Multi-Attribute Task Research Articles

Cross-subject workload classification with a hierarchical Bayes model

Most of the current EEG-based workload classifiers are subject-specific; that is, a new classifier is built and trained for each human subject. In this paper we introduce a cross-subject workload classifier based on a hierarchical Bayes model. The cross-subject classifier is trained and tested with data from a group of subjects. In our work, it was trained and tested on EEG data collected from 8 subjects as they performed the Multi-Attribute Task Battery across three levels of difficulty. The accuracy of this cross-subject classifier is stable across the three levels of workload and comparable to a benchmark subject-specific neural network classifier.

Using the Analytic Hierarchy Process to Examine Judgment Consistency in a Complex Multiattribute Task

This paper investigates the impact of framing and time pressure on human judgment performance in a complex multiattribute judgment task. We focus on the decision process of human participants who must choose between pairwise alternatives in a resource-allocation task. We used the Analytic Hierarchy Process (AHP) to calculate the relative weights of the four alternatives (i.e., <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{1}$</tex></formula> , <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{2}$</tex></formula> , <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{3}$ </tex></formula> , and <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{4}$</tex></formula> ) and the judgment consistency. Using the AHP, we examined two sets of hypotheses that address the impact of task conditions on the weight prioritization of choice alternatives and the internal consistency of the judgment behavior under varying task conditions. The experiment simulated the allocation of robotic assets across the battlefield to collect data about an enemy. Participants had to make a judgment about which asset to allocate to a new area by taking into account three criteria related to the likelihood of success. We manipulated the information frame and the nature of the task. We found that, in general, participants gave significantly different weights to the same alternatives under different frames and task conditions. Specifically, in terms of ln-transformed priority weights, participants gave significantly lower weights to <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{2}$</tex></formula> and <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{4}$</tex></formula> and higher weight to <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex Notation="TeX">${\rm C}_{3}$</tex></formula> under gain frame than under loss frame, and also, under different task conditions (i.e., Tasks #1, #2, and #3), participants gave significantly higher weight to <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${ \rm C}_{4}$</tex></formula> in Task #1, lower weights to <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{1}$</tex> </formula> and <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{4}$</tex></formula> , higher weight to <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{3}$</tex></formula> in Task #2, and lower weight to <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm C}_{3}$</tex></formula> in Task #3. Furthermore, we found that the internal consistency of the decision behavior was worse, first, in the loss frame than the gain frame and, second, under time pressure. Our methodology complements utility-theoretic frameworks by assessing judgment consistency without requiring the use of task-performance outcomes. This work is a step toward establishing a coherence criterion to investigate judgment under naturalistic conditions. The results will be useful for the design of multiattribute interfaces and decision aiding tools for real-time judgments in time-pressured task environments.

Transcranial Doppler Assessment of Workload Transition in a Complex Task

Previous research has demonstrated that the level of performance achieved for a given workload level is affected not only on task demands and individual differences, but also by sequential effects of prior workload. The majority of this research has been done with vigilance-type tasks, wherein performance is an intermittent response to critical signals. Workload is then varied either by adjusting the signal rate or adding a secondary task. This study sought to demonstrate that workload can be assessed with Transcranial Doppler (TCD) in addition to performance and other physiological measures, and that workload transition results generalize to more complex tasks. TCD and electrocardiographic (ECG) data were collected while participants performed the Multi-Attribute Task Battery (MATB, Comstock &amp; Arnegard, 1992). Two levels of workload were individually defined based on performance via a pretest calibration procedure; one level was relatively easy with a mean composite TLX score of 23 and the other relatively difficult with a mean composite TLX score of 63. These two levels were run in isolation and in transition runs that switched from one to the other halfway through. TCD and ECG revealed that both blood flow velocity and heart rate increased with increasing task difficulty, while the transition in task difficulty produced nonsignifcantly lower performance than runs of consistent difficulty. The physiological data provides evidence that additional physical resources are demanded and consumed in response to the change in task difficulty, while performance data collected to date has not revealed transition effects in the complex task.

