Vigilance and Sustained Attention

Vigilance and Sustained Attention: What Is Vigilance and What Accounts for Its Decay?

Vigilance, or sustained attention, is a phenomenon that plays an essential role in our daily lives. The requirement for attention is ubiquitous—in school, work, relationships, and just about every aspect of life. Naturally, sustained attention—vigilance—has become a topic of extreme interest to many researchers. In today’s post, we examine the work of three research groups on vigilance and the factors that influence it. We begin with a summary of the key points of each group’s work, with a particular focus on the main experimental questions posed, key results obtained, interpretations of those results, and inferences made. After this review of results, we perform a comparative analysis of these studies to determine whether the researchers’ views are compatible. We are interested in the similarities and points of departures in the three groups’ results and interpretations. A deeper analysis will follow, in which we examine whether the researchers’ inferences are justified; here, we will raise questions concerning the experimental design and interpretation of results as well as point out possible alternative interpretations. Lastly, partially because I find this topic intriguing and partially because I couldn’t really resist as a neuroscience junkie, we will briefly explore the links between the psychological study of vigilance and neuroscience before concluding with some remarks on possible future directions.

Overview/Summary of Data and Results

 Jerison and Pickett

In their 1964 paper on “Vigilance: The Importance of the Elicited Observing Rate,” Jerison and Pickett took a decision-theory based approach to answering the question “what causes vigilance to degrade, as measured by the decreased ability to detect rare signals in a ‘prolonged vigil’” (1964). Specifically, the authors argue that “a failure of attention” is to blame for this phenomenon; furthermore, they attribute this failure of attention to the “observer’s decision about whether or not to observe.” As the authors note in the paper, this approach implies that the demand on observing behavior is more important than the rate of the signals presented. To test their hypothesis, Jerison and Pickett tested subjects under two conditions: in one, subjects were asked to detect signals against a series of slowly-occurring (5 times/minute) background events in a time period of 80 minutes. In the other, the subject were asked to do the same thing against a series of faster-occurring background events (30 times/minute). In direct contrast to the arousal hypothesis, they found that subjects in the first group (with the low background event rate) only missed about 10% of the signals, whereas the subjects in the second group (with high background event rate) missed about 70% of the signals. Since variables such as memory and time were controlled for in this study, Jerison and Pickett inferred from their results that “the observer’s decision about whether or not to observe is associated with the average payoff or ‘expected value’ of attending to the regularly recurring events.” Specifically, the authors inferred that the observer choose to behave most economically, and thus are more likely to attend and have higher vigilance when the probability of detecting the signals is high (during low background event rate), and less likely to attend when the probability is low, such as during high background event rate.

Broadbent and Gregory

Broadbent and Gregory’s 1965 paper investigated the effects of noise and signal rate on vigilance, a phenomenon that was analyzed through the lens of decision theory. More specifically, they address the question whether changes in vigilance are due to changes in the evidence the operator uses to make decisions or changes in the operator’s criteria in making and reporting such decisions. To approach this question, Broadbent and Gregory considered two variables: d’, which they defined as a measurement of the strength of the signal (and thus independent of the operator’s criterion), and beta, which is defined as a measure of the operator’s “degree of caution” in doubtful situations. The authors tested subjects by asking them to report signal observation with a “yes,” “no,” or “maybe.” The consideration of doubt in their study is relatively novel and departs from previous studies such as that of Jerison and Picket. Interestingly, contrary to previous findings by Mackworth and Taylor, the authors of this paper found that there was no decrease in d’ over time, though a rise in beta was noted. Thus, subjects became more wary of reporting signal observation as time went on. Furthermore, they found high levels of noise increased the subjects’ confidence while low levels of noise was correlated with lower confidence. The authors inferred from this result that the operator’s criteria for a signal become more strenuous as a function of time. Furthermore, the authors also inferred from their results that at low signal rates, the operator is less willing to report uncertain signals, whereas the opposite phenomenon is observed at high signal rates. Thus, the authors attribute a possible decrease in vigilance to a reduction in the sensitivity of the observer as a function of time during the vigilance task.

Scerbo

In his 1998 paper on “What’s So Boring About Vigilance,” Scerbo performed a series of studies in which he attempted to induce boredom via utilizing the vigilance boredom. These studies represented the first endeavor at studying boredom experimentally and were performed with the following objectives: to elucidate the nature of boredom and to determine the role of task-associated factors and personality in the perception of boredom. We will provide a brief summary of the key points and findings of the four experiments that Scerbo performed to this end.

