Abstract
Ergonomics, which MeSH classifies under applied psychology, is the study of how work, tools, and environments should be fitted to the capacities and limits of the human operator. It grew from wartime studies of pilots who made lawful, predictable errors with badly designed cockpits, and it treats such errors as properties of the system rather than the person. Its cognitive branch models the operator as an information processor with measurable throughput: aimed movement obeys Fitts’s law, choice reaction time obeys Hick’s law, and sustained attention decays as the vigilance decrement. Modern human factors extends these laws to mental workload, situation awareness, and human–automation interaction, where the designer’s task is allocating function between a fallible human and an imperfect machine.
Keywords: human factors, ergonomics, mental workload, situation awareness, reaction time
Ergonomics, also called human factors, asks a deceptively simple question: given what is known about human perception, cognition, and action, how should a task be arranged so that a person can perform it quickly, accurately, and without undue strain? The field began with the recognition that skilled, motivated operators still failed at rates that tracked the design of their equipment rather than their training. When several models of aircraft produced the same confusion between the flap and landing-gear levers, the sensible conclusion was not that pilots were careless but that identical adjacent controls invite identical adjacent errors (Fitts, 1954).
The cognitive core of ergonomics treats the operator as an information-processing system with quantifiable properties. Movement time to a target, the time to choose among alternatives, the capacity of working attention, and the decay of vigilance over a long watch are all lawful and measurable. Those regularities let a designer predict performance before a single user is tested, and they turn interface design from taste into engineering. This article follows that thread from the classic timing laws through mental workload and vigilance to situation awareness and the modern problem of sharing control with automation.
Key Takeaways
- Ergonomics fits the task to the operator, treating error as a designed-in property of the system rather than a personal failing.
- Fitts’s law makes aimed-movement time a logarithmic function of distance and target width; Hick’s law makes choice time logarithmic in the number of options.
- Mental workload is finite and multidimensional; performance breaks down when demand exceeds the operator’s resources, measured by tools such as the NASA-TLX.
- Sustained monitoring degrades over time — the vigilance decrement — so critical signals should not depend on unaided human watchkeeping.
- Situation awareness and sound function allocation, not raw automation, determine whether a human–machine system is safe.
What Ergonomics Is
Ergonomics is the scientific discipline concerned with the interactions among humans and the other elements of a system, and the application of that knowledge to make systems compatible with the needs and limitations of people. The word joins the Greek ergon (work) and nomos (law): the laws of work. In American usage the field is more often called human factors, and the two terms are interchangeable. Its defining commitment is that the human is the fixed point. A rivet, a menu, or a cockpit can be redesigned at will; the reach envelope of an arm, the resolution of the fovea, and the span of working memory cannot. Good design therefore begins from those constants and arranges everything else around them.
This inverts the older engineering assumption that a properly trained operator could adapt to any well-built machine. The wartime aviation studies showed that assumption to be false in a specific, useful way: the errors that skilled pilots made were not random lapses but systematic consequences of the equipment. Controls that looked and felt alike were confused; displays that required mental rotation were misread under stress; warnings that competed for the same channel were missed. Because the errors were lawful, they were predictable, and because they were predictable, they were designable-away. That insight — that human error is largely a property of the interface — remains the field’s founding idea and its most consequential contribution to safety (Reason, 2000).
The operator as an information-processing system
Types of Ergonomics
Ergonomics sits in the MeSH tree beneath applied psychology on the psychological side and beneath engineering on the technical side — a polyhierarchy that reflects its dual heritage. Directly beneath the descriptor, MeSH files four narrower headings, each naming a distinct sub-problem of fitting work to the operator. They are largely orthogonal: a single workstation study can touch how information is displayed, how the human and machine divide the work, how the task itself is structured, and how effort is distributed across time, all at once. MeSH is an indexing vocabulary for retrieving literature, not a theory of the mind, so these categories organize the published record rather than carving cognition at its joints.
| Subtype | In brief |
|---|---|
| Data Display | How information is presented visually so that a reader can extract it accurately and fast — the coding, layout, and formatting of instruments, screens, and printed output. |
| Man-Machine Systems | The coupled human–machine unit treated as one system, including how function is allocated between operator and device and how control passes between them. |
| Task Performance and Analysis | The decomposition of a task into its component steps and demands, and the measurement of how well and how quickly they are carried out. |
| Time Management | The scheduling and prioritizing of activity so that limited time and attention are spent where they matter most, at the level of both the task and the working day. |
Performance Models
The oldest and best-tested results in ergonomics are the timing laws that describe how long a motor or perceptual act takes. Their power is that they are quantitative: given a target’s size and distance, or the number of choices an operator faces, they return a predicted time in milliseconds that survives replication across decades and devices.
