The Vigilance Decrement Reflects Limitations in Effortful Attention, Not Mindlessness

Transcription

1 The Vigilance Decrement Reflects Limitations in Effortful Attention, Not Mindlessness Rebecca A. Grier, Joel S. Warm, William N. Dember, Gerald Matthews, and Traci L. Galinsky, University of Cincinnati, Cincinnati, Ohio, James L. Szalma, University of Central Florida, Orlando, Florida, and Raja Parasuraman, Catholic University of America, Washington, D.C. Robertson, Manly, Andrade, Baddeley, and Yiend (1997) proposed that the decline in performance efficiency over time in vigilance tasks (the vigilance decrement) is characterized by "mindlessness" or a withdrawal of attentional effort from the monitoring assignment. We assessed that proposal using measures of perceived mental workload (NASA-TLX) and stress (Dundee Stress State Questionnaire). Two types of vigilance task were employed: a traditional version, wherein observers made button-press responses to signify detection of rarely occurring critical signals, and a modified version, developed by Robertson et al. to promote mindlessness via routinization, wherein button-press responses acknowledged frequently occurring neutral stimulus events and response withholding signified critical signal detection. The vigilance decrement was observed in both tasks, and both tasks generated equally elevated levels of workload and stress, the latter including cognitions relating to performance adequacy. Vigilance performance seems better characterized by effortful attention (mindfulness) than by mindlessness. Actual or potential applications of this research include procedures to reduce the information-processing demand imposed by vigilance tasks and the stress associated with such tasks. INTRODUCTION Vigilance or sustained attention tasks typically require observers to monitor displays over extended periods for the occasional occurrence of critical events (signals). The signals are usually embedded in a context of recurrent nonsignal (neutral) events that, unlike signals, require no overt response from observers. Signal probability is low: Typically, in laboratory research, only 2% to 5% of the events displayed are critical signals. In "real-world" instances, signal probability may be even lower (Parasuraman, Warm, & Dember, 1987). Vigilance tasks and the processes that influence their performance are of interest to human factors/ergonomics specialists because of the vital role that vigilance plays in automated human-machine systems in transportation, process and quality control, medicine, and - of special interest currently - baggage inspection at airport security checkpoints (Hancock & Hart, 2002; Wickens & Hollands, 2000). Traditionally, vigilance tasks have been considered as tedious but benign assignments that place little demand on monitors, and the vigilance decrement - the decline in signal detections over time that typifies vigilance performance -has been viewed as resulting from task underload and consequent underarousal (Nachreiner & Hanecke, 1992). Recent studies indicate, however, that vigilance tasks, though tedious, impose a substantial mental burden on monitors, as reflected in high scores on the NASA Task Load Index (NASA-TLX), a measure of the perceived mental workload incurred in performing a task (Hitchcock, Dember, Warm, Moroney, & See, 1999; Temple et al., 2000; Warm, Dember, & Hancock, Address correspondence to Joel S. Warm, D tmetuof Psychology, University of Cincinnati, Cincinnati, OH 45221; joel.warnm@uc.edu. HUMAN FACTORS, Vol45 Nj3 Fall 2003, pp Copyright 2003, Human Factors and Ergonomics Society. All rights reserved.

