically necessary step, such a demonstration is not sufficient to verify its utility in an active control task. Further, there is the need for an adequate description of the task demands, allowing a proper mapping between what the controller is trying to achieve and the information available to accomplish the task; no single cue (or set of cues) will be appropriate for all vehicular control tasks. Finally, given the robust and opportunistic nature of the human perceptual system, it is possible that the visual cues used for vehicle control will vary from individual to individual, or even within an individual depending on which cues are available and salient in the control environment. The participants in this panel are well versed in the challenges of studying visually based vehicular control. Their presentations will reflect the lessons learned in this field, as well as insights regarding how current and future research can better realize the promise of this domain.
Earlier research has shown that a task-irrelevant sudden onset of an object will capture or draw an observer's visual attention to that object's location (e.g., Yantis & Jonides, 1984). In the four experiments reported here, we explore the question of whether task-irrelevant properties other than sudden-onset may capture attention. Our results suggest that a uniquely colored or luminous object, as well as an irrelevant boundary, may indeed capture or guide attention, though apparently to a lesser degree than a sudden onset: it appears that the degree of attentional capture is dependent on the relative salience of the varied, irrelevant dimension. Whereas a sudden onset is very salient, a uniquely colored object, for example, is only salient relative to the other objects within view, both to the degree that it is different in hue from its neighbors and the number of neighbors from which it differs. The relationship of these findings to work in the fields of visual momentum and visual scanning is noted.
The present study explored whether instructions given to subjects at the outset of a vigilance experiment impact their ratings of frustration and workload. The present task consisted of monitoring a video display terminal (VDT) of uniform color for 30 min. Subjects were asked to respond to occasional 3 ms "flickers" to a different color. Half of the subjects were told to relax by focusing on the display, but to respond to any flickers observed. The remaining subjects were given traditional vigilance instructions emphasizing the importance of detecting as many "critical signals" or flickers as possible. Before and after the vigil, subjects completed the NASA Task Load Index (TLX; Hart and Staveland, 1988) which measured the subjective workload for the vigil. Hits and false alarm data were recorded for each 10 min period within the 30 min vigil. A significant reduction in mean number of hits was observed over the three periods for all subjects. A subsequent analysis showed that perceptual sensitivity also declined significantly over time. Performance, however, was not affected by instruction type. Subjects who received relaxation-emphasis instructions did report significantly lower workload and frustration for the vigil than those receiving detection-emphasis instructions. These results indicate that much of what individuals find unpleasant about participating in vigilance experiments may lie with the expectations outlined in the initial instructions.
Perceived workload was measured via the NASA TLX following a visual vigilance task. Five task durations (10, 20, 30, 40, or 50 min) were combined pictorially with two levels of discrimination difficulty (easy, hard) in a between groups design. Detection probability, computed from the final 10 min of watch in each duration condition, varied inversely with signal salience and declined over time. Overall workload varied directly with salience and increased linearly over time. The temporal growth in perceived workload was independent of signal salience. This result suggests that the rate of gain in workload is based upon general features of the vigilance situation rather than specific psychophysical demands such as signal salience.
Twenty-four male volunteers participated in a study in which the feature extraction and response choice information processing stages of a visual memory task and the environment in which the task was performed were manipulated in order to assess the specificity of energetic demands. Ambient auditory noise was a state variable that was predicted to directly affect the energetic mechanism supporting the feature extraction processing stage and to indirectly affect the energetic mechanism supporting the response choice processing stage. Several classes of dependent measures were taken including performance, cardiovascular, pulmonary, and metabolic in order to isolate effects upon energetic mechanisms. The present study did not support the proposition of multiple energetic mechanisms. Instead, the combination of a single variable-capacity energetic resource with a flexible allocation policy to the processing stages was a better fit to the results.
This article describes two studies that were conducted as part of a systematic effort at the University of Michigan to develop computational and comprehensive models of complex human performance and to improve the scientific basis for these models. The focus of the first study was the integration of models of divided attention and models of selective attention in performing complex tasks. Two experiments were conducted, which required the subjects to perform a simple information acquisition task in the first experiment and a complex information integration task in the second experiment. The two types of tasks were performed either alone or concurrently with a tracking task, and involved either spatial or verbal material. The location of the relevant spatial and verbal material was displayed with 4 levels of spatial uncertainty, but with approximately the same expected distance for visual scanning. The results demonstrated the strengths and limitations of existing models. The potential value of power functions in quantifying different aspects of task interference was proposed in the paper. A queuing network model, that was proposed recently (Liu, 1993a) as an unifying theory and an integrated computational model of human multi-task performance, was also tested in this study. In the second study, a computational model was derived from models of memory scanning and visual scanning and evaluated through an experiment to examine the integration of the two aspects of human performance modeling. We report here the first of a series of experiments of the study, which required the subjects to search through an organized array of circles to decide whether any of the circles carried any of the memorized items in their working memory. The joint effects of two experimental factors were investigated: the number of items in working memory and the number of circles need to be searched.
Topographic EEG measures were compared in 12 adult male subjects during performance of a signal recognition task, presented at three difficulty levels. EEG data were recorded from 17 standard cortical sites, referenced to linked earlobes. Digitized mean spectral magnitude values were calculated for sequential 2 second epochs for each condition, log transformed and subjected to statistical analysis. A good and a poor performance group was established on the basis of scores registered at the highest difficulty level and confirmed statistically. Within-group comparisons showed different EEG patterns for the two performance groups, both within and across difficulty level. The poor performance group showed a progressive pattern of disengagement (increase in 8-12 Hz activity) which diminished gradually as difficulty escalated and was replaced by a pattern of increasing engagement (decrease in 8-12 Hz activity). Good performers showed the same level of engagement independent of difficulty. Performance data alone failed to differentiate between groups under low and moderate task demands. Detailed evaluation of the underlying mechanisms revealed a tendency for all subjects to develop brief periods of disengagement after each stimulus presentation. This pattern became increasingly generalized in poor performers during the low gain task but was also present at the most difficult test level. These findings provide some insight into the dynamics of Central Nervous Systems regulatory mechanisms which modulate sustained cognitive performance under varying demand conditions. They document a propensity for some individuals to become disengaged over time, thereby requiring greater cognitive resource mobilization as task demand increases. Assessment of this trait may be useful in the prediction of performance capability under high demand conditions.