Fitts’s law governs rapid aimed movement. The time to move a limb or a cursor to a target of width W at distance D rises with the ratio of the two, but only logarithmically: doubling the distance adds a fixed increment rather than doubling the time. Formally, movement time is MT = a + b · log₂(2D/W), where a and b are empirical constants for the device and limb, and the logarithmic term is the index of difficulty in bits (Fitts, 1954). The law explains why a large button near the pointer is reached far faster than a small one far away, and it is the reason screen edges and corners — effectively infinitely wide targets — are prime real estate in interface design.
Fitts’s Law: pointing time
Movement time rises with the ratio of distance to width, but only logarithmically.
Predicted movement time MT = 150 + 100 × ID = 550 ms
Doubling the width removes one bit of difficulty and subtracts a fixed 100 ms, no matter how far away the target is — which is why enlarging small, frequent targets buys the most speed.
Hick’s law describes the other half of the timing story: the time to choose among alternatives rather than to move to one. Reaction time rises logarithmically with the number of equally likely options n, as RT = a + b · log₂(n + 1) (Hick, 1952). The information-theoretic reading is that the operator is transmitting bits, and each doubling of the choice set adds one bit and one fixed time increment. Hyman confirmed that it is the information in the stimulus set, not the raw count of stimuli, that sets the time, by varying probabilities and sequential dependencies and finding reaction time tracked entropy in every case (Hyman, 1953). Together the two laws let a designer estimate how long a menu selection or a targeting action will take before it is built.
Hick’s Law: choice time
The time to choose rises with the information in the choice set, one fixed step per doubling.
Predicted reaction time RT = 200 + 150 × log₂(n + 1) = 548 ms
Because the cost is logarithmic, one broad menu of many items is often faster than a deep chain of small ones: each extra doubling of choices adds only a single fixed increment.
Mental Workload
Timing laws assume the operator has resources to spare. Mental workload is the study of what happens when they do not. The central claim is that attention and processing capacity are finite, and that performance holds up only while the demands of the task stay within that budget; beyond it, errors climb and speed collapses. Workload is not a single quantity, however. Wickens’s multiple resource theory holds that the operator draws on several partly independent pools — visual versus auditory perception, spatial versus verbal coding, manual versus vocal response — so that two tasks interfere far more when they compete for the same pool than when they draw on different ones (Wickens, 2008). The practical corollary is that a driver can follow a spoken instruction while steering more easily than a written one, because speech and the visual road scene tax different resources (Wickens, 2002).
Because workload is multidimensional, measuring it means sampling several dimensions at once. The NASA Task Load Index does exactly that, combining an operator’s ratings of mental demand, physical demand, temporal demand, performance, effort, and frustration into a weighted workload score (Hart & Staveland, 1988). It remains the field’s standard subjective instrument. On the objective side, task-analytic models such as the keystroke-level model predict how long a skilled user will take on a routine computer task by summing the times of its elementary operators — keystrokes, pointing, homing, and mental preparation — and so estimate workload and time from a task description alone (Card, Moran, & Newell, 1980).
Vigilance and Error
Sustained attention is the operator’s scarcest resource over long periods. When a person must watch for infrequent, unpredictable signals — a fault on a display, a blip on a radar screen — detection performance falls measurably within the first half hour on watch. Mackworth demonstrated this vigilance decrement with a clock whose hand occasionally made a double jump that observers had to report; the proportion caught dropped sharply as the watch wore on, and the decline was steeper the rarer the signal (Mackworth, 1948). The finding is a direct warning against designs that rest safety on a human staring at a rarely-changing display, which is precisely the situation that highly reliable automation creates for its human monitor.
The Vigilance Decrement
Detection of rare signals falls within the first half hour of a sustained watch.
The decline is steepest early and rarer signals fall faster, so safety-critical monitoring should not rest on unaided human watchkeeping over long, uneventful periods.
When monitoring fails and an accident follows, the ergonomic analysis looks past the operator to the system. Reason’s distinction between active failures — the operator’s unsafe act at the sharp end — and latent conditions — the design, staffing, and procedural weaknesses lying dormant in the system — reframes most disasters as the alignment of several latent holes rather than a single culprit’s blunder (Reason, 2000). The corresponding design goal is not to demand more care from operators but to remove the latent conditions and to build in the defenses that catch an active failure before it propagates.