2 350 Fall Human Factors 350 FaIl Human Factors 1996). Such results are consistent with the notion that performance failures in vigilance tasks result from a decline in information-processing resources rather than from a diminution of arousal (Matthews, Davies, Westerman, & Stammers, 2000). In addition to imposing a substantial workload, vigilance tasks can be highly stressful. Studies employing self-report indices have demonstrated task-induced negative mood shifts and increases in restlessness, subjective fatigue, sleepiness, and headaches across the watch (Hancock & Warm, 1989; Warm, 1993). Given current transactional models of stress, in which stress is viewed as arising from individuals' appraisal of their environment as taxing or exceeding their coping resources (Matthews, 2001), these results are also consistent with the idea that detection failures in vigilance stem from capacity drain brought about by prolonged effortful attention. An alternative interpretation of increasing detection failures in vigilance over time comes from a suggestion by Shallice and his associates (Shallice, 1988; Stuss, Shallice, Alexander, & Picton, 1995): that when confronted with repetitive tasks in which signals are separated by long intervals, as in the case of vigilance, a supervisory attentional system loses its potency and observers cease to focus their awareness on the task at hand. Accordingly, Robertson and his colleagues (Manly, Robertson, Galloway, & Hawkins, 1999; Robertson, Manly, Andrade, Baddeley, & Yiend, 1997) have advanced the proposition that the repetitive nature of vigilance tasks leads to a "mindless" lack of attentional focus and thence to failures of signal detection. According to this view, mindlessness is defined as a thoughtless, routinized approach characterized by withdrawal of effortful attention away from the task at hand. It reflects an endogenous modulation of attention rather than the decline in wakefulness and vigor accompanying lowered arousal (Dickman, 2002). Based on their conception, Robertson and his associates (Manly et al., 1999; Robertson et al., 1997) have introduced a modification of the standard vigilance paradigm designed to increase the degree of mindlessness: Monitors are required to respond overtly to nonsignals and to withhold responding to signals. Increased detection failures in the modified vigil can presumably be attributed to routinization, automaticity, and lapses of attentional focus (all features of "mindlessness"; see LaBerge, 1995; Langer, 1989) generated by uniform, repetitive responding to the considerably more numerous nonsignal events. The Robertson group (Manly et al., 1999; Robertson et al., 1997) found support for the general role of mindlessness in vigilance and for its specific role in their modified task in their observation that absentminded individuals, defined by high scores on the Cognitive Failures Questionnaire (CFQ; Broadbent, Cooper, Fitzgerald, & Parkes, 1982), do more poorly on the modified task than do those who have low scores on the CFQ. The validity of this line of argument is contested, however, by the finding (not cited by Manly et al. or Robertson et al.) that absentminded monitors perform as well on a traditional vigilance task as nonabsentminded monitors but rate that task as more mentally demanding on the NASA-TLX than do nonabsentminded individuals (Grubb et al., 1994). Apparently, in order to perform as well as low-cfq monitors, those high on the CFQ must devote more mental resources to the task, suggesting that neither group is performing in a "mindless" manner. The present study was designed to further pursue the validity of that issue by providing the initial experimental comparison of workload and stress ratings in the modified vigilance task with those in a more traditional vigilance task. Assuming that mindlessness involves the withdrawal of effortful attention away from the task at hand, the workload and stress scores associated with the modified task should be lower than those associated with a traditional vigilance task that induces less mindlessness. Such results would support the Robertson group's (Manly et al., 1999; Robertson et al., 1997) argument that mindlessness plays a role in controlling vigilance performance in general. In contrast, finding similarly high workload and stress scores in the modified and traditional vigilance tasks would challenge the view that mindlessness plays some role in vigilance in general and a key role in the modified task in particular, and, by default, that it strengthens the resource depletion account.

3 VIGILANCE AND MINDLESSNESS VIGILANCE AND MINDLESSNESS 351 Participants METHOD Sixty-four University of Cincinnati undergraduates (32 women, 32 men) participated to satisfy a course requirement. All of the monitors had normal or corrected-to-normal vision and were naive as to the purpose of the study. Of these, half (16 women, 16 men) were assigned at random to a standard-vigil condition and the other half to a modified-vigil condition. In addition, supplementary groups of 32 women and 32 men, drawn from the same population of undergraduates as used in the vigil, performed a card-sorting task, the purpose of which will be described later in this paper. Apparatus All monitors took part in a 50-min vigil, divided into five consecutive 1 0-min periods of watch, during which they assumed the role of industrial quality control inspectors in an electronics factory, monitoring images of parts appearing on a video display terminal (VDT). They were tested individually in a 1.95 x 1.90 x 1.88 m Industrial Acoustics Sound Chamber. The VDT was situated on a table approximately 66 cm from the seated observer. Ambient illumination in the chamber (4.46 cd/m 2 ) was provided by a 25- watt bulb mounted in a parabolic reflector and placed above and behind the monitors to preclude glare on the VDT. As illustrated in Figure 1, the images of the parts to be inspected were displayed on a gray screen and consisted of a large central green dot (1.50 cm in diameter) flanked along a horizontal vector by two smaller dots (0.5 cm in diameter). The transluminance of the screen was cd/m 2 ; that of the display elements was 22.5 cd/m 2. Good parts (neutral events) were those in which both flanking dots were equidistant (1 cm) from the centering dot. Damaged parts (critical signals for detection) were instances in which one of the flanking dots (left or right) was farther (1.5 cm) from the centering dot than the other. In all conditions, parts appeared on the VDT at a rate of 20/min with a dwell time of 200 ms (stimulus onset asynchrony = 3000 ms). Ten damaged parts (critical signals) appeared on a random schedule during each 10-min period of the vigil (signal probability=.05/period). A Power Macintosh 6100/66 computer orchestrated stimulus presentations and response recording. Perceived mental workload and task-induced stress were measured by the NASA-TLX and the Dundee Stress State Questionnaire (DSSQ: Matthews et al., 1999, 2002), respectively. The NASA-TLX, which has a test-retest reliability of.83 (Hart & Staveland, 1988), is considered one of the most effective measures of perceived workload currently available (Wickens & Hollands, 2000). The DSSQ, a well-validated instrument for assessing transient states associated critical signal Figure 1. An illustration of the display showing acceptable (neutral events) and unacceptable (critical signals) parts.

