Phenotypes and genotypes

Erik Hollnagel

Ph.D., Professor, Professor Emeritus

 

Phenotypes and genotypes (of erroneous actions)

As industrial safe moved from the age of technology to the age of the human factor, 'human errors' became the near universal explanation for everything that went wrong. (See The NO view of 'human error' for further details.) Many models and a multitude of taxonomies were proposed to help understand the true nature of 'human error', although most of them were variations of a few basic ideas. A large number of books and a countless number of papers were written about the subject - and are still being written. Woe to the human factors specialist who cannot propose a taxonomy, or a least a category, of his/her own.

While hard-core human factors (and safety experts) subscribed to the old view of 'human error', a small but growing group of dissenters argued that the term 'human error' was misleading in its connotations. It would therefore be more correct to say that the human actions could be erroneous vis-à-vis current plan and goals, hence that one should talk about erroneous actions rather than 'human errors'. Erroneous actions were present in all types of work and in all domains. Erroneous actions wer regarded as inevitable in the sense that there was some minimal level of occurrence below which they could be reduced, no matter what was done.

  • One view was that erroneous actions were unavoidable consequences of the imperfection of humans (and specifically of human information processing), and that they would occur even under the best conditions. Erroneous actions were by their nature unpredictable in both form and frequency, although some patterns could be found. According to this view the best remedy was to replace humans by technology wherever possible. In this case it became important to know the statistics of erroneous actions and to find the areas that were most vulnerable. It was considered less necessary to elaborate the analysis and understand erroneous actions on a deeper level.
  • Another view was that erroneous actions were the results of unfavourable (working) conditions or an unforgiving environment. (This later became known as the 'new' view.) Erroneous actions were predictable both in form and frequency and it was therefore possible - in principle as well as in practice - significantly to reduce their number. According to this view the best remedy was to improve the working conditions by providing humans with appropriate tools and by amplifying human strengths rather than reducing human weaknesses. In this case it became important to know more about the nature of erroneous actions, both with regard to the conditions under which they occured and with regard to the causal “mechanisms” that could explain their manifestations. In other words, it was necessary to develop a detailed theory of erroneous actions that could be used to analyse events and guide the development of solutions.

Put differently, it became necessary to make a distinction between the phenotype and the genotype of erroneous actions. The former refer to the observable forms of erroneous actions. Other commonly used terms were error modes, surface forms, or manifestations. The latter correspond to the assumed causes or "mechanisms" of the erroneous actions. Whereas the phenotypes can be observed, the genotypes can only be inferred - leaving out the case of introspection. The phenotype-genotype categories are thus primarily applied to acknowledged erroneous actions and provide a more consistent way of describing and accounting for erroneous actions, both as part of an event analysis and in performance prediction.

The phenotypes and genotypes represented two fundamentally different ways to consider erroneous actions. The phenotype was concerned with how an erroneous action appeared in overt action, how it could be observed, hence the empirical basis for a classification. The genotype was concerned with the possible cause, in particular the functional characteristics of the human cognitive system that were assumed to contribute to an erroneous action - or in some cases even be the complete cause! The phenotypes and genotypes required two different set of categories, one for observation and one for interpretation. In cases where a strong or well established theory existed there was little risk in using the categories for interpretation to define the categories for observation, hence to combine observation and interpretation to some extent. But in the absence of a strong theory, this combination should be avoided as far as possible. The behavioural study of erroneous actions - whether it was called psychology, human factors , or cognitive engineering - is an example of this. As long as there were no strong theories of human action, observation and interpretation should be kept separate in empirical investigations. Needless to say, this advice was rarely followed.

Today - in the second decade of the 21st Century - the belief in human information processing / cognitive functions as a valid foundation for analysing and describing human performance is on the vane. Developments in the study of safety, most notably resilience engineering, have also reduced - although not yet quite eliminated - the need to distingush between correct and incorrect human actions. The consequence of this has been called the no view (of 'human error'), according to which 'human error' is a meaningless category. This means that there is no need to use the concept of the genotypes to explain incorrect performance, since things that go right and things that go wrong happen in the same way. On the other hand, there is still a legitime academic and practical need to develop less confusing and more consistent categories to describe the manifestations of actions - whether the outcomes are deemed acceptable or unacceptable.

Copyright © Erik Hollnagel 2014 All Rights Reserved.