Personality and Reaction Time after Sleep Deprivation

The relationship between reaction time and both state and trait personality variables was investigated in 37 participants after 30 h of sleep deprivation. Regression analyses suggested that endorsement of greater Novelty Seeking, anger/hostility, and depression/dejection, and less confusion, was associated with greater reaction time declines on one Multi-Attribute Task Battery index after sleep deprivation. Further, greater Novelty Seeking and depression/dejection, and less vigor/activity, was associated with greater reaction time declines after sleep deprivation on another Multi-Attribute Task Battery index. Additional correlational analyses indicated that better reaction times were associated with greater Novelty Seeking and lower anger/hostility prior to sleep deprivation, and less confusion/bewilderment following sleep deprivation. Findings suggest that both state and trait personality variables are associated with reaction time performance following sleep deprivation.

Physiological responses to workload change. A test/retest examination

The purpose of this study is to examine the test/retest consistency of physiological responses induced by mental tasks. Fifteen healthy male university students were recruited as participants. They were instructed to perform a 5-min Multi-Attribute Task Battery (MATB) trial three times successively. The task difficulty level of the tracking task of the second trial was set as medium (M). The first one was set as more difficult (H) and the last trial was easiest (L). The difficulty levels of the other two tasks (resource management and system monitoring) of the MATB were identical for all three trials. The participants repeated this procedure on three different days separated by at least a 1-day interval. The order of the tasks was the same for all repeated trials, i.e., H–M–L. Tissue blood volume from the tip of the nose using a laser Doppler blood flow meter, skin potential level (SPL), ECG from three leads on the chest, systolic time intervals (pre-ejection period, left ventricular ejection time), and hemodynamic parameters (stroke volume, cardiac output) were recorded during the task trials and before and after 5-min resting periods. The participants reported their subjective workload via NASA-TLX after each task trial. Autonomic nervous system parameters derived from the above-mentioned signals, subjective workload scores, and performance indices of MATB were analyzed, and test/retest reliability was investigated. The results showed that a significant test/retest correlation was obtained for SPL for more participants than in the other parameters, although there were large individual differences.

Effects of loratadine/montelukast on vigilance and alertness task Performance in a simulated cabin environment

Sedating effects of some medications used to treat allergic rhinitis (AR) symptoms can impair an individual's ability to function optimally. The objective of this study was to characterize the effects of a single dose of loratadine/montelukast (L/M) versus placebo and diphenhydramine on daytime somnolence and psychomotor performance in healthy volunteers. In this single-center, randomized, double-blind, placebo- and active-controlled,three-way crossover study, healthy volunteers received single doses of placebo, L/M 10 mg/10 mg, and diphenhydramine 50 mg. Subjects (n=23) were evaluated under simulated cabin pressure using the following tools: Vigilance and Tracking Task (VigTrack), measuring vigilance and tracking performance; the Multi-Attribute Task Battery (MAT), measuring ability to perform multiple tasks simultaneously; and the Stanford Sleepiness Scale (SSS), measuring sedative effects of medication, at baseline and each hour from 1 to 6 hours postdose. Safety was monitored via adverse events and vital signs. Performances on VigTrack and MAT from 1 to 6 hours after dosing were not significantly different between L/M and placebo groups; in contrast, diphenhydramine resulted in significant impairment of tracking for up to 5 hours (P< or =0.01) and vigilance performance for up to 3 hours (P< or =0.05) on VigTrack versus placebo. Scores of subjective sleepiness as measured by SSS were similar for patients treated with L/M versus placebo, whereas significant increases in sleepiness occurred between 1-5 hours posttreatment in diphenhydramine-treated patients versus placebo-treated patients (P< or =0.05). L/M is similar to placebo in effects on daytime somnolence and psychomotor performance. L/M treatment resulted in significantly less sleepiness and impairment of vigilance and tracking than diphenhydramine.

Preliminary Examination of Timesharing in United States Air Force ROTC Cadets

Previous research has shown that timesharing ability--the ability to perform multiple tasks simultaneously--is related to various indices of aircraft pilots' proficiency. Superior timesharing has also been demonstrated by experienced pilots. The present study examined timesharing of United States Air Force Reserve Officer Training Corps (AFROTC) cadets, who are relatively inexperienced (some with no flying experience). 10 AFROTC cadets (M age = 21.4 yr., SD = 1.0) and 18 noncadets (M age = 19.8 yr., SD = 0.9) were evaluated. Participants performed a tracking task on the Multi-Attribute Task battery in a single-task condition, a dual-task condition, and a tri-task condition. Cadets' tracking performance was superior to that by noncadets in the dual-task and tri-task conditions.