4 352 with stress, arousal, and fatigue, is designed to reflect the multidimensional nature of stress and has been used extensively to measure the stress of sustained attention (Helton, Dember, Warm, & Matthews, 2000; Matthews et al., 1999, 2002; Szalma et al., in press; Temple et al., 2000). The instrument contains 10 scales: Energetic Arousal (alertness-sluggishness), Tense Arousal (nervousness-relaxation), Hedonic Tone (pleasant vs. unpleasant mood), Task Motivation, Self-Focused Attention (self-preoccupation, daydreaming, etc.), Self-Esteem, Concentration, Confidence and Control, Task-Related Cognitive Interference (worry about performance adequacy), and Task-Irrelevant Cognitive Interference (self-oriented thoughts that are not task related). Alpha coefficients for the scales range from.76 to.89 (Matthews et al., 1999). Factor analysis of the DSSQ scales (Matthews et al., 2002) has identified three higher-order factors labeled worry, distress, and task engagement. The worry factor is defined by cognitive dimension of stress state, encompassing primarily the Task-Related and Task-Irrelevant Cognitive Interference, Self-Focused Attention, and Self-Esteem scales. The distress factor reflects the importance of unpleasant arousal and low perceived control in states of affective distress. It encompasses the Tense Arousal, Hedonic Tone, and Confidence and Control scales. Task engagement relates to alertness and task-directed effort. It encompasses the Energetic Arousal, Motivation, and Concentration scales. For the purposes of this study, task-induced stress was measured in terms of the three factor elements of the DSSQ plus the Task-Related Cognitive Interference scale. This scale constitutes a subset of the items in the Cognitive Interference Questionnaire, which has been used extensively in anxiety research (Sarason, Sarason, & Pierce, 1990). Procedure Monitors in the standard vigilance condition signified their rejection of damaged parts (detection of critical signals) by pressing a response key on a computer keyboard. In the modified condition, monitors signified their acceptance of good parts by pressing the response key and their rejection of damaged parts (detection of critical signals) by not pressing the key. In the Fall Human Factors standard condition, monitors were instructed to execute a rejection response after the image of the damaged part left the video screen but before the image of the next part appeared. Key presses signifying rejection occurring within this 2800-ms window were recorded as correct detections or hits. All other responses were recorded as inappropriate rejections or false alarms. Failures to execute key presses upon the presentation of damaged parts were recorded as detection failures or misses. In the modified condition, monitors were instructed to execute an acceptance response to a good part within the 2800-ms window between the offset of the image of that part and the onset of the image of the succeeding part. Key presses occurring within this window were recorded as correct acceptances. Monitors were credited with the correct rejection of a damaged part (correct detection) if they did not press the response key during the 2800-ms window between the offset of the image of the damaged part and the appearance of the next part. Inappropriate acceptances (detection failures or misses) were defined by key presses to damaged parts, whereas inappropriate rejections (false alarms) were defined by failures to execute a key press upon the appearance of good parts. Prior to the initiation of the main vigil, monitors in both the standard and modified conditions had an 8-min practice session with computercontrolled feedback in which a female voice informed the monitor of correct detections (hits) and detection failures (misses), based on the type of response (key press or the lack thereof) appropriate for the condition in which the observers were participating. Monitors in the standard condition also clearly understood that a key-press response unaccompanied by verbal feedback constituted an inappropriate rejection or false alarm. Likewise, those in the modified group understood that the withholding of a key press unaccompanied by verbal feedback constituted a false alarm. Feedback was not available during the main vigil. Monitors were required to have a minimum of 75% correct detections during the practice session to be retained in the study. Within the standard and modified groups, 16 monitors (8 women, 8 men) were selected at random to complete the NASA-TLX, and the