            Scerbo’s initial experiment was a test of O’Hanlon (1981)’s theory of boredom, which held that boredom is a psychophysiological state that occurs when the optimal arousal level necessary for a given task is unable to be maintained. Specifically, he argues that boredom is created by the stress associated with “fighting habituation.” Thus, it follows that more demanding tasks should yield higher levels of boredom. To test this theory, Scerbo had subjects observe a display for white lines that increased in conspicuity, as increases in conspicuity have been positively correlated to signal detection. Boredom was assessed with a variation of Barmack’s method. Interestingly, although perpetual sensitivity was higher for high conspicuity signals, boredom ratings were similar for both low and high conspicuity conditions, a result that is inconsistent with O’Hanlon’s original theory. Additionally, Scerbo also found that the onset of boredom occurs very quickly in vigilance tasks, usually peaking after 10 minutes.

            Scerbo’s second study investigated the validity of using movement level as a proxy for the subject’s degree of boredom. In accordance with results of previous studies, Scerbo hypothesized that movement was directly proportional to a subject’s degree of boredom. Thus, the more bored a subject is, the more movement he/she should exhibit during the vigil. In this study, movement was assessed by the extent to which subjects moved a mouse cursor during the vigil. Results were in accordance with his hypothesis; he found subjects who moved the mouse the most also reported the greatest levels of boredom. Furthermore, he also found subjects who reported the greatest boredom also had the highest “mental workload” in vigilance tasks and thus had the worst performance. Scerbo made several inferences from the results of this study; specifically, he inferred that subjects who find vigilance “particularly boring” may not be good candidates for sustained attention tasks; furthermore, he suggests that the high mental workload experienced during vigilance may be associated with the boring nature of the task.

            Scerbo’s third study examined the effects of stimulus variety and satiation on boredom. Subjects were presented with two tasks: one conventional vigilance task and a kaleidoscope task, in which various “exciting” stimuli (kaleidoscope-like screen saver) were presented and subjects asked to press a key when the pattern was composed of predominately white lines. In accordance with their hypothesis, Scerbo et. al. found subjects reported lower degrees of boredom for the kaleidoscope task than for the vigilance task. As for satiation, Scerbo reasoned that giving subjects the option to quit during a vigilance task should significantly decrease the level of reported boredom, as subjects can quit when the task becomes overly boring. Interestingly, however, he found mixed results; while most subjects quit before the experimental duration, most subjects still reported feelings of boredom

            In Scerbo’s fourth and final experiment, he investigated the effect of personality factors in the perception of boredom. In short, he found that subjects had varying levels of boredom proneness (BP), and individuals with higher BP reported greater levels of boredom after a vigilance task. Furthermore, he also found that subjects with high BP were sensitive to instruction manipulation whereas subjects with low BP were largely unaffected. From the collective results of these studies, Scerbo inferred that vigilance “can indeed be quite boring.” Furthermore, subjects can feel boredom very strongly and minor changes in task-related parameters may be insufficient to relieve boredom. Additionally, individual-specific factors, such as boredom proneness vary among subjects and should be taken into consideration in vigilance tasks. Importantly, individuals with great BP would most likely make for poor test subjects. Scerbo concludes with his belief that it may be important to continue studying the source of boredom in vigilance. 

Comparative Analyses of Studies

The authors for today were all interested in the paradigm of vigilance and performed studies in attempts to elucidate factors that impact vigilance and contribute to the degradation of vigilance over time. Although there exist several similarities in their views, there are also multiple important distinctions. Here, we comparatively examine the authors’ views and ultimately suggest that are only compatible to a certain degree. For instance, Jerison and Pickett considered only one variable in their study of vigilance, and as they noted in their paper, background event rate was credited with being the major determinant in the decay of vigilance. Broadbent and Gregory performed a similar study on vigilance, and although they took a decision theory-based approach to their study (similar to Jerison and Pickett), they went one step further and took into account the existence of doubt in the subjects’ reports. Their conclusion, namely that the effects seen during a low-signal rate vigilance task are in part due to a “reduced willingness to report uncertain signals” (1965), is compatible with Jerison and Pickett’s findings, but provides an alternative explanation for the latter’s results. Thus, while the two research groups come to a similar conclusion, their interpretations of that conclusion are different. Scerbo’s series of studies on boredom in vigilance contributes a further viewpoint to the discussion on the decay of vigilance over time. For instance, the results of his experiments indicate that boredom can be overwhelming for subjects and can significantly affect their performance in vigilance tasks. He goes on further to suggest that individuals with particularly high boredom proneness would make for poor test subjects in vigilance studies, as they are almost intrinsically primed to do poorly. Are Scerbo’s viewpoints incompatible with those of Jerison and Pickett and Broadbent and Gregory? Certainly not. But they’re not completely compatible either. Scerbo raises an important question, namely the role of boredom in vigilance and its source, that the two previous research groups did not consider. One can think of the overarching question of what accounts for the decay of vigilance as a mosaic, and each author’s work as specific pieces of that mosaic. By itself, no one author or research group provides the complete view. Together, however, they raise interesting questions and challenge each other to help construct a better view of the mosaic.