This study assessed the sorts of biases induced in a pilot's estimates of velocity, elapsed time, and distance traveled in a flight simulation environment. The validity of an "internal equation" relating the concepts of subjective distance, velocity, and time was assessed. Pilots flew a series of flight legs which varied in their distance, time and velocity, during which they were asked to make judgments of time passage (group 1), speed (group 2), and distance traveled (group 3). These judgments were made under both single and dual task conditions. There was little consistent effect of task loading on any of the subjective estimates, but variations in interval duration, speed, and distance effected the subjective estimates of these quantities in different ways. There was no evidence found for support of an internal equation.
This paper reports a programme of laboratory research to assess the potential of a new form of aircraft attitude indicator. The Ambient Attitude Indicator (AAI), designed to exploit the characteristics of the ambient visual system, provides a continuous source of world-stabilized orientation information to the pilot's ambient visual system. In laboratory experiments, the first of which is reported here, the effects of peripheral visual cues on subjects' control of continuous self-stabilization in roll are assessed. Subjects, seated on a roll turntable, were required to null a continuous quasi-random forcing function using a velocity-control joystick. Their objective was to maintain a stable upright orientation. Three visual conditions were studied: no visual cues, world-stabilized peripheral visual cues, and turntable-stabilized peripheral visual cues. Significant effects of visual condition were found for measures of joystick RMS displacement, turntable RMS error, and frequency of changes of direction of the turntable. World-stabilized conditions resulted in the highest joystick RMS displacement, the lowest turntable RMS error, and the highest frequency of changes of direction of the turntable. A linear transfer function, with autoregressive noise process to represent the remnant, was fitted to the data using maximum likelihood estimation. Details of subjects' frequency response are presented and the effects of practice on these measures are also considered.
Cognitive Systems Engineering (CSE) is primarily a blend of technological opportunities, findings from cognitive research, and Cognitive Task Analysis. Using CSE, we were able to produce an efficient and effective redesign of the AWACS Weapons Director (WD) station. The design effort was completed in a relatively short period of time. A Cognitive Task Analysis was performed using two interview techniques: Concept Mapping and the Critical Decision method (CDM). The information obtained using these methods pinpointed specific cognitive areas which any redesign of the WD interface must address. A revised interface was developed and evaluated. During the evaluation, the training of the participants on the revised interface was quite brief (4.5 hours). As a result, the WDs did not achieve the same degree of familiarity or automatization with the revised interface that they have with the current interface. Yet, when WDs were using the revised system their performance improved. This was indicated by an increase in performance for a number of process and outcome measures. Also, a skilled WD provided blind ratings of WD performance. These global ratings were significantly higher for the revised interface. The effectiveness of the revised interface suggests that it is possible to pinpoint cognitive task requirements and to make these the driving factors in a design effort. Moreover, these Cognitive Systems Engineering activities do not consume a great deal of time or effort. The use of CSE may be a feasible aspect of the design process, enabling system developers to achieve a much stronger effectiveness at relatively low cost.
The efficacy of two methods of training dual-task skills was examined in this experiment. Thirty older subjects (Mean age = 67.8 years) were trained using either variable priority or fixed priority training. Subjects performed two tasks, a gauge monitoring task and a letter arithmetic task, both separately and together. Subjects in the variable priority group were trained to vary their processing priorities between the letter arithmetic and monitoring tasks. The fixed priority subjects were trained to devote equal priority to the two tasks. Subjects then transferred to a complex scheduling task which was paired with a paired-associates task. Variable priority subjects exhibited an initial performance cost relative to fixed priority subjects. By the end of training, however, variable priority subjects exhibited superior performance as compared to fixed priority subjects. The performance of variable priority subjects was also superior on transfer tasks with which the subjects had no prior experience, suggesting that variable priority training may involve a generalizable time-sharing skill.
The present experiment investigated ability-performance relationships for two memory skills, each of which required associative learning. Evidence suggests that, after practice, young and old adults have equivalent associative learning abilities (Fisk and Rogers, 1991; Kausler, 1982). We provided 41 young and 52 old adults with extensive practice on consistently and varied versions of a memory search task and a noun pair look-up task (Ackerman and Woltz, 1993). Only consistent practice allows associative learning because the stimulus items are consistently paired; in varied practice, item pairings change across practice and associative learning is not possible. We also assessed a wide range of abilities for each subject and were thus able to investigate ability-performance relationships across practice conditions and across age groups. These relationships provide an indication of the underlying abilities related to task performance (Ackerman, 1988). The mean data suggested that both young and old adults demonstrated successful associative learning in the two CM tasks. The individual differences data suggest, however, that different abilities may be driving performance across the two age groups. These data have important implications for predicting whether or not older adults will successfully acquire a new skill. If the target skill requires associative learning, older adults, may perform as efficiently as young adults if they are provided with sufficient, consistent practice. The ability-performance data suggest that predictions about which individuals will be most successful at skills requiring associative learning, may be dependent on the age of the target population.