Mechanisms of knowledge degradation in a resource management task ¶

Previous research has focused more on learning relative to forgetting despite both being generally recognized as important human performance issues. This research pursued an improved understanding of variables that impact the effectiveness with which humans can retrieve procedural knowledge, such as mathematics or programming. Two studies reviewed yielded varying results in experiments that investigated factors that can inhibit the retrieval of procedural knowledge from memory. To explore theoretical explanations for the opposing results, 32 human participants performed a resource management task in the multi-attribute task battery (MATB) manipulating task strategy and colour symbology. Human performance resulted in significant interference effects for a switch in strategy, but not with a change in the implicit meaning of colours. Therefore, the participants’ sensitivity to the implicit meaning of visual stimuli was not as significant as a more comprehensive change in rule structure associated with procedural knowledge. This work was performed while the first and second authors were at Wright State University.

Effects of Long Audio Communication Delays on Team Performance

Mediated communication in sociotechnical systems is quite common but the effects of long audio delays remain unexplored. In the current study, two-person teams (N=67) completed a modified NASA Multi-Attribute Task Battery with closed-loop audio communication delays 0 to 16 seconds in length. In addition to a joint fuel management task, each team member had simultaneous responsibility for either a compensatory tracking or a system monitoring task. As communication delay increased, performance on the joint task degraded in a cubic fashion while individual tasks were unaffected. These results imply that audio communication delay degrades team performance in a non-linear fashion while simultaneous tasks not requiring communication are unaffected. Implications for the design of sociotechnical systems with long audio communication delays are discussed.

Performance and Psychophysiological Measures of Fatigue Effects on Aviation Related Tasks of Varying Difficulty

Fatigue is a well known stressor in aviation operations and its interaction with mental workload needs to be understood. Performance, psychophysiological, and subjective measures were collected during performance of three tasks of increasing complexity. A psychomotor vigilance task, multi-attribute task battery and an uninhabited air vehicle task were performed five times during one night's sleep loss. EEG, ECG and pupil area were recorded during task performance. Performance decrements were found at the next to last and/or last testing session. The EEG showed concomitant changes. The degree of impairment was at least partially dependent on the task being performed and the performance variable assessed.

A quantitative model of the human–machine interaction and multi-task performance: A strategy function and the unity model paradigm

A human–machine-interaction (HMI) model is developed for the human operator (HO) performing five simultaneous tasks and characterized by a strategy function. Five levels of total machine-initiated baud rate ( B IN ) are generated by the multi-attribute task battery (MATB) and five HO baud rates ( B O ) are then recorded. Total baud ratio ( B ¯ ) is defined as the ratio of B O to B IN . Results indicate that with increasing B IN levels: (1) there is an overall increase in B O , and (2) there is an overall decrease in B ¯ . These results are due to a decreasing HMI performance and divergence of the strategy function from a unity model paradigm.

Physiological Reactivity to a Multi-Task Environment

A lot of physiological parameters, mainly cardiovascular indexes, were obtained from fifteen male participants. They were asked to perform a 5-minute multi-task (MATB: Multi-Attribute Task Battery) for three times. The MATB consisted of four tasks, i. e., compensatory tracking task, system monitoring task, resource management task and communication task. The communication task was not operating in this experiment. Two participants out of fifteen showed bradycardia during the performance of MATB comparing with before and after resting periods. However, many of other participants did not show clear heart rate decreases. Therefore, above-mentioned two participants should be grouped as Vascular Reactors who show dominant Total Peripheral Resistance increment (and heart rate decrements consequently) to mental challenge rather than Cardiac Reactors who show increments in Cardiac Output. In vascular Reactors, increased vagal tone (parasympathetic nervous system activity) is elicited by mental tasks. Therefore, when we evaluate the mental workload using autonomic nervous system measures, we may be confused by the results obtained if there are Vascular Reactors, Cardiac Reactors and Mixed Reactors in participants' group at the same time.

