A nurse sees a patient’s oxygen reading, hears a change in breathing, and knows that the next few minutes may require a response. A number on a screen is only part of what the nurse needs to know. Situation awareness concerns how a person or team notices relevant features of a changing environment, understands what they mean for a goal, and anticipates what may happen next. It is especially useful for explaining action in settings where information changes while people are working: care, transport, emergency response, and complex digital operations. The concept is not a synonym for alertness, intelligence, or successful performance. It has to be defined against the situation and decision at hand.
Perception, comprehension, and projection
Endsley’s influential account distinguishes three levels: perception of relevant elements, comprehension of their meaning, and projection of their likely near future. The levels help diagnose different failures. An operator may never see a warning, may see it but misunderstand its relation to other data, or may understand the present while failing to anticipate a rapid change. These are analytically different problems and suggest different design responses. The three-level framework is a model of situation awareness, not proof that the brain completes three separate serial stages in every case. Perception, interpretation, and anticipation can interact continuously.
The same display can support one level while undermining another. A dashboard may make every sensor value visible, improving access to current measurements, yet scatter related values so widely that the user must remember and integrate them. A trend view may help project where a variable is heading but hide a short-lived anomaly. Designers should therefore ask what a user needs to decide, not merely how much information the interface can show. More visible data do not automatically create better awareness. The meaning of a cue depends on the task, expected trajectory, uncertainty, and the possibility of acting in time.
Situation awareness is relational. A pilot needs different information during approach than while taxiing; an emergency responder needs different information when locating a person than when planning a safe exit. A person may know many true facts and still lack the specific information necessary for the next decision. Conversely, a person may not know every available detail and still have adequate awareness for the immediate goal. A research study should define which facts and relationships were relevant before testing awareness. Otherwise a score can reward trivia while missing the essential state of the system.
Endsley’s model also distinguishes awareness from decision and action. Someone can correctly understand a situation yet choose a risky action because of competing goals or constraints. Someone can act successfully by following a routine without a detailed mental account of every variable. Performance is important but cannot serve as the sole measure of awareness. A failure may arise from limited information, misinterpretation, a poor policy, equipment limits, or teamwork. Calling all failures “loss of situation awareness” explains too little and may hide the design conditions that made a decision hard.
How researchers study awareness
The Situation Awareness Global Assessment Technique, or SAGAT, uses pauses in a simulation to ask participants about relevant aspects of the current situation. Endsley’s measurement paper describes the logic: probes are compared with the actual state of the simulated system. A well-designed probe can test whether someone noticed a key fact, understood a relation, or anticipated a development. This is more direct than inferring awareness solely from task outcome. But the freeze interrupts activity, and answers depend on the chosen questions and the simulated world. Probe content must be validated against the decisions that matter.
Other measures include post-task self-ratings, observer judgments, eye tracking, communication analysis, and performance indicators. They answer different questions. A self-rating tells us how aware someone believed they were, but confidence can diverge from accuracy. Eye tracking shows where a person looked, not what they understood. An observer can see behavior but not every internal inference. A recent meta-review of situation-awareness measurement found many tools and limited psychometric evidence for most, with probing methods such as SAGAT having comparatively stronger validity evidence and practical costs. Method choice should follow the question rather than convenience.
A probe is only as useful as its reference state. In a simple simulation, researchers may know exactly which aircraft is approaching a runway or which monitor alarm has changed. In real work, the relevant state may be uncertain or distributed across people and systems. A question such as “What is happening?” may have several defensible answers depending on role. This is why studies need domain experts, explicit task analysis, and transparent scoring rules. They should also report how participants were briefed and whether the interruption itself changed subsequent decisions.
Situation awareness changes over time. A single freeze can reveal what a participant knows at that moment, but may miss how quickly they update after a change. A design may help detect a new cue yet slow integration with the broader picture. Repeated probes or carefully selected event transitions can test updating, although repeated interruptions can make the simulation artificial. Analysts can complement probes with logs of communication, information seeking, and decisions. The aim is to reconstruct a process, not declare that a person possesses a permanent amount of awareness.
Individuals, teams, and distributed systems
Many tasks are performed by teams. A nurse may recognize a trend, a physician may interpret its clinical significance, and a technician may know the equipment status. No single person has to hold every fact for the team to act well, provided important information reaches the right person at the right time. Stanton’s distributed situation-awareness perspective treats awareness as emerging across people and artifacts, including displays, procedures, and communication channels. This challenges an exclusively individual account. It also raises a measurement problem: a team can hold complementary knowledge rather than a single shared mental picture.
Shared awareness and complementary awareness should not be confused. Team members may all need to know the location of an emergency exit, while only one specialist needs a detailed equipment parameter. Too much duplication can waste attention; too little overlap can make handoffs brittle. A useful team analysis maps who needs which information, when, and for what decision. Communication quality is not simply message volume. A short, well-timed statement that identifies a change and its implication can do more than a stream of undifferentiated updates. Team awareness is therefore linked to coordination, not identical with it.
Tools become part of that coordination. A whiteboard, shared timeline, or patient chart can preserve information across a shift and make it available to people who were not present when an event occurred. Yet an artifact can also become stale or ambiguous. If one person assumes a display is current while another knows it has not updated, the team may act on a false common picture. Design should make time stamps, uncertainty, and responsibility visible. Such recommendations are Institute synthesis grounded in human-factors principles; they require testing in the setting where people actually use the tool.