Visual search is an important component of many real world tasks such as industrial inspection and driving. Several studies have shown that age has an impact on visual search performance. In general older people demonstrate poorer performance on such tasks as compared to younger people. However, there is controversy regarding the source of the age-performance effect. The objective of this study was to examine the relationship between component abilities and visual search performance, in order to identify the locus of age-related performance differences. Six abilities including reaction time, working memory, selective attention and spatial localization were identified as important components of visual search performance. Thirty-two subjects ranging in age from 18-84 years, categorized in three different age groups (young, middle, and older) participated in the study. Their component abilities were measured and they performed a visual search task. The visual search task varied in complexity in terms of type of targets detected. Significant relationships were found between some of the component skills and search performance. Significant age effects were also observed. A model was developed using hierarchical multiple linear regression to explain the variance in search performance. Results indicated that reaction time, selective attention, and age were important predictors of search performance with reaction time and selective attention accounting for most of the variance.
This experiment sought to determine if posture- and exercise-induced neuro-stimulation influences age differences in reaction (RT) and movement (MT) time, and whether obtained effects varied with physical fitness level. Thirty-six healthy male participants (18 young (19-29 yrs) and 18 old (60-69 yrs), with each group divided into the fit or unfit) performed both simple and two-choice visual reaction time tasks under six arousal/activation conditions: three postural changes (supine, sitting, standing) and three different relative workloads on a cycle ergometer (free pedaling, 20% HRR{sub:max}, 40% HRR{sub:max}). Consistently, RTs were slower for the older vs. young adults but the elderly performed fastest when Standing than when Sitting or Lying, whereas posture effects were negligible in the young. During exercise SRTs in the young and Old Fit were not greatly influenced by fitness level or arousal/activation condition, but the Old Unfit benefited from moderate (20% HRR{sub:max}) exercise-induced neuromuscular activation thereby accounting for a portion of age-related cognitive slowing by providing evidence that the elderly function at a less activated (aroused) level than young adults and may benefit from circumstances which elevate these levels. An opposite pattern occurred in MTs for the Old Unfit -- for both posture and exercise: increases in arousal/activation caused increases in MT but in a fashion not supporting an RT-MT tradeoff in response strategy. Posture and exercise does affect speed of response, and may reduce age differences especially for those who possess already slowed response latencies.
Gerontechnology includes the research and development of techniques and technological products, based on the knowledge of aging processes, for the benefit of a preferred living and working environment and adapted medical care for the elderly. Physical and mental fitness are prerequisites to the satisfactory performance of daily tasks. Functionality decreases when perceptive motor abilities or skills diminish, when task demands are too high and/or when the product characteristics, the user-interface or the environmental conditions are in conflict with human skills. The introduction emphasizes the difference between approaching the elderly as patients or consumers. A concept of social interaction and active participation is described. It is discussed which support can be offered by technology and what the state of the art in gerontechnology is.
Adults over the age of 65 consume more prescription medication than individuals in any other age group. Researchers have also reported that rates of noncompliance for prescribed regimens are alarmingly high for the elderly. Thus, it is extremely important to understand the factors that may contribute to the high noncompliance rates in older adults. Although there is a vast body of literature on medication nonadherence available, there has been little systematic attention paid to the relationship of cognitive ability factors to nonadherence. This topic is particularly germane to the adherence behavior of older adults because it is likely that this group of individuals will experience age-related declines in memory and comprehension processes that may affect their ability to adhere. Increasing evidence in the literature also suggests that age-related decrements in comprehension and memory may substantiate much of the nonadherent behaviors found in elderly individuals. In addition to the lack of focus on the influence of cognitive factors on adherence behaviors, none of the methods that have been used to measure adherence in past research (i.e., verbal reports and pill counts) have permitted precise monitoring of the specific medications taken and the date and time when they were taken. Therefore, the objective of this symposium is to present innovative research techniques from different laboratories that have been designed to address the methodological problems with past studies on medication adherence in the elderly. Special emphasis will also be given to unique cognitive strategies that have been developed to aid medication compliance in older adults.
We have collected data on medication adherence in several studies from samples of younger as well as elderly adults. Samples have included hypertensive adults, adults taking medications for a range of illnesses, and adults with osteoarthritis. The time range for collecting adherence data has varied from two weeks to two months, and the level of analysis has varied from examination of individuals medications across 60 days to monthly estimates of overall adherence rates. Finally, our research group has extensive experience with two microelectronic techniques for measuring adherence: the Videx time wand system which relies on bar code scanners to measure adherence, and the Medication Event Monitoring System (MEMS) which involves pressure sensitive lids that record the date and time a lid is removed from a prescription medication. Issues involved in measurement of adherence are presented and various techniques for presenting and analyzing data are discussed.
Elderly patients (N=617) with scheduled appointments in a public health clinic tuberculosis clinic either received or did not receive an automated telephone reminder the evening before their appointment. Patients in this population were primarily non-English speaking immigrants who received reminders in their own language. Automated reminders decreased nonadherence 21% (from 29% to 23%), and this impact did not differ across ethnic groups.
We examined if medication instructions were better remembered when organized in terms of older adults' preexisting schemes for taking medication. A preliminary study suggested that older adults share a general scheme with medication information grouped into 3 categories: (a) General Information (e.g., medication purpose), (b) How to take (dose), and (c) Possible Outcomes (side-effects). In the present study, we investigated age differences in this scheme and in instruction recall. We also examined if individual differences in organization related to cognitive abilities, health care beliefs, and medication taking experience. For the most part, the results provided further evidence that older adults share a scheme for taking medication and revealed few age differences in this organization. Verbal ability was more important than health attitudes for predicting individual differences in instruction organization. Most important, older and younger subjects preferred and better remembered instructions that were organized in terms of their medication taking scheme.
