Human Error Explained: 4 Failure Modes That Change an Investigation

People often write “human error” in an incident report before they know what happened. That shortcut closes the inquiry too early because the same visible action can result from a memory lapse, a poor plan, an impossible rule, or a deliberate shortcut that the work system has quietly rewarded.

Human error is an unintended action, decision, or omission that contributes to an unwanted outcome, but the label alone does not explain why the action made sense at the time. In safety investigations, classifying the failure mode helps teams choose the right evidence, test the surrounding controls, and avoid turning a system problem into a person’s character judgment.

What does human error mean in an investigation?

Human error describes the form of a failed action, not the complete cause of an incident. James Reason’s work is valuable here because it separates active failures from the latent conditions that make those failures more likely and more consequential.

That distinction matters when a report says that an operator “forgot,” “misread,” or “did not follow the procedure.” Each phrase may be accurate at the surface while still leaving the important questions unanswered. Was the task interrupted? Was the procedure usable? Did the design make the wrong action easier? Had supervisors accepted the same workaround for months?

Why does classification matter before blame is assigned?

Classification matters because different failure modes demand different evidence. Treating every action as carelessness produces generic retraining, while the exposure remains in the task design, staffing model, control logic, or management decision.

Andreza Araujo’s Sorte ou Capacidade frames accidents as systemic outcomes rather than random events. That position supports a practical investigation rule: name the action precisely, then widen the lens until the barriers and decisions around it are visible.

For a second perspective, compare this taxonomy with the evidence discipline described in Evidence Chain Explained. The classification is only useful when the evidence chain can support it.

What are the four human-error failure modes?

The four modes below are not excuses and they are not interchangeable labels. They describe different relationships between intention, planning, execution, and rules.

Slip

A slip occurs when the person intends to perform the correct action but executes it incorrectly. Examples include selecting the wrong switch, placing a control in the wrong position, or starting a familiar sequence before the final verification is complete.

Lapse

A lapse is a memory or attention failure in which an intended action is omitted, delayed, or forgotten. Interruptions, fatigue, competing alarms, and poorly designed handovers can increase the likelihood of a lapse without changing the person’s underlying intention.

Mistake

A mistake occurs when the plan or judgment is wrong, even if the execution follows that plan accurately. The investigator should examine the information available, the assumptions used, the competence expected, and whether the decision rules matched the actual conditions.

Violation

A violation is a deliberate departure from a rule, procedure, or expectation. That description should trigger more questions, not fewer, because repeated violations can reveal a rule that conflicts with production, a control that is impossible to use, weak supervision, or a local norm that leadership has tolerated.

How do investigators differentiate the four modes in practice?

The first question is whether the person intended the action that occurred. If the answer is no, the review should distinguish execution from memory. If the answer is yes, the review should ask whether the plan was sound and whether the rule was knowingly bypassed.

Which evidence should change the investigation route?

A slip or lapse should not automatically become a training action. Review the conditions that shaped attention and execution, then test whether a design or verification control could make the error easier to detect.

A mistake should move the investigation toward decision quality. The relevant evidence may include the risk assessment, the assumptions behind it, the quality of field information, and the point at which uncertainty was escalated.

A violation should move the review toward rule and governance quality. The investigator needs to know whether the rule was clear, compatible with the work, supervised consistently, and supported by the time and equipment required to follow it.

These routes connect with the distinction between causal layers in Causal Factors Explained. A label on the action is not the end of the causal map.

When should leaders use this model?

Leaders should use the model after incidents, near misses, recurring deviations, and failed critical-control checks. It is especially useful when a report contains a vague phrase such as “human error” but does not show what the person intended, what information was available, or what control should have caught the failure.

The model also helps leaders compare weak signals without collapsing them into a single behavior metric. A recurring lapse in a high-interruption task is different from a recurring violation of a rule that no one can apply during a normal shift.

What should change after the investigation?

The outcome should be a sharper control decision. A slip may justify a design change or independent verification. A lapse may require a handover, workload, or interruption control. A mistake may require a better decision boundary. A violation may require a rule redesign, clearer authority, or a governance response.

Use the classification to test whether the corrective action addresses the conditions that made the action likely. If the only response is to tell the worker to be more careful, the investigation has probably named the event without explaining it.

For a broader precursor view, read Heinrich Pyramid Explained and ask which repeated signals deserve control review before a serious outcome occurs.