Judgment and human error

Definition

Heuristics are simplified strategies of judgment that are usually efficient but can produce systematic errors called biases. Satisficing is the choice of the first option that meets an acceptable threshold instead of the optimal one. Situation awareness is a person’s perception and understanding of the elements of a situation and the projection of their future state. Human error, in the cognitive sense, is a planned action that fails to achieve its intended outcome without the intervention of chance; the machine safety sense of human error concerns its role in hazards and risk reduction.

Historical development

Herbert Simon (1955, 1956) described bounded rationality and introduced the term satisficing. Amos Tversky and Daniel Kahneman (1974) described the availability, representativeness and anchoring heuristics and the biases they produce; Gerd Gigerenzer and colleagues later argued that simple heuristics are often well adapted to real environments. Jens Rasmussen (1983) distinguished skill-based, rule-based and knowledge-based behavior. Donald Norman (1981) classified slips, and James Reason (1990) systematized errors into slips and lapses — failures of execution and memory when the plan is adequate — and mistakes — failures of the plan itself, rule-based or knowledge-based — and distinguished violations as deliberate deviations. Mica Endsley (1995) defined three levels of situation awareness: perception, comprehension and projection.

Main models

1. Heuristics and biases. Judgments based on ease of recall, similarity or an initial value deviate predictably from normative standards.
2. Bounded rationality and satisficing. Decisions are limited by information, time and cognitive capacity.
3. Skill-, rule- and knowledge-based behavior (SRK). Routine actions run automatically; familiar problems are handled with stored rules; novel problems require reasoning from a model of the system.
4. Slips, lapses and mistakes. Error types correspond to the levels of behavior and call for different countermeasures.
5. Situation awareness. Many decision failures originate in missed or misinterpreted information rather than in the decision itself.

Function and purpose

The models explain why competent people make errors, why errors cluster in particular situations — interruptions, unusual conditions, changes from a familiar routine — and why different errors need different support. They replace the question of who made an error with the question of what conditions made it likely.

Applications in technical communication

Documentation practice draws consequences for each level, as set out in the context card Human error types and troubleshooting. Slips and lapses in routine tasks are addressed with checklists, clear step states and cues at points where steps are often omitted. Rule-based mistakes are addressed with troubleshooting information organized by observable symptoms, so that readers select the correct rule; DITA troubleshooting topics give this a fixed structure of condition, cause and remedy. Knowledge-based situations require concept information that supports reasoning about the system and information on how to reach a safe state and recover. Decision points in procedures state the criteria for each branch explicitly instead of relying on judgment under time pressure. Information on reasonably foreseeable misuse and on residual risks follows from the risk assessment; error models help to present it, but they do not determine what must be included, and documentation is not a substitute for design measures and safeguards.

Limitations and evidence

Error classifications are applied after the event and are open to hindsight bias: an action is called an error once its outcome is known. Researchers in safety science have criticized “human error” as an explanation that ends an investigation too early. The core heuristics findings have been widely replicated, but some effects associated with the wider literature on judgment, including claims about willpower as a depletable resource behind “decision fatigue”, did not hold up in large preregistered replications, and popular summaries often overstate the reach of individual biases. Situation awareness is a construct with competing definitions and measures. None of these models predicts the probability of a specific error in a specific task without analysis of that task.

Relation to other areas

Judgment uses what attention selected, memory holds and comprehension modeled. Its models connect technical communication with human factors, task analysis and machine safety, where error is treated in the risk assessment.

Conclusion

Research on judgment and error shows that errors have types and conditions. Documentation responds with different support for routine execution, rule selection and reasoning about novel situations, and with recovery information, while design and safeguards remain the primary means of risk reduction.