A visco-elastic model of the mechanical properties of muscle was used to describe age-differences in the buildup of force in isometric elbow flexion. Given information from the literature on age-related physiological changes, such as decreasing connective-tissue elasticity, one would expect changes in the mechanical properties of skeletal muscle and their related model parameters. Force vs. time curves were obtained for 7 young (aged 21-27) and 7 old (aged 69-83) female subject. There were significant age group differences in steady-state force level and the best fitting model parameters. In particular, the viscous damping element of the model plays a large role in describing the increased time to reach steady-state force levels in the older subject group. Implications of this research include incorporating parameter differences into more complex models, such as crash impact models.
The time drivers require to react in braking situations underlies many practices in highway design and operations. There is concern whether the perception-reaction time (PRT) values used in current practice adequately meet the requirements of many older drivers. This study compared on-the-road brake PRTs for unsuspecting drivers in three age groups: 20-40, 65-69, and 70-plus years old. The method included features to enhance the ecological validity of the observed reactions: subjects drove their own vehicles in their normal manner; driving was on actual roadways; extended preliminary driving put the driver at ease and without expectation of unusual events at the time of the braking incident; the incident occurred at a location lacking features that might enhance alertness (e.g., curves, crests, driveways). Subjects drove an extended route, under the guise that they were making periodic judgments about "road quality." At one point, a large crash barrel was remotely released from behind brush on a berm and rolled toward the driver's path. Although most of the fastest observed PRTs were from the young group, there were no differences in central tendency (mean = 1.5 s) or upper percentile values (85th percentile = 1.9 s) among the age groups. Furthermore, the current highway design value of 2.5 seconds for brake PRT appears adequate to cover the full range of drivers.
The U.S. aviation industry is in serious financial trouble. Since 1990 two major carriers have ceased operations (Eastern & Pan Am) while many others have sought court protection from creditors. Boeing and Douglas Aircraft had record losses in 1992, and 1993 looks worse. Engine builders like General Electric and United Technology are having similar problems. Can research in human performance remedy the ailing industry? The answer is, Yes: improving human performance is one of many fixes that will help the industry. This paper offers aviation-specific examples of cost control accomplished through increases in human performance.
By the year 2024, 25% of drivers in the U.S. will be over the age of 65. Older drivers have more crashes and fatalities per mile driven than any other adult age group. Although driving is a highly visual task and vision impairment is more prevalent in the elderly, previous research has failed to identify visual factors which are strongly associated with increased crashes in the elderly. Using a comprehensive approach to assess several aspects of visual processing in a large sample of older drivers, this study has identified
Education is using technology in unprecedented ways. Rather than being used as stand-alone automated tutors, technology is being used to enable children to extend the learning environment beyond the classroom into the home, library, and the community at large. This symposium describes these educational trends and the human factors issues they raise.
This paper describes an experiment that examined the effect of car phone design on simulated driving and dialing performance. The results were used to help develop an easy to use car phone interface and to provide task times as input for a human performance model. Twelve drivers (six under 35 years, six over 60 years) participated in a laboratory experiment in which they operated a simple driving simulator and used a car phone. The phone was either manually dialed or voice-operated and the associated display was either mounted on the instrument panel (IP) or a simulated head-up display (HUD). The phone numbers dialed were either local (7 digits) or long distance (11 digits), and could be familiar (memorized before the experiment) or unfamiliar to the subject. Four tasks were performed after dialing a phone number, two of the tasks were fairly ordinary (listening, talking) and two required some mental processing (loose ends, listing). In terms of driving performance, dialing while driving resulted in greater lane deviation (16.8 cm) than performing a task while driving (13.2 cm). In addition, the voice-operated phone resulted in better driving performance (14.5 cm) than the manual phone (15.5 cm) using either the IP display or HUD. In terms of dialing performance, older drivers dialed 11-digit numbers faster using the voice phone (12.8 seconds) than the manual phone (19.6 seconds). Dialing performance was also affected by the familiarity of numbers. Dialing unfamiliar numbers using the voice phone was faster (9.7 seconds) than using the manual phone (13.0 seconds) and 7-digit unfamiliar numbers were dialed faster (8.2 seconds) than 11-digit unfamiliar numbers (14.5 seconds). Thus, the voice-operated design appears to be an effective way of improving the safety and performance of car phone use, but the location of the display is not important.
The numeric keys on the keypads of devices such as telephones and point of sale terminals have the letters of the alphabet mapped onto the digits 2-9, except for Q and Z, which are missing. The international standards bodies CCITT and ISO/IEC are standardizing the placement of letters on 10-digit keypads, and must decide where to place Q and Z. Two alternatives have been considered: placing Q and Z in alphabetic order on the 7 and 9 keys, respectively, or placing Q and Z on the 1 key. A study was conducted to determine if one of these alternatives is to be preferred on the basis of human keying performance and/or preferences. Performance differences were too small to conclude that one alternative should be preferred, however, the majority of users clearly indicated a preference for alphabetic order. A nation-wide preference and usage survey among active calling card and bank machine users revealed an overwhelming preference for the alphabetic order. Finally, a summary is given of the current status of the 10-key pad standardization process in CCITT and ISO/IEC.
Rapid Order is an Ameritech Services audiotext system which lets consumers learn about and order 12 telephone services. The current menu structure is based on the development history of the services rather than on an intuitive organization for the consumer. Four studies were performed to develop a new menu structure which would be easier for consumers to use. The first study used cluster analysis to develop a menu structure. The second experiment tested the goodness of fit of the new menu compared to the old menu. Results showed that consumers were more accurate in determining where on the menu a service would be found with the new menu. In the third experiment, we changed the name of one of the menu categories and got an even greater improvement in performance. Finally, we tested the same three menu structures using a phone prototype rather than a paper task and the results were similar to those with the paper task. Consumers were 18% more accurate in locating a service on the best new menu than on the existing menu. These results strongly suggest that changing the Rapid Order menu will result in improved customer performance with the menu. They also illustrate how cluster analysis alone will not provide all the answers in menu design. The labeling of the clusters can have a significant effect on the goodness of fit of the menu.