In Search of Workload Context Effects Using a Compensatory Tracking Task

This study examined whether or not prior levels of task difficulty (context effects) impacted subsequent performance and workload ratings. Previous research by Hancock, et al. (1995) found context effects for workload ratings. Similar studies, (Fischer, 1995; Moroney et al., 1993; Rench 2002) using flight simulation tasks had not demonstrated context effects. In the current study, twelve participants completed three sessions of three compensatory tracking trials each using the Multi-Attribute Task Battery (MATB) tracking task. RMSE scores and TLX ratings were analyzed for each trial. A repeated measures design presented the first trial (Baseline) and the third trial (Critical) at a medium difficulty level. The (context inducing) middle trial had three levels of difficulty (low, medium and high). Performance and workload ratings for the baseline and critical trial were compared and showed no difficulty by trial interaction. Therefore, contrary to Hancock, et al., no context effect was obtained.

Individual Difference Effects on Performance in a Complex, Multi-Decision-Making Environment

The present research examines the impact of cognitive style and short-term memory (STM) capacity on the computer interface usability of a pilot simulation. Participants screened for individual differences in cognitive style and STM capacity were randomly assigned to a training condition within the Multi-Attribute Task Battery (MAT; Arnegard &amp; Comstock, 1991). Training conditions varied in automation settings (adaptive versus static). Participants transferred to a novel static condition. Between subjects independent variables were individual differences in STM capacity (high, low), cognitive style (analytic, holistic, quasi-rational), and training (static versus adaptive). Within subjects independent variables were training (four blocks) and transfer (four blocks) sessions. Dependent variables were accuracy and reaction time. Results suggest that individual differences in STM capacity and cognitive style are orthogonal constructs that interact to affect skill acquisition and strategy development for visual-spatial tasks. Implications related to computer interface design are discussed.

Prediction of subjective states from psychophysiology: A multivariate approach

Biocybernetic systems utilise real-time changes in psychophysiology in order to adapt aspects of computer control and functionality, e.g. adaptive automation. This approach to system design is based upon an assumption that psychophysiological variations represent implicit fluctuations in the subjective state of the operator, e.g. mood, motivation, cognitions. A study was performed to investigate the convergent validity between psychophysiological measurement and changes in the subjective status of the individual. Thirty-five participants performed a demanding version of the Multi-Attribute Task Battery (MATB) over four consecutive 20-min blocks. A range of psychophysiological data were collected (EEG, ECG, skin conductance level (SCL), EOG, respiratory rate) and correlated with changes in subjective state as measured by the Dundee Stress State Questionnaire (DSSQ). MATB performance was stable across time-on-task; psychophysiological activity exhibited expected changes due to sustained performance. The DSSQ was analysed in terms of three subjective meta-factors: Task Engagement, Distress and Worry. Multiple regression analyses revealed that psychophysiology predicted a substantial proportion of the variance for both Task Engagement and Distress but not for the Worry meta-factor. The consequences for the development of biocybernetic systems are discussed.

The influence of task demand and learning on the psychophysiological response

The level of expertise of an operator may significantly influence his/her psychophysiological response to high task demand. A naïve individual may invest considerable mental effort during performance of a difficult task and psychophysiological reactivity will be high compared to the psychophysiological response of a highly skilled operator. A study on multitasking performance was conducted to investigate the interaction between learning and task demand on psychophysiological reactivity. Thirty naïve participants performed high and low demand versions of the Multi-attribute Task Battery (MATB) over a learning period of 64 min. High and low task demand setting were preset via a pilot study. Psychophysiological variables were collected from four channels of EEG (Cz, P3, P4, Pz), ECG, EOG and respiration rate to measure the impact of task demand and learning. Several variables were sensitive to the task demand manipulation but not time-on-task, e.g., heart rate, Θ activity at parietal sites. The sensitivity of certain variables to high demand was compromised by skill acquisition, e.g., respiration rate, suppression of α activity. A sustained learning effect was observed during the high demand condition only; multiple regression analyses revealed that specific psychophysiological variables predicted learning at different stages on the learning curve. The implications for the sensitivity of psychophysiological variables are discussed.

