Cognitive Workload and Fatigue in a Vigilance Dual Task: Miss Errors, False Alarms, and the Effect of Wearing Biometric Sensors While Working.

The effects of workload, fatigue, and practice on the performance of cognitive tasks are often intertwined. Previous research has shown that these influences can be separated with the two cusp catastrophe models. This study expanded an earlier investigation of the two models for workload and fatigue in a vigilance task to include a wider range of bifurcation variables that could affect the elasticity versus rigidity of the operator in response to workload and added performance variability resulting from fatigue. The study also responded to a concern in the literature that performance on cognitive tasks can be complicated by adaptive responses to artificial task situations and thus distort underlying cognitive events. Therefore, we also explored whether wearing biometric sensors, frequently used in workload studies, can affect performance dynamics. Participants were 279 undergraduates who responded to target stimuli that appeared on a simulated security camera display at three rates of speed while completing a secondary task. Participants worked alone, in pairs, or in pairs wearing GSR sensors. Results supported the efficacy of the two models and isolated the impact of wearing sensors on the fatigue process. The strongest control variables across both the workload and fatigue models were field independence, anxiety, indecisiveness, inflexibility, secondary task completion, working in pairs, and wearing the sensors. The contributing effect of wearing sensors could possibly extend to other types of wearable technologies.

Estimating Appropriate Lag Length for Synchronized Physiological Time Series: The Electrodermal Response.

Physiological synchronization of autonomic arousal between people is thought to be an important component of work team dynamics, therapist-client relationships, and other interpersonal dynamics. This article examines concepts and mathematical models of synchronization that could be relevant to work teams. Before it is possible to deploy nonlinear modeling, however, it is necessary to develop a strategy for determining appropriate lag lengths. If a measurement at time 2 is a function of itself at time 1 and a coupling effect from another source, what is the appropriate amount of real time that should be allowed to elapse between the two measurements in order to observe the coupling effect? This study examined four strategies for doing so. In the experiment, 78 undergraduates worked in pairs to perform a vigilance dual task for 90 min while galvanic skin responses (GSR) were recorded. Lags based on mutual entropy and the natural rate criteria produced corroborating results, whereas strategies based on a critical decline in the linear autocorrelation (max r/e) and Theiler's W did not produce usable results for this situation. Some connections were uncovered between linear autocorrelation strength and lag based on mutual entropy with performance on the tasks and subjective ratings of workload.

Catastrophe models for cognitive workload and fatigue in N-back tasks.

N-back tasks place a heavy load on working memory, and thus make good candidates for studying cognitive workload and fatigue (CWLF). This study extended previous work on CWLF which separated the two phenomena with two cusp catastrophe models. Participants were 113 undergraduates who completed 2-back and 3-back tasks with both auditory and visual stimuli simultaneously. Task data were complemented by several measures hypothesized to be related to cognitive elasticity and compensatory abilities and the NASA TLX ratings of subjective workload. The adjusted R2 was .980 for the workload model, which indicated a highly accurate prediction with six bifurcation (elasticity versus rigidity) effects: algebra flexibility, TLX performance, effort, and frustration; and psychosocial measures of inflexibility and monitoring. There were also two cognitive load effects (asymmetry): 2 vs. 3-back and TLX temporal demands. The adjusted R2 was .454 for the fatigue model, which contained two bifurcation variables indicating the amount of work done, and algebra flexibility as the compensatory ability variable. Both cusp models were stronger than the next best linear alternative model. The study makes an important step forward by uncovering an apparently complete model for workload, finding the role of subjective workload in the context of performance dynamics, and finding CWLF dynamics in yet another type of memory-intensive task. The results were also consistent with the developing notion that performance deficits induced by workload and deficits induced by fatigue result from the impact of the task on the workspace and executive functions of working memory respectively.

Catastrophe models for cognitive workload and fatigue in a vigilance dual task.

OBJECTIVE: This study investigated two cusp catastrophe models for cognitive workload and fatigue for a vigilance dual task, the role of emotional intelligence and frustration in the performance dynamics, and the dynamics for individuals and teams of two participants. BACKGROUND: The effects of workload, fatigue, practice, and time on a specific task can be separated with the two models and an appropriate experimental design. Group dynamics add further complications to the understanding of workload and fatigue effects for teams. METHOD: In this experiment, 141 undergraduates responded to target stimuli that appeared on a simulated security camera display at three rates of speed while completing a jigsaw puzzle. Participants worked alone or in pairs and completed additional measurements prior to or after the main tasks. RESULTS: The workload cusp verified the expected effects of speed and frustration on change in performance. The fatigue cusp showed that positive and negative changes in performance were greater if more work on the secondary task was completed and whether the participants who started with the fast vigilance condition demonstrated less fatigue. CONCLUSION: The results supported the efficacy of the cusp models and suggested, furthermore, that training modules that varied speed of presentation could buffer the effects of fatigue. APPLICATION: The cusp models can be used to analyze virtually any cognitively demanding task set. The particular results generalize to vigilance tasks, although a wider range of conditions within vigilance tasks needs to be investigated further.

The performance-variability paradox, financial decision making, and the curious case of negative Hurst exponents.

This study examined the relationship between performance variability and actual performance of financial decision makers who were working under experimental conditions of increasing workload and fatigue. The rescaled range statistic, also known as the Hurst exponent (H) was used as an index of variability. Although H is defined as having a range between 0 and 1, 45% of the 172 time series generated by undergraduates were negative. Participants in the study chose the optimum investment out of sets of 3 to 5 options that were presented a series of 350 displays. The sets of options varied in both the complexity of the options and number of options under simultaneous consideration. One experimental condition required participants to make their choices within 15 sec, and the other condition required them to choose within 7.5 sec. Results showed that (a) negative H was possible and not a result of psychometric error; (b) negative H was associated with negative autocorrelations in a time series. (c) H was the best predictor of performance of the variables studied; (d) three other significant predictors were scores on an anagrams test and ratings of physical demands and performance demands; (e) persistence as evidenced by the autocorrelations was associated with ratings of greater time pressure. It was concluded, furthermore, that persistence and overall performance were correlated, that 'healthy' variability only exists within a limited range, and other individual differences related to ability and resistance to stress or fatigue are also involved in the prediction of performance.