The introduction of videocommunications via the Integrated Services Digital Network (ISDN) offers the potential of significant changes in the working conditions of a variety of professions including conference interpreters. A study was conducted aiming at identifying the special videocommunications requirements of professional conference interpreters and establishing whether ISDN videotelephony offers a sufficient audio bandwidth and image resolution to support the interpreters in their work. To this aim, a requirements analysis was conducted and four videotelephony systems were set up each implementing a different ISDN combination of audio bandwidth and picture quality. Five professional interpreters took part in the study. Each performed a simultaneous interpretation task of 30 minutes under each of the four conditions (set-ups). Stimulus material to be interpreted consisted of videotapes of conference presentations and TV interviews. Data on the dependent variables (task difficulty, benefits of the picture, sound quality, Social Presence, performance, and acceptance) were collected by means of questionnaires. The most important conclusion is that remote interpreting using ISDN videotelephony is possible only at a cost in terms of both increased fatigue and less satisfaction of the interpreters with their work. Remote interpreting requires at least the image quality that can be provided by a 384 kbit/s transmission rate (until better coding algorithms are available for lower transmission rates) and an audio bandwidth of at least 7 kHz. It is further recommended to conduct a field study with standardized interpreting equipment using more realistic input material.
Call routing involves directing incoming telephone calls from a central number to an appropriate person or department. In the course of an ongoing work project, a quick study was performed to scope the vocabulary requirements for a speech recognition automated call routing application for a large department store. Forty-one participants were given 35 sample shopping tasks and were asked which department they would ask for when calling the store. The range of responses for a given task was large. With a 29 item recognition vocabulary consisting of most frequent responses and root phrases (e.g., "sport" for "sporting goods"), 57% of user responses would be covered. Users were also asked to rate the confidence of their department choice. The greater the variety of responses to a task across all participants, the less confident participants were of their responses.
Time compression increases the rate of speech without altering its pitch. The present study investigated time compression as a means of improving the efficiency of audiotext applications for a variety of user populations. Subjects from three age groups (20-30, 40-50, and 60-70 years old) and two native language groups (native and nonnative English speakers) interacted with a prototype of an Interactive Voice Response system. Four prototypes were
This paper deals with factors affecting Communicative Presence in video-communications. Communicative Presence is defined as the capacity of a system to transfer mutual communicative signals of interlocutors. The experiment the paper reports on examined the effects of various features of videoconferencing systems in terms of several aspects, such as the conferees' feeling of being individually addressed by non-verbal signals (e.g. eye-contact), the flow of conversation, user satisfaction, and the willingness to use videoconferencing systems. One system feature that was systematically varied was the degree of vertical and horizontal eye-contact angles. In order to reduce the horizontal eye-contact angles for the two test subjects acting as conferees at one site of the experimental set-up, two different images (one per conferee) which were in accordance with the perspectives the conferees would have in a similar face-to-face meeting were displayed ("view-per-person" principle). Another feature that was examined was the spatial resolution of the displayed images. The results showed that large eye-contact angles as well as a low resolution decrease the feeling of Communicative Presence. Within the framework of the experiment, the hypothesis concerning the benefits of a view-per-person representation could not be verified.
The writing efficiency of a type of computer-based groupware called a group editor was evaluated in six environments by factorially combining three communication media (audio only, audio plus video, and face-to-face) each with and without the group editor. Twelve subjects familiar with journalistic writing were pretested and matched into dyads. Subjects were trained with the group editor and wrote news articles based on a standard set of questions about actors and objects shown in short video clips. The quality of the writing task products was consistently high and showed no differences among conditions. Results of a balanced, within-subject analysis of variance design indicated that face-to-face conditions took significantly less time to complete than the other two communication conditions. When the group editor was used, however, all communication media were equal in terms of efficiency to face-to-face communications. Significantly less variance occurred in the audio plus video condition with the group editor than in the audio or the audio plus video conditions which did not have the group editor. Users preferred co-authoring with the group editor and considered writing trials with the group editor to be more productive. The results were discussed in terms of the benefits of using the group editor to increase the overall communication structure, to reduce variability of writing time among dyads, and to increase efficiency when face-to-face communication is not practical. Methodological procedures for matching subjects, using within-subject designs, and structuring a writing task were discussed as means of reducing team variability in writing efficiency when investigating computer-supported, cooperative work configurations.
Computer-based products with multimedia user interfaces will need to communicate large amounts of data and functions, as well as concepts and emotional values, to increasingly diverse users. Using metaphors to embody complex structures and processes is one technique available to user interface developers. The article discusses kinds of metaphors and metaphor design scenarios.
Aviation Safety Inspectors make up the inspection team of the Flight Standards Service, which is responsible for enforcing the laws governing civil aviation. We are developing pen-computer-based job performance aids to assist the inspectors in their job duties. These aids provide inspectors with tools for data collection and on-line documentation. The tools eliminate redundant data recording, prevent errors, increase capabilities, and support quick retrieval of up-to-date regulatory information.
There have been numerous methodologies, models, and tools created to support successful user-system interface (USI) design. One such tool is USI design guidelines, which is important for both software developers and human factors professionals in developing a good user interface. This paper discusses the creation of interactive USI design guidelines intended for software developers to use when creating applications in the Microsoft Windows graphical software environment. User-system interface design guidelines are an important part of the software design process and complement other human factors activities that support good USI design. Differences between printed and on-line guidelines documents suggest developing on-line guidelines to support the development of Windows-based GE Information Services applications. The content of the GE guidelines is tailored toward company applications, using examples of both good and bad user interface designs to illustrate guideline principles. The guidelines also include a sample application that incorporates the guidelines in its user interface. Components that contribute to the effectiveness of the guidelines, such as quality, time required to use, relevance, and complexity, are explored.