5 VIGILANCE AND MINDLESSNESS VIGILANCE AND MINDLESSNESS remainder completed the DSSQ. This procedure was employed to minimize possible interscale interactions. Those who evaluated the workload of the vigilance task completed a computerized version of the NASA-TLX once, immediately upon conclusion of the vigil, according to a protocol described by Hart and Staveland (1988). Those who were administered the DSSQ completed a paper-and-pencil pretest version of the scale upon arrival at the laboratory and a posttest version immediately at the end of the vigil using a procedure described by Matthews et al. (1999). Monitors surrendered their watches and cell phones upon entering the testing chamber and had no knowledge of the length of the vigil other than it would not exceed 90 mi. Participants in the card-sort task worked for 240 s sorting for suit while being paced by an auditory indicator (computerized metronome) at the rate of one card/ 1.5 s. Thirty-two of the 64 card-sort participants (16 women, 16 men) were selected at random to complete the NASA- TLX at the end of the task. The remainder completed the pre- and posttask versions of the DSSQ. The card-sort task was included to ensure that the absence of workload and/or stress differences between the two vigilance conditions could not be attributed to a lack of sensitivity on the part of the NASA-TLX or the DSSQ. Separate groups of participants were employed for the vigilance and card-sort task to avoid the possibility of unwanted carryover effects from one task to the other. Performance RESULTS Mean percentages of correct detections in the standard-vigil and modified-vigil conditions are plotted as a function of periods of watch in Figure 2. A 2 (conditions) x 5 (periods) analysis of variance (ANOVA) of an arcsin transformation of the detection scores indicated that overall detection probability declined significantly over time, F(4, 220) = 10.32, p <. 001, (w 2 =.03, and that although there was no overall main effect for condition, (p >.05), there was a significant Condition x Periods interaction, F(4, 220) = 3.86, p <.01, Cw 2 =.01. Separate ANOVAs of the effect of periods conducted for each of the two response conditions (Keppel, 1991) indicated that the vigilance decrement was statistically significant in both conditions, Fstandard(3, 108) = 10.37, p <.001, Cw 2 =.10; Fmodifced(3, 100) = 2.96, p <.05, 0) 2 =.01. It is evident in Figure 2, however, that because the scores for the two vigilance groups were almost identical in the first l i 90g * Standard Modified o 80 0z _ Il I I I I I I Periods of Watch (10 nmi) Figure 2. Percentages of correct detections for the standard-vigil and modified-vigil groups as a function of period of watch (error bars reflect standard errors).

6 FaIl Fall Human Human Factors~~~~~ period of watch but were lower in the standard condition than in the modified condition by the final period, the drop in signal detections over time was more pronounced in the standard than in the modified vigil condition. In the ANOVAs for correct detections and all subsequent analyses involving repeated measures, Box's epsilon was employed when appropriate to correct for violations of the sphericity assumption (Maxwell & Delaney, 1990). Mean percentages of false alarms for the standard and modified conditions are plotted as a function of periods of watch in Figure 3. A 2 (conditions) x 5 (periods) ANOVA of an arcsin transformation of the false alarm scores revealed that false alarms were significantly more frequent in the modified vigil condition (M= 15.8) than in the standard condition (M= 3.0), F(1, 62) = 9.58, p <.001, t 2 =.12. All other sources of variation in the analysis were not significant (p >.05). Workload Mean global workload scores on the NASA- TLX for the standard vigil and modified vigil are displayed in Figure 4. Also shown is the mean workload score on the card-sort task. Under the standard scoring procedures for the NASA-TLX (Hart & Staveland, 1988), global workload values can range from 0 to 100. It is clear in the figure that the means for the standard-vigil and the modified-vigil conditions were almost identical. Both means fell at the upper end of the scale and were considerably higher than the mean for the simple card-sort task, which was located at the lower end of the workload scale. The standard-vigil and modifiedvigil conditions did not differ significantly in their global workload ratings, t(30) < 1. Stress The 10 DSSQ scale scores were standardized against normative data secured from a large British sample (Matthews et al., 1999), using the formula (raw score - norm group factor mean)/ norm group factor standard deviation. Factor scores for worry, engagement, and distress were calculated using regression weights from the normative sample. Factor scores are distributed with a mean of 0 and SD of 1, so that values calculated for a sample represent a deviation from normative values in standard deviation units. Preliminary inspection of the previgil data for the two vigilance conditions and the card-sort condition revealed that the factor scores in all three conditions were close to the normative 20 I- i- _ec U W. c; K 5 I I Periods of Watch (10 min) i Figure 3. Percentages of false alarms for the standard-vigil and modified-vigil groups as a function of period of watch (error bars reflect standard errors).