Situation Awareness and Automation
Knowing what the machine is doing is as critical as operating it well. Situation awareness names that knowledge, and Endsley’s three-level model decomposes it into perceiving the elements in the environment, comprehending their meaning, and projecting their near-future state (Endsley, 1995). Accidents in automated systems frequently trace to a breakdown at one of these levels: the operator perceives the raw data but fails to comprehend that the autopilot has quietly changed mode, or comprehends the present state but cannot project where an automated maneuver will lead.
Automation does not remove the human from the loop so much as change the human’s job from controller to supervisor, and that shift has costs. Parasuraman, Sheridan, and Wickens set out a framework for how much to automate, distinguishing the stage of processing being automated — acquisition, analysis, decision, or action — from the level of autonomy at each stage, and warning that high automation of decision-making erodes the operator’s awareness and skill (Parasuraman, Sheridan, & Wickens, 2000). One design response is to make the machine’s internal state visible: ecological interface design maps the deep structure of the controlled process onto the display so that the constraints the operator must respect are directly perceptible rather than inferred (Vicente & Rasmussen, 1992).
Worked Example
Consider a touchscreen control on which an operator must tap a rectangular button. In the first layout the button is 12 mm wide and its center sits 96 mm from the resting position of the thumb. Fitts’s law describes the movement in two steps: first the index of difficulty, then the predicted time.
The index of difficulty is ID = log₂(2D/W) = log₂(2 × 96 / 12) = log₂(16) = 4.0 bits. Using device constants of a = 150 ms and b = 100 ms/bit — values typical of thumb pointing on a handheld screen — the predicted movement time is MT = a + b · ID = 150 + 100 × 4.0 = 550 ms.
Now widen the button to 24 mm while keeping the distance fixed. The index of difficulty falls to log₂(2 × 96 / 24) = log₂(8) = 3.0 bits, and the predicted time drops to 150 + 100 × 3.0 = 450 ms. Doubling the target width did not halve the time; it removed exactly one bit of difficulty and so subtracted one b-worth of time, 100 ms, or an 18 percent saving. The logarithmic form is why the largest gains in pointing speed come from enlarging the smallest, most-used targets rather than from shortening travel, and why an edge target — where W is effectively unbounded — is reached fastest of all.
Discussion
The enduring lesson of ergonomics is that human performance is lawful enough to be engineered. The timing laws give quantitative predictions; the workload and vigilance literatures set limits on what an operator can sustain; the situation-awareness and automation frameworks describe how those limits play out when control is shared with a machine. Across all of them runs the same commitment: the human’s capacities are the design constants, and the system is the variable to be adjusted.
That commitment has a moral as well as a technical edge. Treating error as a property of the interface rather than the person moves the response to failure away from blame and toward redesign, which is both fairer and more effective, because a fired operator is replaced by another equally susceptible to the same trap while a redesigned control protects everyone who ever uses it (Reason, 2000). The field’s influence is now largely invisible precisely because it succeeded: shape-coded controls, forcing functions that make dangerous actions impossible, and displays laid out to match the task are so ordinary that their absence is what now looks like a design flaw.
Current Directions
The active frontier of human factors is the fully autonomous system, where the operator’s role shrinks to rare, high-stakes intervention. Endsley’s review of the road from automation to autonomy argues that increasing autonomy does not monotonically reduce workload; it can instead produce an out-of-the-loop operator who is slow to detect and correct the automation’s failures, so that the human factors of autonomy are harder, not easier, than those of manual control (Endsley, 2017). Hancock presses the same point for self-driving vehicles, cautioning that the hand-off from machine to unprepared human in an emergency is exactly the moment human capability is weakest, and that promises of effortless autonomy understate this transfer-of-control problem (Hancock, 2019).
A second front applies macroergonomics to whole sociotechnical systems rather than single workstations. The SEIPS family of models treats patient safety as an emergent property of a work system — its people, tasks, tools, physical environment, and organization together — and its most recent form follows the patient’s own journey across settings as the unit of analysis, extending the field’s founding insight from the cockpit to the hospital (Carayon et al., 2020). Both fronts return to the same question the field opened with: how should function be divided between a fallible human and an imperfect machine.
Common Misconceptions
- Ergonomics is just about chairs and posture.
- Physical ergonomics — seating, reach, posture — is one branch. Cognitive ergonomics, the subject of this article, concerns perception, memory, attention, and decision-making, and drives interface, display, and automation design (Wickens, 2008).
- More automation always reduces workload and error.