A Deeper Analysis: Questions and Comments

Now that we have reviewed the authors’ key findings and discussed the compatibility of their viewpoints, it is worthwhile to examine their inferences in detail and determine whether they’re justified. Furthermore, we will also point out alternative interpretations of results and possible confounding factors. Let’s begin with Jerison and Pickett. Although their experiments were properly controlled for the most part and their results indicative of their inferences, they failed to include an adequate sample size for their vigilance tasks. Since their study only had 20 subjects, the trends seen in their results are less than compelling. One can question whether their results are representative of the actual phenomenon. What if there were a few subjects with particularly high boredom proneness (a factor that was not even considered in this study)? While the gender and memory of the subjects were controlled, their personalities were not, and this could skew the results, especially with such a low sample size. Furthermore, aside from these factors, the results obtained by Jerison and Pickett could be accounted for by alternative explanations in addition to the background signal rate theory. For instance, aspects such as the degree of fatigue of individuals prior to commencing the vigilance task and their tendency to become bored would be important factors to consider.

Broadbent and Gregory improved upon Jerison and Pickett’s study on vigilance by considering the element of doubt during each reported decision. Although I felt their results were more comprehensive than those of Jerison and Pickett, I still found some of there inferences partially unjustified. For instance, their conclusion that a reduced willingness to report signals during doubtful situations in low-signal conditions definitely has merit and is supported by the results, but is also not the only possible explanation of the data. Fatigue is a possible physiological explanation. It is possible that a third variable such as fatigue was contributing to the decrease in confidence and thus the decrease in vigilance. This may also be compatible with their finding on the relationship between noise and confidence. Audio stimuli such as noise may bring the individual to a higher level of arousal, which may veil some of the effects of fatigue. Consequently, the subjects’ fatigue levels may be decreased in the presence of noise, which would correlate with an increased confidence.

I found Scerbo’s studies of boredom in vigilance extremely fascinating as one of the first studies to examine boredom experimentally and investigate the presence and role of boredom in vigilance. However, I was bothered by the fact that Scerbo never really provides a definition for boredom and what constitutes boredom. He rejects O’Hanlon’s definition of boredom, but does not provide a discrete definition himself. Although he provides several ways of measuring boredom (Barmack’s method, etc.), he does not provide a way to normalize reports of boredom. As he himself notes in his report, individuals have different boredom proneness; thus subjects may have different perceptions of what constitutes boredom. Without a normalization scheme, the reports used in his study remain rather subjective. The reliance of his inferences were based on such subjective reports may introduce an element of doubt into the level of justification of his inferences.

The largely qualitative nature of the research groups’ experiments is also a possible area of concern. Clearly, it is extremely difficult to quantify something nebulous such as boredom and thus boredom assays and similar assays must rely heavily on subject responses. However, these responses may not be normalized, and one person’s idea of “extreme boredom” or an “8/9” may be very different from that of a different subject. Furthermore, all three studies employ visual tasks in the assessment of vigilance. However, sustained attention is required in several non-visual tasks. For instance, is the vigilance required in an oral interview a different sort of vigilance? If the subjects were subjected to auditory tasks in the assessment of vigilance, would the results be the same?

Links to Neuroscience and Possible Future Directions

I confess that I’m obsessed with neuroscience. So I was really interested in the biological and chemical differences in brains of a subject at rest, under sustained attention, and during boredom. I think it would be interesting to monitor which areas of the brain are activated during vigilance tasks in real time; more specifically, I would be interested to see the differences in activation patterns between the beginning and middle of the vigilance trail. A rudimentary experimental design could be as follows: subjects could be placed in an fMRI machine and shown the visual stimuli associated with the vigilance task. They can then be asked to verbalize a response every time a signal is detected. This experiment may shed light on the neurobiological mechanisms underlying sustained attention as well as a biological explanation for the decay of vigilance. Once we have a better idea of this, we could expand to a more systematic/network level by investigating the pathways and neural networks that contribute to vigilance and sustained attention.

Thanks for reading! And I’m really sorry that it’s too long—I got a little carried away.

 References 

Broadbent, D.E. and Gregory, M. (1965). Effects of Noise and Signal Rate Upon Vigilance Analysed by Means of Decision Theory. Human Factors, 7, 155-162.

Jerison, H.J. and Pickett, R.M. (1964). Vigilance: The Importance of the Elicited Obseving Rate. Science, 143, 970-971.

Scerbo, M. (1998). What’s So Boring About Vigilance? In R.B. Hoffman, M.F. Sherrick, & J.S. Warm (Eds.), Viewing Psychology As A Whole: The Integrative Science of William N. Dember (pp. 145-166). Washington, DC: APA. 

Image References:

Question Marks. http://www.servitokss.com/question-marks/. 31/1/2013/

Study Shows Effects of Combat on Brain. http://vetlawyers.com/vetblog/index.php/2011/02/study-shows-effects-of-combat-on-brain/. 31/1/2013. 

Study Habits Cartoons and Comics. CSL Cartoonstock. http://www.cartoonstock.com/directory/s/study_habits.asp. 31/1/2013.