Metamemory, the ability to interrogate the contents of one's permanent memory, may be an important factor in using one's knowledge to cope with the environment. Metamemorial accuracy and efficiency can be assessed. Metamemory is accurate if it returns correct information about the contents in store. It is efficient if it appropriately controls search durations so that more time is allocated to seeking information actually present, less to information actually absent. Adult subjects in three age groups answered questions on heterogeneous topics, and their responses were timed. Next, metamemorial judgments were made for each subject's set of unanswered questions. The same items were then attempted in multiple-choice format, and confidence ratings in the answers were taken. All age groups showed comparable ability to retrieve answers from memory. All showed accurate and efficient metamemory, with no age differences in either. A signal detection analysis raised the possibility that metamemorial sensitivity increases with age. The data also suggested caution among the elderly in suppressing available but lowconfidence answers.

Future manned space operations will include a greater use of automation than we currently see. 1 For example, semiautonomous robots and software agents will perform difficult tasks while operating unattended most of the time. As these automated agents become more prevalent, human contact with them will occur more often and become more routine, so designing these automated agents according to the principles of human-centered computing is important.In this article, we describe two cases of semiautonomous control software developed and fielded in test environments at the NASA Johnson Space Center. This software operated continuously at the JSC and interacted closely with humans for months at a time. Our approachFor the past seven years, we've worked on developing intelligent software for the control of advanced life support systems. We fielded this control software in an operational environment in which test engineers manually controlled and continuously monitored all life support systems from a console in a test control room. Such operations required the engineers to spend considerable time on routine data monitoring and lowlevel commanding. Our biggest challenges initially were to prove that automated control software was reliable enough to be useful and that automating routine control tasks would be worthwhile.Thus, from the beginning, we had the goal of using automation to reduce the engineer's workload. However, our objective was not to replace humans in operations but to free them from routine tasks (such as vigilant monitoring), thereby enabling them to concentrate on activities that capitalize on human strengths (such as supervisory monitoring). To perform these new tasks, humans still must interact with the control automation. In fact, human interaction becomes more challenging because the human is less involved in routine day-to-day operations and, as a result, might be less aware of the ongoing control situation and could lose anomaly response skills through lack of practice. This is a critical consideration for the human centering of semi-autonomous control systems.We also recognized that the change in test operations resulting from the use of automated control would be fundamental. The human role changes to one of supervisory monitoring with occasional intervention when operations cannot be automated or when exceptional situations occur. During normal operations, engineers supporting these tests will spend most of their time doing activities unrelated to control but will need to be on call should the automation or life support hardware experience problems. In addition, humans will need to supervise and command these continuously operating systems from remote locations (such as their offices) with only infrequent (and possibly asynchronous) interaction. For such operations, human supervisors must be able to quickly form an integrated view of distributed control without having to continuously monitor control data.This concept of test operations, however, represented too radical a change to be quickly acc...

An important aspect of interaction among groups of humans and software agents is supporting collaboration among these heterogeneous agents while they operate remotely and communicate asynchronously. We are developing an architecture that supports multiple humans interacting with multiple automated control agents in such a manner. We are evaluating this architecture with a group consisting of the crew of a space-based vehicle and the automated software agents controlling the vehicle systems. Such agent interaction is modeled as a loosely coordinated group because this model minimizes agent commitment to group goals and constraints while addressing a significant portion of crew and control agent group behaviors. In this paper we give background on human interaction with space-based automation. We identify related research in multi-agent autonomous architectures and single agent human-computer interaction systems. We describe our architecture design for human-software agent groups. And we identify research issues in loosely coordinated human-software groups.

No abstract

ABSTRACTprocessing, mismatches between a human's mental models and the This paper describes an implemented software prototype for the agent's implementation models, lack of adequate adjustability for Distributed Collaboration and Interaction (DCI) system, which the agent's autonomy, lack of notification to the human about helps humans to act as an integrated part of a multi-agent system. important events at appropriate levels of abstraction, and a basic Human interaction with agents who act autonomously most of the incompatibility between a human's natural interface modalities time, such as a process control agent in a refinery, has received and the usual interfaces provided by an autonomous control agent. little attention compared to human interaction with agents whoWe have developed the Distributed Collaboration and Interaction provide a direct service to humans, such as information retrieval.(DCI) system to address these difficulties in human-agent This paper describes how liaison agents within the DCI system interaction and to create an environment in which humans and can support human interaction with other agents that are not, by mostly-autonomous software agents together can form an design, human-centric but must be supervised by, or coordinated integrated multi-agent system. The initial motivation for the DCI with, humans. The DCI system provides a step toward future system arose from our experiences with deployed intelligent seamless integration of humans and software agents into a control agents for NASA advanced life support systems. These cohesive multi-agent system. intelligent control agents monitor and perform process control for regenerative life support systems, which recover usable water or