State of science: mental workload in ergonomics

In the paper

“Research in the 1990s was also concerned with the proliferation of automation, and a significant body of work was directed at the emergence of more advanced physiological metrics of workload.”

In the paper

“Nevertheless, the relative decade-on-decade increase in hits for MWL in the 2000s suggests a more recent prominence over the comparator terms - 36% for 'mental workload' against 17% for 'physical workload' and 'workload'.”

In the paper

“A cursory review of these search results indicates that the focus in the 1980s was much more on the measurement of MWL, while the 1990s saw a shift towards theoretical developments and modelling efforts.”

In the paper

“A cursory review of these search results indicates that the focus in the 1980s was much more on the measurement of MWL, while the 1990s saw a shift towards theoretical developments and modelling efforts.”

In the paper

“Finally, the first decade of the twenty-first century has seen more examples of MWL applications coming to the fore.”

In the paper

“Of these, driving stands out as a particular focus area especially in the last decade, although we should also note the rapid growth of MWL research in the rail industry, reflecting the resurgence in ergonomics and human factors interest within this domain (e.g. Wilson and Norris 2005).”

What the sources say

• Not checked yetcites Wilson and Norris 2005

In the paper

“Another theme emerging from this review surrounds the impacts of skill (or experience) and age on MWL, which can, more generally, be related to the still unsolved puzzle of individual differences (see Damos 1988; Szalma 2009).”

What the sources say

• Not checked yetcites Damos 1988
• Not checked yetcites Szalma 2009

In the paper

“The predominance of transport applications among MWL research over the last two decades cannot be ignored.”

In the paper

“The other key theme in the 1980s was on adaptive interfaces.”

In the paper

“The theoretical underload redline does exist and is illustrated on Figure 2, but identifying or quantifying it remains elusive.”

In the paper

“We would certainly not wish to claim that progress in measurement or definition has ended; MWL looks to be just as nebulous a concept today as it did three decades ago, and researchers continue to debate over definitions to this day.”

In the paper

“But we know this is not the case; underload is just as bad for performance as overload, and leads to the classic inverted-U curve as illustrated in the figures.”

In the paper

“The field has developed from tackling the thorny issue of defining the concept, through the development of metrics of MWL, to the more recent focus on applied research, which, in turn, reflects the technological and societal concerns of our times.”

In the paper

“This is particularly the case as technology progresses in areas such as transportation, which has already dominated the MWL scene in recent years.”

In the paper

“While this notion offers a neat hypothetical explanation of underload and overload, it still does not let us quantify or predict these redlines in an applied setting.”

In the paper

“Indeed, in measurement, new neuroergonomics techniques are emerging as strong contenders for finally ascertaining the physiological correlates of mental demand.”

In the paper

“Research in this field has so far failed to provide absolute limits on human performance due to the complex nature of the subject and of the processes involved.”

In the paper

“Earlier in the paper, we reviewed research from the 1980s promising workload-based adaptive automation. Some 25 years later, that promise is yet to be fulfilled.”

In the paper

“Mental workload (MWL) is one of the most widely invoked concepts in ergonomics research and practice (Flemisch and Onken 2002; Loft et al. 2007; Parasuraman and Hancock 2001; Tsang and Vidulich 2006; Wickens 2008).”

What the sources say

• Not checked yetcites Flemisch and Onken 2002
• Not checked yetcites Loft et al. 2007
• Not checked yetcites Parasuraman and Hancock 2001

In the paper

“It is easier to state what underload is not: it is not vigilance (since the consensus has been for some time that vigilant monitoring is actually a highly demanding task; Warm, Dember, and Hancock 1996) nor is it boredom (usually it is about doing very little rather than doing nothing at all).”

What the sources say

• Not checked yetcites Warm, Dember, and Hancock 1996

In the paper

“Indeed, the current opinion is that there is an optimum range of MWL which is associated with best performance (Hancock and Warm 1989; see also Figure 1).”

What the sources say

• Not checked yetcites Hancock and Warm 1989

In the paper

“Automatic processing is associated with expert performance and is characteristically fast, unconscious and almost completely liberated from attentional resource constraints.”

What the sources say

• Not checked yetcites Schneider and Shiffrin 1977

In the paper

“If demands begin to exceed capacity, skilled operators can either adjust their strategy to compensate or else performance necessarily degrades.”

In the paper

“Nevertheless, it is probably fair to say that the jury is out as to whether underload is distinct from other influences on performance such as automation-related complacency and supervisory control.”

In the paper

“There may also be different kinds of resources available, as in other team members, or technological support to cope with demand.”

In the paper

“Therefore, there is no direct guarantee that simply reducing MWL improves performance, and in fact the opposite may be true.”

