CREAM.
Cognitive Reliability and Error Analysis Method
Originator Erik Hollnagel (1998)
Paradigm Second-generation human reliability analysis
Unit of analysis Cognitive function in context
Primary domains Aviation, nuclear, rail, maritime, healthcare

CREAM is a second-generation Human Reliability Analysis (HRA) method that treats human performance as context-driven rather than as a set of inherent error probabilities. It uses a cognitive model (COCOM), a classification of erroneous actions, and a set of Common Performance Conditions (CPCs) to diagnose and predict human-system performance.

Overview of the framework

CREAM (Hollnagel, 1998) responds to the limitations of first-generation HRA techniques such as THERP by making context the centre of analysis. It couples three building blocks: (1) the Contextual Control Model (COCOM), which maps operator control into four modes — scrambled, opportunistic, tactical, strategic — ranked by increasing reliability; (2) a classification of phenotypes (observable error modes — timing, duration, sequence, object, force) and genotypes (individual, technology and organisation-related causes); and (3) nine Common Performance Conditions — adequacy of organisation, working conditions, adequacy of MMI & support, availability of procedures/plans, number of simultaneous goals, available time, time of day, adequacy of training & experience, and crew collaboration quality. The Basic CREAM provides a qualitative probability of action failure; the Extended CREAM produces nominal and context-adjusted probabilities for specific cognitive functions (observation, interpretation, planning, execution).

9 Common Performance Conditions (CPCs) Organisation Working cond. MMI & support Procedures # of goals Available time Time of day Training Crew collab. → Context score → Control mode Scrambled Opportunistic Tactical Strategic very low reliability very high reliability Nominal × context-adjusted P(action failure)
Figure 1. CREAM logic: nine Common Performance Conditions characterise the context, which maps to one of four COCOM control modes and, ultimately, a context-adjusted probability of action failure for specified cognitive functions.

When to use it

Typical applications

  • Retrospective analysis of incidents/accidents with dominant human-performance issues.
  • Prospective HRA of new procedures, automation or staffing levels.
  • Screening analyses where a qualitative probability of action failure is sufficient.
  • Integration with QRA/PRA as the human-reliability submodel.

Aviation relevance

  • Flight-deck task analysis; go-around, unstable-approach, rejected-take-off decisions.
  • ATC task reliability under time pressure and high traffic load.
  • Line-maintenance task analysis and dispatch decisions.
  • Published aviation adaptations and extensions include CARA and fuzzy-CREAM.

Benefits

Analytical strengths

  • Anchors HRA in a cognitive model (COCOM) rather than opaque error-probability tables.
  • Explicit treatment of context through CPCs makes assumptions visible and auditable.
  • Bridges qualitative diagnosis (phenotypes/genotypes) and quantitative screening (action-failure probability).
  • Scalable: Basic CREAM for screening, Extended CREAM for detail.

Practical strengths

  • Widely taught; substantial body of applied case studies in aviation, rail and nuclear.
  • Compatible with existing task analyses (HTA, SHERPA).
  • Supports both retrospective and prospective work with the same taxonomy.
  • Numerous fuzzy, Bayesian and evidential extensions published since 2010.

Limitations

  • Quantification uncertainty. CPC scoring and the mapping to control modes rely on expert judgement.
  • Coarse discrete scale. Only four control modes and fixed CPC levels; extensions often add fuzzy/continuous scales.
  • Second-generation, not systemic. Less suited than FRAM/STAMP for emergent socio-technical dynamics.
  • Resource-intensive in Extended form; Basic screening may be too coarse for certification use.
In short CREAM operationalises the insight that people are not unreliable — contexts are. It remains a reference method for human-reliability analysis in aviation and other safety-critical domains, particularly where COCOM-style control modes fit the decision process.

References (APA 7)

Hollnagel, E. (1998). Cognitive reliability and error analysis method (CREAM). Elsevier Science.

Hollnagel, E. (1993). Human reliability analysis: Context and control. Academic Press.

Kirwan, B. (1998). Human error identification techniques for risk assessment of high-risk systems — Part 2: Towards a framework approach. Applied Ergonomics, 29(5), 299–318. https://doi.org/10.1016/S0003-6870(98)00011-6

Stanton, N. A., Salmon, P. M., Rafferty, L. A., Walker, G. H., Baber, C., & Jenkins, D. P. (2013). Human factors methods: A practical guide for engineering and design (2nd ed.). Ashgate.

Bedford, T., & Cooke, R. M. (2001). Probabilistic risk analysis: Foundations and methods. Cambridge University Press.

Further reading

Kirwan, B., Gibson, W. H., Kennedy, R., Edmunds, J., Cooksley, G., & Umbers, I. (2004). Nuclear action reliability assessment (NARA): A data-based HRA tool. Proceedings of PSAM 7 / ESREL '04.

Yang, Z. L., Bonsall, S., Wall, A., Wang, J., & Usman, M. (2013). A modified CREAM to human reliability quantification in marine engineering. Ocean Engineering, 58, 293–303. https://doi.org/10.1016/j.oceaneng.2012.11.003

Akyuz, E., & Celik, M. (2015). A methodological extension to human reliability analysis for cargo tank cleaning operation on board chemical tanker ships. Safety Science, 75, 146–155. https://doi.org/10.1016/j.ssci.2015.02.008

Ung, S.-T. (2015). A weighted CREAM model for maritime human reliability analysis. Safety Science, 72, 144–152. https://doi.org/10.1016/j.ssci.2014.08.012

SKYbrary. (n.d.). Cognitive Reliability and Error Analysis Method (CREAM). https://skybrary.aero/articles/cognitive-reliability-and-error-analysis-method-cream