7 VIGILANCE AND MINDLESSNESS 355 _9 --- ou I 70 l- 60H 50H I I.0 ie 3 40K 30 H 20 ~- T 10O Card sort Task o Standard Modified Condition Figure 4. Mean global workload scores on the NASA-TLX for the standard-vigil and modified-vigil groups (error bars reflect standard errors). Scores can range from 0 to 100. mean of 0. Scores did not differ significantly from each other or from the norm group on any of the DSSQ factors. Posttask (vigil and card sort) standard scores based on the DSSQ norms for each of the three factors were also determined for each participant. Mean standard scores for the card sort and the two vigilance conditions are displayed in Figure 5. A 3 (conditions) x 3 (factors) ANOVA of the postvigil data revealed significant effects for conditions, F(2, 45) = 5.04, p <.02, W 2 =.03, and factors, F(1, 67) = 16.14, p <.001, o2 =.13, and a significant Conditions x Factors interaction, F(3, 67) = 15.06, p <.00 1, w 2 =.25. To explore the interaction more fully, separate ANOVAs were computed on the differences between conditions within each DSSQ dimension. Significant condition differences were found within each dimension, Fworry( 2, 45) = 8.77, p <.01, 2 =.24, FDjstess(2, 45) = 14.79, p <.001, o 2 =.36; FEngagement( 2, 45) = 11.77, p <.001, W2 =.31. Supplementary Scheffe tests using an alpha of.05 indicated that although the means for the two vigilance conditions (standard and modified) did not differ significantly from each other on any DSSQ dimension, the means for both vigilance conditions were significantly greater than the mean for the cardsort task on the worry and distress factors and significantly lower than the card-sort mean on the task engagement factor. To compare the posttest DSSQ scores of the card-sort condition and the two vigilance conditions against the DSSQ norm group mean, t tests against the hypothesis of no difference (a standard score of 0) were carried out on the mean of the card-sort group on each DSSQ factor and the combined mean of the two vigilance conditions on each factor. The Bonferroni correction was employed to ensure that the alpha level was.05 for the six tests employed. The mean of the card-sort condition was significantly below that of the norm group on the

8 356 Fall Human Factors 356 FaIl Human Factors U) 0~0 u " T -2- Worry Distress Engagement DSSQ Factors Figure 5. Postvigil standard scores on the DSSQ worry, distress, and engagement factors for the card-sort group and the two vigilance groups (error bars reflect standard errors). worry dimension but did not differ significantly from the DSSQ norms on the remaining dimensions. In contrast, the combined mean of the two vigilance conditions did not differ significantly from the DSSQ norm on the worry dimension, but it was significantly greater than the norm on the distress dimension and significantly below the norm on task engagement. Microanalysis Although the postvigil scores for the two vigilance groups on the worry dimension did not differ significantly from each other or from those of the norm group, one of the scales constituting that factor, Task-Related Cognitive Interference, has special implications for the mindlessness hypothesis proposed by Robertson and his associates (Manly et al., 1999; Robertson et al., 1997). That scale measures respondents' thoughts and cognitions regarding the task and the adequacy of their performance. If the modified condition indeed promotes mindlessness, observers in that condition should be expected to show fewer task-related thoughts than those in the standard condition. Consequently, data from the Task-Related Cognitive Interference scale of the worry factor were selected for additional analysis. Previgil and postvigil standard scores for Task-Related Cognitive Interference were determined for each observer in the two vigilance conditions using the norm group mean and standard deviation for that subscale. Mean previgil and postvigil z scores for the standard vigil (standard errors are in parentheses) were (.19) and.48 (.23), respectively. Comparative values for the modified vigil were -.05 (.28) and.64 (.22), respectively. A 2 (conditions) x 2 (phase) ANOVA of the data for the Task- Related Cognitive Interference scale indicated that, overall, self-reports of task-relevant concern increased significantly over the vigil, F(1, 30) = 44.47, p <.001, W 2 =.14. The absence of a significant main effect for conditions and of a significant Conditions x Phase interaction (p >.05 in each case) indicated, however, that the previgil and postvigil scores on the Task-Related Cognitive Interference scale were similar in the standard and modified conditions. Tests of significance using an alpha level of.05 and the Bonferroni correction indicated that the previgil mean for the combined vigilance conditions (M = -. 20) did not differ significantly from a z score of 0, t(31) = -1.19, but that the postvigil combined mean (M =.56) was significantly greater than 0, t(3 1) = Thus, whereas the