In the paper

“Fixed resource models of workload are not without their drawbacks, though. For instance, absolute capacity demands are illusive; they do not directly consider nonattentional factors, such as experience, or more slowly changing variation in attentional resources due to learning or the reduction of capacity with age, or even the willingness to invest effort.”

In the paper

“Although numerous definitions have been offered, it is obvious that there is no universal agreement between these disparate statements.”

In the paper

“It is easier to state what underload is not: it is not vigilance (since the consensus has been for some time that vigilant monitoring is actually a highly demanding task; Warm, Dember, and Hancock 1996) nor is it boredom (usually it is about doing very little rather than doing nothing at all).”

In the paper

“Demands (stress) can have multiple facets, such as time pressure and task complexity.”

In the paper

“The converse is controlled processing, and in the practical world these two elements lie on a continuum.”

In the paper

“Suboptimal workload can mean either overload or underload (Brookhuis and de Waard 2000).”

What the sources say

• Not checked yetcites Brookhuis and de Waard 2000

In the paper

“MWLhasthus become a topic of increasing importance as modern technology imposes ever greater cognitive demands.”

In the paper

“Consequently, if effort is invested, then MWL will be increased (cf. Shaw et al. 2013) - which may be a positive adaptation in circumstances of underload, but could be further detrimental when faced with overload.”

In the paper

“This is consistent with the attentional resources approach, as automatic processing releases attentional resources for other tasks, with a resulting decrease in MWL.”

In the paper

“Thus, MWL as a multidimensional construct, is determined by characteristics of the task (e.g. demands, performance), of the operator (e.g. skill, attention) and, to a degree, the environmental context in which the performance occurs.”

In the paper

“In the present authors' view, there has to be some engagement in the task, but such engagement is exceptionally low.”

In the paper

“Underload itself, while an intuitive concept, is in many ways a more nebulous term than MWL, as different researchers take it to mean different things.”

In the paper

“From this view, MWL in real-world tasks is determined by the balance of automatic and controlled processing involved.”

In the paper

“That said, there is a strong consensus that mental underload can be just as detrimental to performance as mental overload, with both leading to performance degradation, attentional lapses and errors (Wilson and Rajan 1995).”

What the sources say

• Not checked yetcites Wilson and Rajan 1995

In the paper

“There are, however, commonalities among the various interpretations, which do help to shed light here.”

In the paper

“Although the stress/strain comparison has been criticised for being too simplistic (e.g. Bainbridge 1974), even the international standard on MWL (ISO 10075, 2000) is heavily dependent on the analogy for its terminology.”

What the sources say

• Not checked yetcites ISO 10075, 2000
• Not checked yetcites Bainbridge 1974

In the paper

“MWL is a peculiar concept that has intuitive appeal, but remains surprisingly difficult to define (see also situation awareness; Smith and Hancock 1995).”

What the sources say

• Not checked yetcites Smith and Hancock 1995

In the paper

“Finally, the trade-off between stress and strain may have different effects on the human - as measured by the different objective and subjective metrics which we describe later (see, e.g., Bevan and Macleod 1994).”

What the sources say

• Not checked yetcites Bevan and Macleod 1994

In the paper

“Resources, in this arena of discussion, often refer to attentional resources (e.g. Wickens 1980, 2002); thus, MWL becomes a product of the resources available to meet task demands (Welford 1978).”

What the sources say

• Not checked yetcites Welford 1978
• Not checked yetcites Wickens 1980, 2002

In the paper

“An analogy is often made between mental and physical load, in that each expresses two components - stress (i.e. task demands) and strain (impact on the human; cf. Schlegel 1993).”

What the sources say

• Not checked yetcites Schlegel 1993

In the paper

“Thus, performance can be maintained at the cost of individual strain or vice versa.”

In the paper

“Overload occurs, for instance, when the operator is faced with more stimuli than (s)he is able to handle while maintaining their own standards of performance.”

In the paper

“As workload increases, the response time to, and the chance of missing, a signal increases.”

In the paper

“Visual perception is crucial for drivers while concurrent execution of another in-vehicle visual task (for instance, looking at a cell phone) competes for visual attention with the primary driving task.”

In the paper

“MWL can increase heart rate and decrease heart rate variability at the same time.”

In the paper

“Physiological measures are a natural type of workload index since work demands physiological activity by definition.”

In the paper

“So far, these methods have been restricted to laboratory conditions, and the equipment is extremely expensive to purchase and use.”

In the paper

“Recently, Posner (2012) illustrated that new methods to study human operators at work are gaining momentum, advancing our understanding of brain plasticity.”