CARDIAC CONTROL DURING DUAL-TASK PERFORMANCE OF VISUAL OR AUDITORY MONITORING WITH VISUAL-MANUAL TRACKING

Thirty university students (15 females) performed tasks from the NASA Multi-attribute Task Battery, which was used to manipulate the attentional processing resource demands of the tasks. Separate groups performed visual or auditory monitoring tasks singly and in combination with visual-manual tracking while heart period, respiratory sinus arrhythmia (RSA), and pre-ejection period (PEP) were recorded to determine the modes of cardiac control during task performance. The tracking single-task suppressed RSA and was interpreted as eliciting an uncoupled parasympathetic inhibition mode of cardiac control because of its demand upon perceptual and manual processing resources. The monitoring task both suppressed RSA and shortened PEP and was interpreted as eliciting a reciprocally-coupled sympathetic activation and parasympathetic inhibition mode of cardiac control because of its demand for perceptual/central processing resources. Dividing attention elicited uncoupled parasympathetic inhibition that was greater for the visual monitoring and tracking dual-task than for the auditory monitoring and tracking dual-task, which we interpreted as indicating a greater sharing of perceptual/spatial and manual processing resources for these tasks. Practice lengthened PEP in the dual-tasks and was interpreted as eliciting uncoupled sympathetic inhibition that was indicative of improved central processing resource efficiency. The results are evaluated in terms of how knowledge of cardiac control modes may be used to assess mental workload in applications such as adaptive automation.

An effectiveness analysis of information channeling in choice decision making with attribute importance weights

In multi-alternative, multi-attribute choice decision tasks, decision makers use either alternative-based or attribute-based information processing patterns. Evidence suggests that channeling of access patterns may be effective. Restricted search of only key information attributes may be further encouraged when importance weights for attributes are predetermined and provided to the decision makers. This study examines the effectiveness of alternative-based channeling and attribute-based channeling with or without the provision of attribute importance weights. Both alternative-based and attribute-based channeling improves the decision accuracy when attribute weights are provided. In addition, the results indicate statistically significant effects on decision accuracy for the type of information display.

Establishing the Psychophysiological Variables that can Identify &amp; Predict Operator Subjective State

Biocybernetic adaptation describes a system wherein automation is regulated by the operator's physiology. The identification of real-time psychophysiological variables capable of detecting and predicting operator subjective states is an important requisite for biocybernetic control. The present study investigated how psychophysiology may index negative subjective states, namely, task-disengagement, distress and worry. To encourage negative states, 35 participants performed the (high demand) Multi-Attribute Task Battery (MATB), for 100 minutes in total. Subjective state was assessed using the Dundee Stress State Questionnaire (DSSQ). The psychophysiological measures recorded included: GSR, EOG, EEG, ECG, and respiration. Analysis of the data revealed performance was sustained throughout the task although subjective states became more negative over time. This was accompanied by various psychophysiological changes, e.g. increased respiration rate. Regression analyses indicated that psychophysiological changes were predictive of changes in subjective states. The results and implications for biocybernetic systems are discussed.

Real-time assessment of mental workload using psychophysiological measures and artificial neural networks.

The functional state of the human operator is critical to optimal system performance. Degraded states of operator functioning can lead to errors and overall suboptimal system performance. Accurate assessment of operator functional state is crucial to the successful implementation of an adaptive aiding system. One method of determining operators' functional state is by monitoring their physiology. In the present study, artificial neural networks using physiological signals were used to continuously monitor, in real time, the functional state of 7 participants while they performed the Multi-Attribute Task Battery with two levels of task difficulty. Six channels of brain electrical activity and eye, heart and respiration measures were evaluated on line. The accuracy of the classifier was determined to test its utility as an on-line measure of operator state. The mean classification accuracies were 85%, 82%, and 86% for the baseline, low task difficulty, and high task difficulty conditions, respectively. The high levels of accuracy suggest that these procedures can be used to provide accurate estimates of operator functional state that can be used to provide adaptive aiding. The relative contribution of each of the 43 psychophysiological features was also determined. Actual or potential applications of this research include test and evaluation and adaptive aiding implementation.