9 VIGILANCE AND MINDLESSNESS 'A57r level of Task-Related Cognitive Interference for observers in the two vigilance groups was no different from that of the DSSQ norm group prior to the vigil, it was significantly greater than the norm group's after the vigil. DISCUSSION The results of this study clearly do not support the notion proposed by Robertson and his associates (Manly et al., 1999; Robertson et al., 1997) that vigilance tasks in general and the modified vigil in particular promote a mindless lack of attentional focus. As in prior studies (Hitchcock et al., 1999; Temple et al., 2000; Warm et al., 1996), vigilance performance in this investigation induced a high level of workload, and that workload was similar in the standard and modified conditions. In addition, monitors in this experiment, as in others using the DSSQ (Helton et al., 2000; Matthews et al., 1999; Szalma et al., in press; Temple et al., 2000), reported the vigilance task to be stressful. Monitors in both vigilance conditions were significantly more worried postvigil than were the card-sort participants and also showed significantly less task engagement and greater feelings of distress. Moreover, monitors in the two vigilance conditions also showed significantly less task engagement and significantly more distress than did the DSSQ norm group postvigil. Finally, interfering cognitions regarding performance adequacy increased similarly over the course of the testing session in the two vigilance conditions. The high workload and stress scores in the modified vigil are hardly compatible with the view that this procedure induces mindlessness. Those results are compatible, however, with the view that regardless of the nature of the indicator response to be made, vigilance tasks in general, and visual vigilance tasks in particular, can be quite taxing. Although similar to the standard-vigilance condition in workload and stress, the modifiedvigilance condition did differ from the standard condition in performance efficiency: The modified condition had a smaller vigilance decrement and a much higher false alarm rate than did the standard condition. A possible basis for the high false alarm rate in the modified condition is muscle fatigue. The frequent pressing of the response key required in this condition could have induced muscle tiredness that led observers to hold back the required motor response to nonsignal events. That possibility seems unlikely, however, in that postexperimental interviews did not elicit complaints of tiredness, muscle pain, or finger discomfort from monitors in the modified vigil. Essentially, monitors in the modified vigil were confronted with a situation in which they were required to make a motor response to a frequently occurring imperative stimulus - the appearance on their VDT of a display indicating an acceptable part - and to make "the response of not responding" when the imperative stimulus occasionally changed such that it indicated an unacceptable part. This situation is reminiscent of the "stop-signal paradigm" often used in studies of motor control wherein operators are required to make a specified motor response to a frequently occurring imperative stimulus and to withhold that response when an infrequent countermanding command is issued to stop ongoing action (Sanders, 1998). As described by Logan and Cowan (1984), competition between antagonistic action modules (excitation/inhibition) in this paradigm can lead to inappropriate responses. In the present case, such competition could have led monitors in the modified vigil to have difficulty in initiating the motor response signifying an acceptable part, thus elevating their false alarm rate. Viewed in this way, the false alarm results fit with the workload and stress findings in indicating that the dynamics of the modified vigilance task transcend the routinized mindlessness envisioned by Robertson and his associates (Manly et al., 1999; Robertson et al., 1997). Had the findings of this study supported the view that failures of vigilance performance reflect mindless behavior on the part of monitors, there would be little to offer designers and managers of operational systems in the way of remedy, short of full automation (with its attendant limitations; Parasuraman & Riley, 1997) or imposing drastic punitive action on those performing vigilance tasks. However, the finding that our monitors expended considerable attentional resources in both types of the vigilance task suggests that procedures might be