This paper reviews recent developments in the area of software user interface standardization, and identifies issues that must be addressed before an effort to establish an American National Standards Institute (ANSI) software user interface standard can be successfully completed. Arguments supporting and opposing the establishment of software user interface standards are reviewed. Potential resolutions to the critical issues including technical, organizational, logistical, and political issues are identified and discussed.
Many human-computer interface design guidelines have been developed to design good interfaces for various kinds of software. Database systems have been also developed for accessing the guidelines. This paper considers the role of the design guidelines, rather than the role of the database, in improving interface designs. Sixteen software designers, who have no human factors experience, participated in a typical design review task. They were provided with a representation of a bad interface design. Eight designers (the UG participants) were instructed to individually improve the design by using the guidelines. The other designers (the NG participants) were instructed to improve it unaided (without the guidelines). The results indicated that both groups made similar numbers of improvements, but the UG participants produced higher quality improvements. Quality was evaluated using a goodness measure defined in this paper. The NG participants made good improvements but also bad ones that conflicted with the guidelines because only the designers' knowledge, experience and preference were used. On the other hand, the UG participants made fewer bad proposals because they could refer to the guidelines. Guidelines can work as a filter to eliminate inappropriate or false improvements from the designers' original proposals. There is a possibility that the guidelines may hinder the designer from developing new and interesting proposals. Their value is, however, very clear for novice designers who have no human factors experience; they can easily develop high quality proposals.
This paper explores the application of human factors methods to the design and documentation of software installation processes and procedures. The study analyzes the installation of a telecommunications operations support system in a workstation-based, distributed computing environment. The long-range goal of the endeavor is to produce usability objectives and guidelines that project teams can use in the development of software installation related procedures and user interfaces. The areas being studied include software installation planning, documentation, training, procedures, and customer support. Methods used include the development of user profiles, interviews with the system administrators, and field observations. Whereas the actual installation of the software proved to be relatively easy, the planning and pre-installation phases proved to be very complex and time-consuming. We explore some of the reasons for this and identify areas that need to be considered when developing software deployment plans and user interfaces.
One of the main tenets of most company-sponsored quality programs is that the customer is always right. Designers frequently evaluate the goodness of their systems by simply asking users whether or not they like the interface. The fallacy of this approach is that users generally make judgements based on their "preferences" and tend to ignore the more important performance issues. System designers frequently use their own preferences to make decisions, and then make major inferences about how users will perform with their system. Several past studies are reviewed to show that users can perform well and not like a system, or like a system and still not perform well. Two recent studies are reported showing a mismatch between designer's preferences for certain interface decisions, and measured user performance when using the resulting interfaces. It is proposed that better user interfaces are possible if we clearly separate the performance and preference concepts, recognize the limitations of each, and work to optimize one or the other (there is usually not sufficient time to optimize both). The only way to ensure that systems will elicit acceptable levels of performance is to conduct performance-oriented usability tests.
A well-designed user interface is recognized as a benchmark for determining the success of a software product. The proliferation of user interface design guidelines, standards, prototyping tools, and techniques are indicative of the importance placed on quality user interfaces. However, even with the availability of the latest information, tools, and human factors practitioners to software developers, sub-optimal interfaces may result. This is because within a large multidisciplinary software design team, issues such as communication, responsibilities, and cost and schedule constraints may override the usability issues. This paper describes the implementation of concurrent engineering, used to successfully develop user interfaces for a large, complex system. Success is expressed in terms of quality and consistent user interfaces, positive influence of human factors on software development, and customer satisfaction.
Research shows that developers and designers alike use examples in their design process. However, the tools that developers and designers use are difficult to use because they provide few or no examples and require extensive reading and memorization. This study gave developers and designers a background questionnaire and asked them to design an interface for displaying and exploring user interface examples within a user interface design guidance system. Analysis of audio tape, screen layouts, and user and task profile data from the study provided a number of user interface requirements for systems that support the representation and use of user interface examples.
CARD (Collaborative Analysis of Requirements and Design) is a participatory technique for analyzing task flows and for redesigning task flows, in software systems. It provides a macroscopic complement to the more microscopic design activities that are supported by the PICTIVE technique. CARD uses the metaphor of a card game as the vehicle for communication and collaboration among users, developers, and designers. We describe the technique, and provide illustrative session protocols and assessment data. The paper closes with a comparison to other relevant participatory practices, and a discussion of CARD's shortcomings.
Safety considerations require that air carrier maintenance be of the highest quality and essentially error-free. Economic considerations require that maintenance activities be as efficient as possible. Success in meeting both safety and economic goals depends on the performance of those working in the aviation maintenance community. A human factors guide can be of value in supporting and enhancing the performance of aviation maintenance personnel. One end product of the Federal Aviation Administration (FAA) program on "Human Factors in Aviation Maintenance" is a guidebook presenting human factors information oriented specifically toward the air carrier maintenance workforce. The Human Factors Guide presents established principles of job design and work. Coverage is broad and includes a variety of topics considered important in determining maintenance effectiveness and in the control of maintenance error. The preparation of the Human Factors Guide was done in a manner to ensure that needs of users were addressed. Size and format were selected in terms of user preferences. Information obtained through a series of FAA-sponsored human factors meetings plus that gained from a survey of maintenance personnel guided both the selection of topics and the presentation of materials.