- Automating a task shifts the operator from controller to monitor, which introduces vigilance decrements, skill loss, and out-of-the-loop delays. Poorly allocated automation can raise total system risk rather than lower it (Parasuraman, Sheridan, & Wickens, 2000).
- Human error is the root cause of most accidents.
- The operator’s act is usually the last link in a chain of latent design and organizational conditions. Ergonomics treats such error as a symptom of the system, not the explanation for the failure (Reason, 2000).
- Fitts’s law means shorter movements are always faster.
- Distance matters only logarithmically, while target width matters just as much. Enlarging a small target often saves more time than shortening the movement to it (Fitts, 1954).
Glossary
- Ecological interface design.
- A display philosophy that maps the constraints and deep structure of a controlled process directly onto the interface so operators perceive them rather than infer them.
- Engineering psychology.
- The branch of psychology, named by Fitts, that applies experimental findings on human perception, cognition, and action to the design of equipment and systems.
- Fitts’s law.
- The regularity that the time to move to a target is a linear function of its index of difficulty, log₂(2D/W), where D is distance and W is width.
- Function allocation.
- The design decision about which functions in a system are assigned to the human operator and which to the machine.
- Hick’s law.
- The regularity that choice reaction time rises logarithmically with the number of equally likely alternatives, log₂(n + 1).
- Human factors.
- The North American synonym for ergonomics: the discipline of fitting tasks, tools, and environments to human capabilities and limits.
- Index of difficulty.
- The logarithmic term in Fitts’s law, measured in bits, that combines target distance and width into a single measure of a movement’s demand.
- Latent condition.
- A dormant weakness in a system’s design, staffing, or procedures that lies inactive until combined with an active failure to produce an accident.
- Macroergonomics.
- The organizational level of ergonomics that treats safety and performance as emergent properties of a whole work system rather than of a single workstation.
- Mental workload.
- The demand a task places on an operator’s finite attentional and processing resources relative to the capacity available.
- Multiple resource theory.
- Wickens’s account that attention comprises several partly independent pools, so tasks interfere most when they compete for the same pool.
- NASA-TLX.
- The NASA Task Load Index, a multidimensional subjective rating scale that combines six demand dimensions into an overall workload score.
- Situation awareness.
- An operator’s perception of the elements in the environment, comprehension of their meaning, and projection of their future state.
- Vigilance decrement.
- The decline in signal-detection performance that occurs within the first half hour of a sustained monitoring task, steeper for rarer signals.
Key Researchers
Pascale Carayon
(living). Human factors engineer who developed the SEIPS family of models applying macroergonomics to healthcare and patient safety; Professor Emerita, University of Wisconsin–Madison. ORCID · Wikipedia · Wikidata
Alphonse Chapanis
(1917–2002). American psychologist regarded as a father of human factors engineering; his wartime cockpit studies established shape-coded controls to prevent the confusions then blamed on pilot error. Wikipedia · Wikidata
Mica R. Endsley
(living). Engineer who developed the three-level theory of situation awareness and the SAGAT measurement technique; founder of SA Technologies and former Chief Scientist of the U.S. Air Force. ORCID · Wikipedia · Wikidata
Paul M. Fitts
(1912–1965). American psychologist who founded engineering psychology; originator of Fitts’s law relating movement time to target distance and width, and of the Fitts List for allocating functions between humans and machines. Wikipedia · Wikidata
Peter A. Hancock
(born 1953). Human factors researcher known for work on stress, mental workload, vigilance, time perception, and trust in human–automation interaction; Provost Distinguished Research Professor, University of Central Florida. ORCID · Wikipedia · Wikidata
Donald A. Norman
(born 1935). Cognitive scientist and usability engineer; author of The Design of Everyday Things and popularizer of user-centered design, affordances, and the concept of the designed-in error. Wikipedia · Wikidata
Christopher D. Wickens
(living). Engineering psychologist who originated multiple resource theory of attention and workload and the SEEV model of attention allocation; Professor Emeritus, University of Illinois at Urbana-Champaign. ORCID
Frequently Asked Questions
What is the difference between ergonomics and human factors?
There is none of substance. Ergonomics is the term of choice in Europe and much of the world, human factors in North America, and professional bodies increasingly pair them as human factors and ergonomics. Both name the discipline of fitting work to the human.
Is ergonomics part of psychology or engineering?
Both. MeSH files it under applied psychology and under engineering at once, reflecting a field that applies psychological knowledge of perception, cognition, and action to the design of engineered systems. Cognitive ergonomics leans psychological; physical and organizational ergonomics lean toward engineering and management.