What the sources say

• Not checked yetcites Posner 2012

In the paper

“Other measures of interest are the activity of the brain (Brookhuis and de Waard 1993; de Waard and Brookhuis 1991) and of certain facial muscles, such as the corrugator supercilii (Hoogendoorn et al. 2010; Jessurun, Steyvers, and Brookhuis 1993).”

What the sources say

• Not checked yetcites Brookhuis and de Waard 1993
• Not checked yetcites de Waard and Brookhuis 1991
• Not checked yetcites Hoogendoorn et al. 2010

In the paper

“Closely coupled to heart rate and also sensitive to workload is blood pressure (Rau 2004).”

What the sources say

• Not checked yetcites Rau 2004

In the paper

“Suboptimal workload may also emerge from disruptions in the operator's physiology, for instance when under stress (Hancock and Warm 1989), although other non-workload stressors (such as from the environment or sleep loss) can influence these indices.”

What the sources say

• Not checked yetcites Hancock and Warm 1989

In the paper

“In the complex, safety-critical systems where MWL research is usually most pertinent and/or most pursued (such as transport, as we shall see later), both underload and overload are very real concerns.”

In the paper

“Operator conditions in fixed positions such as operating motor vehicles, trains and airplanes are suitable for this next generation of MWL field studies.”

In the paper

“While numerous physiological measures are now relatively easily measured in the operational environment, attention is less easily monitored in ambulant situations - but not impossible.”

In the paper

“MWLis a subjective state as well; people are able to express themselves in words or indications on scales in post-task responses (Zijlstra 1993).”

What the sources say

• Not checked yetcites Zijlstra 1993

In the paper

“Beta activity (1230 Hz) is predominant when the participant in the study is generally awake and alert, while the activity dropping to Alpha activity (8-12 Hz) indicates developing drowsiness, and going further down into the theta region (5-8 Hz) may even lead to falling asleep.”

In the paper

“As we have noted, the multidimensional nature of MWL is reflected in the variety of workload metrics available (see also, e.g. Gawron 2008).”

What the sources say

• Not checked yetcites Gawron 2008

In the paper

“But while both low and high MWL are undoubtedly basic precursors to errors, an exact relationship between MWL and accident causation is not easily established, let alone measured in practice.”

In the paper

“Modern integrated, wireless measurement facilities enable more easily accessible EEG measurement (e.g. Makeig et al. 2002), while more integrated approaches such as neuroergonomics, suggested by Parasuraman (2011), open new windows to the study of MWL.”

What the sources say

• Not checked yetcites Parasuraman 2011

In the paper

“The PDT is based on the premise that visual attention narrows as workload increases.”

In the paper

“Driving on an urban road with traffic, traffic signs and traffic lights leads to increased workload according to these measures (high heart rate, low variability), which immediately reverses when the driver stops if one of the lights turns red.”

What the sources say

• Not checked yetcites De Waard 1996

In the paper

“Modern integrated, wireless measurement facilities enable more easily accessible EEG measurement (e.g. Makeig et al. 2002), while more integrated approaches such as neuroergonomics, suggested by Parasuraman (2011), open new windows to the study of MWL.”

What the sources say

• Not checked yetcites Makeig et al. 2002
• Not checked yetcites Parasuraman 2011

In the paper

“Brookhuis and de Waard (2000) discriminated between underload and overload by referring to error sources, the former leading to reduced alertness and lowered attention, the latter to distraction, diverted attention and insufficient time for adequate information processing.”

What the sources say

• Not checked yetcites Brookhuis and de Waard 2000

In the paper

“In fact, recent research has explored the interaction between physical and MWL across some of these metrics, espousing the idea that light physical activity could actually compensate for the deficits imposed by mental underload (Basahel, Young, and Ajovalasit 2010).”

What the sources say

• Not checked yetcites Basahel, Young, and Ajovalasit 2010

In the paper

“Both physical workload and MWL are well known to have a clear impact on heart rate, heart rate variability and respiration (Mulder 1992), and delineating the effects of one over the other can prove troublesome in applied research.”

What the sources say

• Not checked yetcites Mulder 1992

In the paper

“In the central region, workload gradually increases while performance is at its best, remaining relatively constant.”

In the paper

“Although the apparent concern is more with overload than underload, it is clear that issues of optimising workload and defining workload 'redlines' or thresholds have occupied researchers since some of the earliest work in this field.”

In the paper

“Thus, higher task demand leading to increased MWL does not have to affect performance, and not all measures have to be strongly correlated.”

What the sources say

• Not checked yetcites de Waard 1996

In the paper

“Also, Hayashi (1988; Hayashi and Kosugo 1987) examined the MWL of software engineers in programming tasks.”