A Usability Inspection is a specific formal design review methodology to identify and prioritize potential usability defects in a given design. More formal than heuristic evaluations (Nielsen & Molich, 1990), less complex than cognitive walkthroughs (Poison et al., 1992), the methodology is a task-oriented review process that started as an adaptation of the Kepner-Tregoe Potential Problem Analysis and further evolved to heavily leverage a formal generic inspections methodology in use at Hewlett-Packard. This evolution by adaptation of standard generic methods already understood by engineers in Hewlett-Packard has been a conscious design strategy in the hope that this would ease the adoption process. The key question for a practitioner in the field is -- does the method provide sufficient return on investment to be worth including in a product development process. There is some evidence that design review methods carried out by non-Human Factors engineers do successfully find defects (Nielsen & Molich, 1990, Lewis et al., 1990, Jeffries et al., 1991). Bailey (1992) raised the question as to whether heuristic evaluations may be counterproductive in that they actually may create work rather than save it. To date, the research has largely focussed on the contribution a method may provide to a specific product or interface. This paper argues the case for a wider definition and understanding of potential utility or contribution from a Human Factors method and further claims that usability inspections not only provide a sufficient payback to warrant their inclusion in a product development lifecycle but also provide the seeds of organization change that are needed to make usability engineering a reality.
A high-fidelity prototype of an extended voice mail application was created. We tested it using three distinct usability testing paradigms so that we could compare the quantity and quality of the information obtained using each. The three methods employed were (1) heuristic evaluation, in which usability experts critique the user interface, (2) think-aloud testing, in which naive subjects comment on the system as they use it, and (3) performance testing, in which task completion times and error rates are collected as naive subjects interact with the system. The three testing methodologies were roughly equivalent in their ability to detect a core set of usability problems on a per evaluator basis. However, the heuristic and think-aloud evaluations were generally more sensitive, uncovering a broader array of problems in the user interface. Implications of these findings are discussed in terms of the costs of doing the evaluations and in light of other work on this topic.
The "mole" is a foot-operated input device for a computer which emulates a mouse in functionality. It is used to control the position of the pointer on a PC screen. The "mole" device studied in the present research is the second-generation version of the device which was developed by Pearson (1988). In Pearson's previous studies, (Pearson and Weiser 1988), he demonstrated that individuals unfamiliar with the "mole" could learn to use the device to hit small on-screen targets comparable in size to small font size characters. The present studies were designed to provide data for evaluation of both the second-generation device and its software. The first study tested the use of the "mole" in controlling the pointer to solve the problem of the Towers of Hanoi. The second study involved a short (60 second) tracking task at each of three different target sizes and three different target speeds. In the third study, six individuals each practiced for four hours using the foot control and the mouse in the same tracking tasks. The results thus far are encouraging in providing data that shows increases in performance over short (one hour) and long (four hour) periods of practice. After the short amount of practice,
An increasing need exists for both a theoretical basis and practical human factors guidelines for designing and selecting high degree-of-freedom (DOF) computer input devices for 3D interactive environments such as telerobotic and virtual reality systems. This study evaluates elastic versus isometric rate control devices, in a 3D object positioning task. An experiment was conducted with a stereoscopic virtual reality system. The results showed that the elastic rate controller facilitated faster task completion time in the first of four phases of the experiment. The results are discussed in light of psychomotor literature. While the richer proprioceptive feedback afforded by an elastic controller is necessary for achieving superior performance in the early stages of learning, subjects performed equally well with the isometric controller in later learning stages. The study provides evidence to support a theory of skill shift from closed-loop to open-loop behaviour as learning progresses.
Virtual environments promise an almost limitless vista of expansion for human capabilities. They are being touted as the interface of the future and have begun to generate an expectation of a revolution in human-computer interaction greater than any seen to date. Like much hyperbole that cascades around innovative technology, little is based in knowledge while much is based on expectation or sheer speculation. In previous work, we have examined some of the basic human factors questions about usability. Here we specifically examine the ability to reproduce a complex spatial layout of objects having experienced them under a variety of viewing conditions. Subjects teamed the spatial layout of nine common objects arranged on a flat plane. Viewing conditions consisted of free binocular virtual, free binocular real, and monocular real. The first two allowed active exploration of the environment while the latter condition allowed only a single viewpoint. The dependent variables were mapping accuracy and triadic comparisons of relative inter-object distances. Mapping results showed a significant effect of viewing condition, where in contrast to expectations the single eye was superior to the virtual and real binocular conditions. Results for the triadic comparisons showed a significant interaction of gender by viewing condition. The spatial representation formed by using virtual reality appears equivalent to that of the representation with the actual objects. The are a number of implications of these data with respect to interface design, in particular the question of potential information overload in virtual interfaces.
Three dimensional hierarchies may be used to visualize and manipulate a variety of types of information, including tables of contents, taxonomies, fault trees, and code libraries in languages such as C++. Based on a review of the literature, it is suggested that visual scanning and perceived organization are two main processes involved in the understanding and use of 3D hierarchies. A general system (Info-TV) is described which can be used to explore a 3D design space. Info-TV is an information tree visualization system that takes hierarchical information and displays it in a three dimensional format. It has been designed to include the major parameters of the tree visualization design space and can be used to explore this space. Two experiments are reported that assess the effect of different tree shapes on different tasks. The results of these experiments are then discussed in terms of their implications for the design space of three-dimensional hierarchies.
Considerable research has been conducted on head or helmet-mounted displays (HMDs) for aviators, but little, if any, for ground soldiers. This study investigated the ability of individuals to perform combined tasks of walking over irregular terrain (simulated) and processing information from an HMD. Volunteers performed reaction time tasks while either standing or walking on a treadmill, with and without obstacles. Results show uniform effects of obstacles on three levels of cognitive difficulty. Walking on level ground with no obstacles had no more effect on information processing than merely standing, whereas minor and moderate levels of obstacle avoidance had equal impact. Spatial displays were easier than verbal and numeric displays during all levels of mobility. Subjective effort matched reaction time performance. Results support existing models of attention and will help provide guidance for unique applications for the ground combatant.