Automation changes what humans can perceive. A system may filter hundreds of data points and display only a prediction. This can help a user focus, but it may hide the evidence behind the prediction or the conditions under which it fails. If the system unexpectedly hands control back, the person may need to rebuild a picture under time pressure. Situation-awareness research suggests evaluating whether the human can explain current system state, likely next state, and available options at transition points. A high automation accuracy rate during routine cases does not settle the transfer-of-control problem.
Boundaries with attention and mental workload
Attention makes information available, but attending to a cue is not the same as understanding a situation. A person can stare at a graph without seeing its trend; another may infer the trend after a brief glance because of prior knowledge. Situation awareness includes interpretation and anticipation relevant to a decision. The distinction matters when designing alerts. A louder alarm may capture attention but add little awareness if it fails to say which condition changed or what action is possible. The goal is to support sensemaking, not merely interruption.
Mental workload is related but separate. A person may work hard to maintain good awareness in a demanding environment. Another may experience low workload because automation handles routine monitoring yet be poorly prepared for an unusual event. A design could lower perceived workload while hiding critical dependencies, or raise workload temporarily while helping someone build an accurate picture. Workload ratings and awareness probes can therefore disagree without either being defective. Their relationship is itself a useful result if the study defines both constructs and includes the task outcome.
Expertise can improve awareness by helping a person notice patterns and project developments. It can also create expectation-driven errors when a rare event resembles a familiar one. A novice may check each value slowly, while an expert may recognize a trajectory quickly. Comparing the two requires tests that include both typical and deceptive cases. Training should not simply teach a checklist of facts; it should help users see relations among cues and notice when an expected pattern fails. A simulator can test this under controlled scenarios, but real-world transfer must be measured independently.
Time pressure exposes the distinction between knowing and acting. A responder may understand that conditions are deteriorating but lack the equipment or authority to intervene. A later incident review should not translate every poor outcome into deficient awareness. Conversely, timely action can occur by luck or routine without a robust understanding of the developing situation. Good analysis considers information access, interpretation, decision rules, resources, and communication. This broader view can identify system changes that improve safety without relying on blame.
Testing design claims
Suppose a new dashboard groups related measurements and displays trends. A credible evaluation would specify which predictions it is meant to improve, select realistic scenarios, and compare it with the current display under counterbalanced conditions. Researchers could probe awareness at decisive moments, measure workload and task outcome, and inspect how users communicate about the change. If users answer more projection questions correctly but take longer to locate a raw value, the tradeoff should be reported. The study should also include uncertain and abnormal cases, where a tidy visual story may conceal ambiguity.
Representative participants matter. A student using a generic simulation may learn an interface quickly, but that does not show how a practicing team will use it under interruption and institutional routines. Field studies can reveal missing constraints, though they introduce more variation and make causal attribution harder. A staged program can use laboratory work to isolate mechanisms, simulation to test coordinated action, and field follow-up to examine real use. Each stage supports a different inference. Claims should match the stage actually completed.
Measurement also needs humility about what counts as the “correct” situation. Dynamic environments contain uncertainty, and experts may disagree on the significance of an early cue. Scoring rules should allow justified uncertainty rather than reward overconfident guesses. A participant who says “I cannot tell yet” may have better calibrated awareness than one who names a precise future incorrectly. Good projection questions test plausible trajectories and confidence where possible. That makes awareness more than recalling a display value; it tests whether the person has a useful account of what may happen.
What the evidence does not show
The evidence does not prove that Endsley’s three levels are discrete brain modules, that one awareness score predicts all performance, or that every team should hold identical information. It does not show that making data visible guarantees comprehension, or that a freeze-probe score captures all of an unfolding real-world decision. Endsley’s theory and measurement work, distributed-awareness accounts, and measurement reviews offer complementary tools with distinct limits. The strongest claims identify the role, situation, relevant cues, moment of assessment, and decision the awareness measure was meant to support.
Situation awareness remains valuable because it asks a practical question: does the person or team have a sufficiently accurate and timely understanding to make the next decision? Answering that question requires more than a polished dashboard or a confident self-report. It requires attention to changing information, meaning, projection, and the channels through which knowledge moves. Used carefully, the concept points toward better displays, communication, and training while keeping responsibility with the system of work rather than reducing a complex outcome to an individual’s supposed failure to pay attention.
Research references
- Endsley, M. R. (1995). Toward a theory of situation awareness in dynamic systems. Human Factors, 37(1), 32–64. View source
- Endsley, M. R. (1995). Measurement of situation awareness in dynamic systems. Human Factors, 37(1), 65–84. View source
- Stanton, N. A. (2016). Distributed situation awareness. Theoretical Issues in Ergonomics Science, 17(1), 1–7. View source
- Sorensen, L. J., & Stanton, N. A. (2016). Inter-rater reliability and content validity of network analysis as a method for measuring distributed situation awareness. Theoretical Issues in Ergonomics Science, 17(1), 42–63. View source
- Moens, L. L., et al. (2026). Measuring situation awareness: A meta-review across domains. Human Factors, 68(5), 632–672. View source