Perception-cognition-action

Dr.-Ing. Marcus JenkeSenior Human Factors Engineer | Managing Partner
Last updated: October 2026
Short definition

Perception-cognition-action, PCA for short, describes basic aspects of human information processing when interacting with a product: Users must first perceive information, then interpret and process it, and finally plan and carry out an action. In usability engineering, PCA serves as a practical analysis framework.

The model helps to systematically distinguish perception, comprehension, decision-making and execution problems and to derive suitable measures for designing the user interface. The three areas should not be understood as completely separate or always strictly sequential stages. Perception, cognition and action influence one another and can jointly contribute to the occurrence of a use error.

On this page

The three stages

The model goes back to descriptions of human information processing in cognitive psychology and divides the interaction between human and product into three basic areas:

  • Perception: perceiving. Information from the product or the environment is taken in through the senses: A measured value is read, an alarm is heard, a vibration is felt, or the position of a control is recognized. See perception.
  • Cognition: processing and deciding. What has been perceived is interpreted, compared with prior knowledge and expectations, and evaluated with regard to its meaning. This includes attention, memory, reasoning, problem solving, decision-making and the formation of an intention to act. Cognitive load can limit human information processing here.
  • Action: acting. The intention that has been formed is turned into an action: A button is pressed, a dial is set, a connector is attached, or a medical procedure is performed on the patient.

The three areas mark different points at which problems in the interaction can arise. They are, however, interconnected: Expectations and prior knowledge already influence perception, while feedback from an action that has been carried out in turn triggers new perception and decision processes.

Assigning errors along the stages

The practical value of the model lies in the structured analysis of possible causes of a use error. An observable use error can look the same from the outside even though it has different underlying causes. For developing effective measures it is therefore important to examine whether perception, interpretation, decision-making or the execution of the action contributed to the error. Typical problems are:

  • Perception level: Relevant information is not perceived, or not perceived correctly. A warning is overlooked, an alarm is masked by ambient noise, or a decimal point is misread because of an unsuitable presentation.
  • Cognitive level: The information is perceived but misinterpreted, not remembered or evaluated incorrectly. A symbol is misunderstood, a unit of measure is confused, a system state is misjudged, or a necessary action step is forgotten.
  • Action level: The intention is fundamentally correct but is not carried out as intended. The user presses an adjacent button, turns a dial in the wrong direction, attaches a connector incompletely, or applies unsuitable force.

A use error should not be hastily assigned to a single category alone. For example, a display that is too small can first make perception harder, an unclear unit can lead to a misinterpretation, and the resulting wrong action can go undetected because of missing system feedback.

PCA analysis complements the established distinction between execution errors and errors in planning or goal formation. Slips and lapses typically concern the execution or the recall of a fundamentally correct intention. Mistakes, by contrast, result from incorrect goal formation, interpretation or planning. These error types are, however, not fully congruent with the three PCA areas.

Suitable measures for each stage

Different design measures can be derived from the cause analysis. Measures should not be selected solely on the basis of a single PCA category. What matters is which factors actually contributed to the use error. Typical measures are:

  • For perception problems: Improve contrast, size, position and display duration, take noise levels and frequency ranges into account, use several sensory channels, and, where necessary, code safety-relevant information redundantly.
  • For cognitive problems: Adapt terminology to the user group, present system states unambiguously, design interactions that match expectations, reduce memory demands and mental calculations, and provide relevant information at the place and time of the decision.
  • For action problems: Separate controls spatially, make them clearly distinguishable, safeguard critical inputs, prevent mix-ups through design, and provide appropriate visual, auditory or haptic feedback. See error tolerance.

A common mistake in practice is to try to solve a perception, comprehension or action problem solely through an additional note in the instructions for use or through supplementary training. Such measures can remain ineffective if the relevant information cannot be perceived, understood, remembered or turned into a safe action in the actual situation of use. It should therefore first be examined whether the problem can be avoided or reduced through the design of the user interface.

Application in task analysis and risk analysis

In the task analysis, PCA can be used as a systematic grid. For each action step, the following is examined:

  • What information must the user perceive?
  • What information must the user understand or remember?
  • What decision must the user make?
  • What action must the user carry out?
  • What feedback does the user need to recognize that the action was successful?

This view uncovers requirements that are easily overlooked in a process description that focuses only on actions. Examples include tacitly assumed knowledge, necessary mental calculations, system states that are hard to recognize, or missing feedback.

In the use-related risk analysis, the grid supports the systematic identification of possible use errors and their causes. For each relevant action step, the following can be examined:

  • What might not be perceived, or be perceived incorrectly?
  • What might be misunderstood, confused or forgotten?
  • What wrong decision might be made?
  • What might be carried out incorrectly, incompletely or not at all?
  • What feedback might be missing or misinterpreted?