Aircraft inspection is still primarily a visual activity, thus the accuracy and efficiency of this visual inspection is ultimately determined by the adequacy of the visual environment. Based upon site visits to various commercial aircraft sites, the existing visual environment in aircraft inspection has been found to be generally inadequate. This study demonstrates a procedure which can be utilized to assist in selecting the appropriate lighting equipment for aircraft tasks. An evaluation was undertaken at a single commercial aircraft maintenance facility which included task analyses of typical inspection jobs combined with photometric evaluations of the ambient and task lighting. Portable and personal lighting sources were sampled and evaluated in the laboratory and on the hangar floor for both photometric performance and ease of use. In addition, inspector perceptions were collected from four facilities to obtain a wider base for comments and concerns related to the personal and portable lighting and the visual environment. Recommendations are made based upon the task demands, visual requirements, and other selected lighting considerations.
An experiment is reported in which the relative effectiveness of color coding, texture coding, and no coding of target borders to speed visual search is determined. The following independent variables were crossed in a within-subjects factorial design: Color coding (present or not), Texture coding (present or not), Distance between similarly coded targets (near or far), Group size of similarly coded targets (1, 2, 3, or 4), and a Replication factor of target Border width (10, 20, or 30 pixels). Search times, errors, and subjective rankings of the coding methods were recorded. Results showed that color coding improved search time compared to no coding, but that texture coding was not effectively used by subjects, resulting in nearly identical times to encoded targets. Subjective preference rankings reflected the time data. The adequate power of the experiment along with the results of preparatory pilot studies lead us to the conclusion that texture coding is not an effective coding method for improving visual search time.
Subjects viewed a series of alphanumeric tables containing information regarding the attributes (cost, amount, etc.) of different objects (utilities such as gas and electricity). They answered questions that required them to locate specific pieces of information in the table, perform simple integration between pieces, or complex integration (division, multiplication), and information for questions was either located within a table panel (close separation) or between panels (distant separation). The table was either organized by objects within attributes, or attributes within objects. Table organization had no effect on response time or accuracy. However, accuracy suffered with increased separation, but only for the complex integration questions, a finding that implicates the interference between visual search and the working memory demands of information integration. A computational model of the mental operations required for task performance accounted for 69% of the variance in response time, and provides a useful basis for developing more elaborate models of display layout.
Advanced road traffic management systems provide numerous opportunities for the application of sophisticated computer visualization concepts. The operating staff in a traffic control center are required to assimilate large quantities of incoming data in order to determine the real state of traffic flow and congestion. Part of the incoming data relates to vehicular speed and density, and is often not subjected to sufficient pre-processing before presentation in tabular form on a video display terminal (VDT). Improvements in the format of the tabular information are therefore worthy of investigation. A traffic control simulation experiment was conducted to examine how human subjects extract information from VDT data presented in several different formats. Subjects were asked to respond to exceptional values which occurred randomly in tabular columns of frequently changing data. Their accuracy and reaction time were measured for data columns which were sorted or unsorted, and for data which was presented either numerically or color-coded. Analysis of the results suggests that both sorting and color-coding are significant in reducing response time, and that color-coding is appreciably more effective in this regard.
Performance was evaluated in a single and dual task environment which simulated basic tasks carried out by sonar operators, using processed, visual representations of acoustic data. Two general classes of performance functions were obtained. One group of subjects performed at high levels in both single and dual task conditions and responded to increasing demand with increased throughput. The second group showed performance levelling or decrement once information load reached a critical level. The strategies used by the two groups (heuristics/pattern recognition for the former, and serial analysis for the latter) have clear importance for the type of decision aids which need to be provided in future generation sonar systems.
This paper describes an effort to understand the nature of decision tasks in the cockpit, their underlying cognitive requirements, the types of errors associated with each, and how crews can best be trained or aided. A scheme based on cue clarity and response availability was used to identify the cognitive requirements associated with classes of decision situations and to predict types of errors. Data from flight crews in full-mission simulators and from NTSB accident reports were analyzed to validate the analytical scheme.
There are several reasons to study decision errors. We can learn about reasoning processes from instances where they break down. We can also try to find ways to reduce the chances for decision errors. While there have been a number of excellent studies of naturally-occurring errors (e.g., Fitts & Jones, 1947; Norman, 1981; Rasmussen, 1982; Reason, 1990; Swain & Guttmann, 1983), these studies have examined the full range of human errors. The goal of my study was much more modest: to examine errors linked to faulty judgment, or inadequate problem solving or decision making, in order to gain a sense of why such errors would arise in field settings.
We examined decision-making in the real-world environment of trauma patient resuscitation and anesthesia in a Level One Trauma Center. The present paper focuses on the risk factors in the trauma treatment environment that can lead to errors or misjudgments, and strategies that may be helpful in reducing these risks. Video and audio recordings were made of a number of trauma cases involving tracheal incubation, including both emergency intubations performed during resuscitation and "elective" intubations prior to surgery. Post-treatment questionnaires completed by anesthesia personnel suggested that their perceived misjudgments were primarily procedural errors caused by lack of preparation for low probability events, inadequate monitoring of available indices, or carelessness. However, video analyses of a subset of the cases by a non-participant anesthesiologist, in conjunction with examination of patient management records, not only confirmed the occurrence of such errors but also identified instances of knowledge-based errors, which caused subsequent cascades of adverse events. Video analysis also documented the shortcuts that are characteristic of emergency intubations. The post-treatment questionnaires also suggested an association between team interactions and anesthesiologist performance. To follow up on this, we transcribed and categorized verbal communications for several minutes before, during, and after incubation in a subset of cases. This analysis indicated that during emergency intubations not only was more information communicated than during elective intubations, but that there were increases specifically in the incidence of directives, comments conveying plans or strategies, and comments both seeking and offering needed information. The discussion presents a number of strategies that emerged from the present analyses for reducing the risk factors involved in trauma treatment decision-making.