10 358 Fall Human Factors developed to enhance system efficiency and to lower the workload and stress experienced by monitors in operational settings. Toward those ends, several leads have appeared in the vigilance literature. These include enhancing the signal-to-noise ratio, the use of auditory displays instead of visual displays when possible, the provision of feedback regarding performance efficiency and reliable warnings of impending untoward events, reducing postural constraint and eyestrain in visual vigilance tasks, and increasing the monitors' perception of control over stimulus events (Craig, 1984; Hitchcock et al., 1999, 2003; Scerbo, 1998; Szalma et al., in press; Warm et al., 1996). ACKNOWLEDGEMENTS This work was supported by the Department of Defense Multidisciplinary University Research Initiative (MURI) program administered by the Army Research Office under grant DAAD , P. A. Hancock, Principal Investigator. The views expressed in this work are those of the authors and do not necessarily reflect official Army policy. REFERENCES Broadbent, D. E., Cooper, P., Fitzgerald, P., & Parkes, K. (1982). The cognitive failures questionnaire (CFQ) and its correlates. British Journal of Clinical Psychology, 21, Craig, A. (1984). Human engineering: The control of vigilance. In J. S. Warm (Ed.), Sustained attention in human performance (pp ). Chichester, UK: Wiley. Dickman, S. J. (2002). Dimensions of arousal: Wakefulness and vigor. Human Factors, 44, Grubb, P. L., Miller, L. C., Nelson, W. T., Warm, J. S., Dember, W. N., & Davies, D. R. (1994). Cognitive failure and perceived workload in vigilance performance. In M. Mouloua & R. Parasuraman (Eds.), Human performance in automated systems: Current research and trends (pp ). Hillsdale, NJ: Erlbaum. Hancock, P. A., & Hart, S. G. (2002). Defeating terrorism: What can human factors/ergonomics offer? Ergonomics in Design, 10(1), Hancock, P. A., & Warm, J. S. (1989). A dynamic model of stress and sustained attention. Human Factors, 31, Hart, S. G., & Staveland, L. E. (1988). Development of the NASA- TLX (Task Load/Index): Results of empirical and theoretical research. In P. A. Hancock & N. Meshkati (Eds.), Human mental workload (pp ). Amsterdam: North-Holland. Helton, W. S., Dember, W. N., Warm, J. S., & Matthews, G. (2000). Optimism, pessimism, and false failure feedback: Effects on vigilance performance. Current Psychology, 18, Hitchcock, E. M., Dember, W. N., Warm, J. S., Moroney, B. W., & See, J. E. (1999). Effects of cueing and knowledge of results on workload and boredom in sustained attention. Human Factors, 41, Hitchcock, E. M., Warm, J. S., Matthews, G., Dember, W. N., Shear, P. K., Tripp, L. D, & Mayleben, D. W. (2003). Automation cueing modulates cerebral blood flow and vigilance in a simulated air-traffic control task. Theoretical Issues in Ergonomic Science, 4, Keppel, G. (1991). Design and analysis: A researcher's handbook (3rd ed.). Upper Saddle River, NJ: Prentice-Hall. LaBerge, D. (1995). Attentional processing: The brain's art of mindfulness. Cambridge, MA: Harvard University Press. Langer, E. J. (1989). Mindfulness. Reading, MA: Addison-Wesley. Logan, G. D., & Cowan, W. B. (1984). On the ability to inhibit thought and action: A theory of an act of control. Psychological Review, 91, Manly, T., Robertson, I. H., Galloway, M., & Hawkins, K. (1999). The absent mind: Further investigations of sustained attention to response. Neuropsychologica, 37, Matthews, G. (2001). Levels of transaction: A cognitive science framework for operator stress. In P. A. Hancock & P. A. Desmond (Eds.), Stress, workload, and fatigue (pp. 5-33) Mahwah, NJ: Erlbaum. Matthews, G., Campbell, S. E., Falconer, S., Joyner, L., Huggins, J., Gilliland, K., et al. (2002). Fundamental dimensions of subjective state in performance settings: Task engagement, distress and worry. Emotion, 2, Matthews, G., Davies, D. R., Westerman, S. J., & Stammers, R. (2000). Human performance: Cognition, stress and individual differences. Hove, UK: Psychology. Matthews, G., Joyner, L., Gilliland, K., Campbell, S., Falconer, S., & Huggins, J. (1999). Validation of a comprehensive stress state questionnaire: Towards a state "big three"? In 1. Mervielde, I. J. Deary, F. DeFruyt, & F Ostendorf (Eds.), Personality psychology in Europe (Vol. 7, pp ). Tilburg, Netherlands: Tilburg University Press. Maxwell, S. E., & Delaney, H. D. (1990). Designing experiments and analyzing data: A model comparison perspective. Belmont, CA: Wadsworth. Nachreiner, F., & Hanecke, K. (1992). Vigilance. In A. P. Smith & D. M. Jones (Eds.), Handbook of human performance: Vol. 3. State and trait (pp ). London: Academic. Parasuraman, R., & Riley, V. A. (1997). Humans and automation; use, misuse, disuse, abuse. Human Factors, 39, Parasuraman, R., Warm, J. S., & Dember, W. N. (1987). Vigilance: Taxonomy and utility. In L. S. Mark, J. S. Warm, & R. L. Huston (Eds.), Ergonomics and human factors (pp ). New York: Springer-Verlag. Robertson, I. H., Manly, T., Andrade, J., Baddeley, B. T., & Ytend, J. (1997). "Oops!": Performance correlates of everyday attentional failures in traumatic brain injured and normal subjects. Neuropsychologica, 35, Sanders, A. F. (1998). Elements of human performance: Reaction processes and attention in human skill. Mahwah, NJ: Erlbaum. Sarason, 1. G., Sarason, B. R., & Pierce, G. R. (1990). Anxiety, cognitive interference, and performance. Journal of Social Behavior and Personality, 5, Scerbo, M. W. (1998). What's so boring about vigilance? In R. R. Hoffman, M. F. Sherrick, & J. S. Warm, (Eds.), Viewing psychology as a whole: The integrative science of William N. Dember (pp ). Washington, DC: American Psychological Association. Shallice, T. (1988). From neuropsychology to mental structure. Cambridge, UK: Cambridge University Press. Stuss, D. T., Shallice, T., Alexander, M. P., & Picton, T. W. (1995). A multidisciplinary approach to anterior attentional functions. Annals of the New York Academy of Sciences, 769, Szalma, J. L., Warm, J. S., Matthews, G., Dember, W. N., Weiler, E. M., Meier, A., et al. (in press). Effects of sensory modality and task duration on performance, workload, and stress in sustained attention. Human Factors. Temple, J. G., Warm, J. S., Dember, W. N., Jones, K. S., LaGrange, C. M., & Matthews, G. (2000). The effects of signal salience and caffeine on performance, workload, and stress in an abbreviated vigilance task. Human Factors, 42, Warm, J. S. (1993). Vigilance and target detection. In B. M. Huey & C. D. Wickens (Eds.), Workload transition: Implications for individual and team performance (pp ). Washington, DC: National Academy. Warm, J. S., Dember, W. N., & Hancock, P. A. (1996). Vigilance and workload in automated systems. In R. Parasuraman & M. Mouloua (Eds.), Automation and human performance: Theory and applications (pp ). Mahwah, NJ: Erlbaum.

11 VIGILANCE AND MINDLESSNESS li;rl Wickens, C. D., & Hollands, J. G. (2000). Engineering psychology and human performance (3rd ed.). Upper Saddle River, NJ: Prentice-Hall. Rebecca A. Grier is an engineering psychologist in the Human Systems Integration Group at the Naval Air Warfare Center, Dahlgren Division, Dahlgren, Virginia. She received her M.S. in experimental psychology/human factors in 2000 at the University of Cincinnati, where she is pursuing her Ph.D. in experimental psychology/human factors. Joel S. Warm is a professor of psychology and director of graduate studies in psychology at the University of Cincinnati. He received his Ph.D. in experimental psychology from the University of Alabama in William N. Dember is dean emeritus and professor emeritus at the University of Cincinnati. He received his Ph.D. in experimental psychology from the University of Michigan in Gerald Matthews is a professor of psychology at the University of Cincinnati. He received his Ph.D. in experimental psychology from the University of Cambridge in Traci L. Galinsky is an adjunct assistant professor of psychology at the University of Cincinnati and a research psychologist in the Organizational Science and Human Factors Branch of the National Institute for Occupational Safety and Health in Cincinnati. She received her Ph.D. in experimental psychology/ human factors from the University of Cincinnati in James L. Szalma is a postdoctoral fellow in the Department of Psychology at the University of Central Florida. He received his Ph.D. in experimental psychology/human factors from the University of Cincinnati in Raja Parasuraman is professor of psychology and director of the Cognitive Science Laboratory at the Catholic University of America, Washington, D.C. He received his Ph.D. in psychology from the University of Aston in Date received: August 29, 2001 Date accepted: February 27, 2003

12 COPYRIGHT INFORMATION TITLE: The Vigilance Decrement Reflects Limitations in Effortful Attention, Not Mindlessness SOURCE: Hum Factors 45 no3 Fall 2003 WN: The magazine publisher is the copyright holder of this article and it is reproduced with permission. Further reproduction of this article in violation of the copyright is prohibited. Copyright The H.W. Wilson Company. All rights reserved.