PCA can also serve as a structuring aid in the root cause analysis of problems observed in formative or summative evaluations. The assignment should not, however, rely solely on the observed action or on a study participant's self-report. It should be derived, where possible, from several sources of information, for example:

  • Observation of the actual use behavior
  • Targeted root cause probing
  • Interview statements from the participants
  • Analysis of system states and system responses
  • Video, audio or log data
  • Information about the context of use, distractions and working conditions

Limits of the model

The division into perception, cognition and action is a useful simplification but does not fully represent human information processing.

The areas do not always proceed in strict sequence. Prior knowledge, expectations and attention already influence which information is perceived. At the same time, actions change the situation and generate new information that in turn has to be perceived and processed.

The boundaries between the categories are also not always clear. If, for example, an alarm is heard but not recognized as safety-relevant, this can be interpreted as a perception problem as well as a cognition problem. For design purposes, what matters is therefore less the formal category than a traceable description of the underlying factors.

In addition, PCA takes only limited account of: collaboration and team communication, handovers and changes of responsibility, organizational processes, workload and staffing, interruptions and competing tasks, social and cultural influences, interactions between several products, and management and organizational decisions. Such questions require complementary methods, for example a workflow analysis, a context analysis or a systems ergonomics perspective.

Perception-cognition-action is thus a helpful analysis tool, but not a complete theory of human behavior and not a sole method for determining the causes of a use error.

Regulatory reference

Perception-cognition-action is not a standalone normative requirement. Rather, the model serves as an aid for analysis and structuring within the usability engineering process.

IEC 62366-1 describes the interaction between the medical device, the user interface, the user and the context of use. Perception, processing and action are fundamental components of the interaction between user and product. The standard does not, however, require that all use errors be documented using a PCA classification. What matters is that foreseeable use errors, use-related hazards and hazard-related use scenarios are systematically identified, analyzed and controlled through suitable measures.

IEC 62366-2 contains supplementary background information and practical guidance for implementing usability engineering. As a technical report, it is informative, contains no binding requirements and is not intended as a standalone basis for regulatory conformity.

In the FDA context as well, perception, cognition and action can serve as helpful categories for analyzing use problems. The PCA approach is used in particular to investigate, during usability evaluations, why a use error, a close call or a use difficulty occurred. For regulatory evidence, however, the formal naming of a PCA category is not what matters. What is essential are a traceable root cause analysis, the assessment of possible consequences and the derivation of suitable measures to reduce use-related risks. The FDA also places human factors within the design and development requirements under 21 CFR 820.10(c) and ISO 13485, clause 7.3 (as of October 2026).

In brief

Perception-cognition-action structures the interaction between user and product along three basic areas: perceiving, processing and deciding, and acting. The model helps to analyze possible causes of use errors and to derive suitable measures for designing the user interface.

The three areas are, however, not completely separate from one another. A use error can arise from several perception, cognition, action and context factors at the same time. PCA is therefore a helpful analysis tool, but not a complete error taxonomy and not a substitute for a comprehensive task, context, risk and root cause analysis.

Frequently asked questions (FAQ)

Is the model normatively required?

No. Perception-cognition-action is not a mandatory analysis method. The model can be used as a helpful grid for task analyses, use-related risk analyses and root cause analyses. What matters is not the use of the term PCA but the systematic and traceable analysis of use-related problems.

What is the difference between an execution error and an intention error?

In an execution error, the intention is fundamentally correct but the implementation does not go as planned, for example the accidental press of an adjacent button. In an intention or planning error, the decision itself is already based on an incorrect interpretation, assumption or goal formation, for example selecting the wrong dose because a unit of measure was misunderstood. Execution errors are often called slips, errors caused by forgetting are called lapses, and errors in planning, interpretation or goal formation are called mistakes.

How do you classify an error when several stages are affected?

An error does not necessarily have to be assigned to a single stage. Often several factors act together. In this case, all relevant contributing causes should be documented, and a distinction can be made between the primary cause, further contributing factors and the conditions under which the error was not detected or corrected in time. The classification should make clear which measures are required.

Why is the classification needed at all?

The classification supports the selection of effective design measures. An overlooked warning is not solved by a clearer wording alone, a comprehension problem is not automatically fixed by a larger font, and a correctly formed intention does not protect against the accidental operation of an unsuitable control. PCA thus helps to avoid hasty or purely document-based measures and to address problems at their cause.

Do you want to not only document use errors but analyze their causes reliably and reduce them through suitable design measures? We support you from task analysis and use-related risk analysis through formative evaluations to human factors validation.

More about our usability engineering

Sources

Related terms

← Back to the wiki overview