Academic journal article Human Factors

Effects of a Biocybernetic System on Vigilance Performance

Academic journal article Human Factors

Effects of a Biocybernetic System on Vigilance Performance

Article excerpt


Adaptive automation refers to systems in which decisions regarding initiation, cessation, and mode of operation are shared between the human operator and the system in real time (Parasuraman, 2000; Scerbo, 1996). Adaptive systems provide for a tighter coupling between changing user needs and the operation of the system (Parasuraman, Bahri, Deaton, Morrison, & Barnes, 1992). The object of adaptive systems is to adjust situational demands, restructure the environment, and maintain more stable levels of workload, thereby enhancing operator performance. Interest in adaptive automation is fueled by concerns about the difficulties operators have when working with complex systems with multiple modes of automation (Woods, 1996). Indeed, recent research with process control operations (Moray, Inagaki, & Itoh, 2000) and piloting decisions (Inagaki, Takae, & Moray, 1999) have begun to show that superior human-system performance can be achieved in an adaptive context.

One critical issue in the development of adaptive automation concerns the mechanism that should be used to initiate system changes. A number of possibilities exist, such as subjective assessments of workload, indices of operator performance, and physiological measures of arousal and workload (e.g., Gomer, 1981; Hancock, Chignell, & Lowenthal, 1985; Kramer, Trejo, & Humphrey, 1996). Byrne and Parasuraman (1996) suggested the use of physiological measures in the design and regulation of adaptive systems because such measures are relatively unobtrusive, as compared with subjective or secondary task measures, and can allow an on-line assessment of operator workload and effort.

Little empirical work has been done on the development and evaluation of physiological measures with respect to adaptive automation. Four recent studies, however, have examined the viability of a system using the operator's own encephalographic (EEG) activity (Freeman, Mikulka, Prinzel, & Scerbo, 1999; Freeman, Mikulka, Scerbo, Prinzel, & Clouatre, 2000; Pope, Bogart, & Bartolome, 1995; Prinzel, Freeman, Scerbo, Mikulka, & Pope, 2000). In each of these studies individuals were asked to perform a compensatory tracking task. Operation of the task would switch back and forth between manual conditions, in which the operator performed the task, and automated conditions, in which the operator monitored the task. Changes between the automated and manual modes were driven by an EEG-based index of engagement. More specifically, the engagement index represents a ratio of EEG power bands [e.g., beta/(alpha + theta)]. Research has shown a direct relationship between beta activity and alertness and an indirect relationsh ip between alpha and theta activity and alertness (e.g., see O'Hanlon & Beatty, 1977).

More recently several investigators have found that ratios among these EEG power bands may provide a better index of alertness or task engagement than does any power band by itself (Lubar, 1991; Lubar, Swartwood, Swartwood, & O'Donnell, 1995; Pope et al., 1995; Streitberg, Rohmel, Herrmann, & Kubicki, 1987). Thus in the adaptive system described previously, the operator's index of engagement was used to trigger switches between task modes in two different ways. Under negative feedback conditions, when the index of engagement increased, the tracking task switched to automatic mode, which in turn resulted in a lower BEG index of engagement. Conversely, if an individual's engagement index decreased, the tracking task switched to manual mode, which in turn would increase workload and subsequently increase the engagement index.

As designed, this procedure should keep the operator at a moderate level of engagement and thereby avoid excessive levels of high or low workload. Under positive feedback conditions, however, high levels of engagement index switched the task to manual mode and low levels switched the task to automatic mode. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.