What does Fitts’s law actually predict?
It predicts how long a rapid, aimed movement to a target will take, as a logarithmic function of the target’s distance and width. A wider or nearer target is reached faster, and because the relationship is logarithmic, enlarging a small target usually buys more speed than shortening the travel to it.
Why does automation sometimes make systems less safe?
Automating a task turns the operator into a monitor, and human monitoring degrades over time through the vigilance decrement and skill loss. If the automation then fails, an out-of-the-loop operator is slow to notice and correct it, so poorly allocated automation can raise overall risk.
What is mental workload and how is it measured?
Mental workload is the demand a task places on an operator’s limited attentional resources. It is assessed subjectively with multidimensional scales such as the NASA-TLX, objectively through performance on a concurrent secondary task, and analytically with models that sum the times of a task’s component operations.
What is situation awareness?
Situation awareness is an operator’s up-to-date model of what is happening: perceiving the relevant elements, understanding what they mean together, and projecting how the situation will develop. Many automation accidents are failures of situation awareness rather than of manual skill.
Does ergonomics only apply to physical comfort?
No. Physical ergonomics addresses posture, reach, and force, but cognitive ergonomics addresses perception, memory, attention, and decision-making, and macroergonomics addresses whole organizations. The cognitive branch drives the design of displays, controls, and automated systems.
Why does ergonomics treat error as a system property?
Because skilled, motivated operators make errors that track the design of their equipment rather than their competence. Redesigning the interface protects every future user, whereas blaming and replacing an operator leaves the same trap in place for the next one.
References
Card, S. K., Moran, T. P., & Newell, A. (1980). The keystroke-level model for user performance time with interactive systems. Communications of the ACM, 23(7), 396–410. https://doi.org/10.1145/358886.358895
Carayon, P., Wooldridge, A., Hoonakker, P., Hundt, A. S., & Kelly, M. M. (2020). SEIPS 3.0: Human-centered design of the patient journey for patient safety. Applied Ergonomics, 84, 103033. https://doi.org/10.1016/j.apergo.2019.103033
Endsley, M. R. (1995). Toward a theory of situation awareness in dynamic systems. Human Factors, 37(1), 32–64. https://doi.org/10.1518/001872095779049543
Endsley, M. R. (2017). From here to autonomy: Lessons learned from human-automation research. Human Factors, 59(1), 5–27. https://doi.org/10.1177/0018720816681350
Fitts, P. M. (1954). The information capacity of the human motor system in controlling the amplitude of movement. Journal of Experimental Psychology, 47(6), 381–391. https://doi.org/10.1037/h0055392
Hancock, P. A. (2019). Some pitfalls in the promises of automated and autonomous vehicles. Ergonomics, 62(4), 479–495. https://doi.org/10.1080/00140139.2018.1498136
Hart, S. G., & Staveland, L. E. (1988). Development of NASA-TLX (Task Load Index): Results of empirical and theoretical research. Advances in Psychology, 52, 139–183. https://doi.org/10.1016/S0166-4115(08)62386-9
Hick, W. E. (1952). On the rate of gain of information. Quarterly Journal of Experimental Psychology, 4(1), 11–26. https://doi.org/10.1080/17470215208416600
Hyman, R. (1953). Stimulus information as a determinant of reaction time. Journal of Experimental Psychology, 45(3), 188–196. https://doi.org/10.1037/h0056940
Mackworth, N. H. (1948). The breakdown of vigilance during prolonged visual search. Quarterly Journal of Experimental Psychology, 1(1), 6–21. https://doi.org/10.1080/17470214808416738
Parasuraman, R., Sheridan, T. B., & Wickens, C. D. (2000). A model for types and levels of human interaction with automation. IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans, 30(3), 286–297. https://doi.org/10.1109/3468.844354
Reason, J. (2000). Human error: Models and management. BMJ, 320(7237), 768–770. https://doi.org/10.1136/bmj.320.7237.768
Vicente, K. J., & Rasmussen, J. (1992). Ecological interface design: Theoretical foundations. IEEE Transactions on Systems, Man, and Cybernetics, 22(4), 589–606. https://doi.org/10.1109/21.156574
Wickens, C. D. (2002). Multiple resources and performance prediction. Theoretical Issues in Ergonomics Science, 3(2), 159–177. https://doi.org/10.1080/14639220210123806
Wickens, C. D. (2008). Multiple resources and mental workload. Human Factors, 50(3), 449–455. https://doi.org/10.1518/001872008X288394