What the sources say

• Not checked yetcites Hayashi 1988
• Not checked yetcites Hayashi and Kosugo 1987

In the paper

“If one measure reflects an increase in workload and another measure does not, then measures are said to dissociate (see also Hancock 1996; Yeh and Wickens 1988).”

What the sources say

• Not checked yetcites Hancock 1996
• Not checked yetcites Yeh and Wickens 1988

In the paper

“All of this chimes with recent suggestions that mind and body cannot be separated in ergonomics research (cf. Marras and Hancock 2014).”

What the sources say

• Not checked yetcites Marras and Hancock 2014

In the paper

“Similarly, Nowakowski (1987) was also concerned with online MWL assessment for a knowledge-based adaptive system, with a particular focus on individual differences between users.”

What the sources say

• Not checked yetcites Nowakowski 1987

In the paper

“For instance, glass cockpits in commercial aircraft have relieved workload in some areas, such as reduced display clutter and more automated flight procedures. However, the same systems have increased workload in other areas, such as more decision options in any given situation (Hilburn 1997), and confusion with respect to operating modes (Ferris, Wickens, and Sarter 2010; Sarter and Woods 1995).”

What the sources say

• Not checked yetcites Hilburn 1997
• Not checked yetcites Ferris, Wickens, and Sarter 2010
• Not checked yetcites Sarter and Woods 1995

In the paper

“The problem with some automation is that it reduces workload when demand is low but problematically increases workload when demand is high. This tendency is evident when the automation is 'dumb and dutiful' as Wiener (1989) cast it.”

What the sources say

• Not checked yetcites Wiener 1989

In the paper

“Dissociation has usually been reported between self-reports and measures of performance, and sometimes between physiological and self-report measures (Myrtek et al. 1994).”

What the sources say

• Not checked yetcites Myrtek et al. 1994

In the paper

“Notwithstanding the progress reported here, the measurement of mental work is founded on the science of psychology and comparatively speaking, psychology is an immature, if no less difficult, science.”

In the paper

“Count variables can be used to predict absolute workload values, both above and below the redline, but multitask interference models cannot easily do so at the current stage of their maturity.”

In the paper

“In any resource-limited system, the most relevant measure of demand is specified relative to the supply of available resources.”

In the paper

“These metrics suggest a resolution of the physical versus MWL question, and perhaps even augur the potential to quantify workload and those redlines of performance.”

In the paper

“It has been suggested (Perrey, Thedon, and Rupp 2010; Shaw et al. 2013) that metrics of brain oxygenation could essentially represent a quantitative measure of attentional resources.”

What the sources say

• Not checked yetcites Perrey, Thedon, and Rupp 2010
• Not checked yetcites Shaw et al. 2013

In the paper

“For example we have noted the redline for working memory at roughly seven chunks of information.”

In the paper

“However, alongside these evident scientific concerns, the need to assess mental work is also very much fuelled by the ever-present practical necessity to measure mental activity and allocate tasks in the modern electronic workplace.”

In the paper

“Alternatively, low demands could be misperceived by the operator, leading to a mismatch in terms of effort invested to perform the task (Matthews and Desmond 1997).”

What the sources say

• Not checked yetcites Matthews and Desmond 1997

In the paper

“Many of the measures are also differentially sensitive in the different regions.”

In the paper

“Young and Stanton's research into vehicle automation led them to propose a theory whereby attentional resources shrink in situations of low workload, thus leading to problems when demands suddenly increase (see Young and Stanton 2002).”

What the sources say

• Not checked yetcites Young and Stanton 2002

In the paper

“The redlines of both underload and overload coincide with the points where the demand and resource lines cross over; performance degrades in both regions of underload and overload when demands exceed resources.”

In the paper

“As we have already seen earlier in this paper, effort is voluntary and is related to resource investment (cf. Shaw et al. 2013), so this could be compatible with Young and Stanton's (2002) idea.”

What the sources say

• Not checked yetcites Shaw et al. 2013

In the paper

“Nevertheless, it is reasonable to suggest that the applied problems will assume even greater prominence as we move forward.”

In the paper

“Advances in mobile and wearable computing, augmented reality and similar technologies no doubt present great challenges for MWL research in the near future.”

In the paper

“This is reflected (albeit with some artistic licence) in Figure 2, as resource supply (based on activation and/or effort) at first lags behind task demands in the underload region, then exceeds it until resources start to asymptote towards the upper capacity limit, and we reach overload.”

In the paper

“These advances also reflect and further emphasise the role of mental work over physical work in the modern workplace.”

In the paper

“Here, very low workload is often coupled with fatigue and automationrelated complacency. Then the sudden unexpected failure rapidly throws the operator into a highly stressful fault diagnosis and failure management mode, well above the redline, in such a way that the prior loss of situation awareness has rendered them unprepared.”