Errors: An Inherent Part of Human-System Performance - Foundations for Designing User-Centered Systems - page 280

success as much as failure because there is often only a fine line dividing the two,
and there are many more instances of success than of failure (e.g., Hollnagel et al.
2006 ). This changed emphasis is a reflection of Ernst Mach's ( 1905 ) prescient
view that ''Knowledge and error flow from the same mental sources, only success
can tell one from the other……the same mental functions, operating under the
same rules, in one case lead to knowledge, and in another, to error….''. Although
couched in slightly different terms, it is a view that others have concurred with and
reiterated, such as Reason ( 1990 , p.1):
Not only must more effective methods of predicting and reducing dangerous errors emerge
from a better understanding of mental processes, it has also become increasingly apparent
that such theorizing, if it is to provide an adequate picture of cognitive control processes,
must explain not only correct performance but also the more predictable varieties of
human fallibility. Far from being rooted in irrational or maladaptive tendencies, these
recurrent error forms have their origins in fundamentally useful psychological processes.
The consequences of errors have also increased over the years. Or perhaps that
should be the consequences are perceived to have increased. The mass media these
days are often very quick to report on air accidents, for example, where a single
accident may give rise to hundreds of casualties. The fact that more people get
killed on the roads, however, goes largely unreported, mostly because each fatal
road accident often only involves a few deaths (Gigerenzer 2004 ).
10.1.3 The Accident was Caused by Human Error, Right?
Most accidents could naively be attributed to human error because the systems that
fail, leading to the accident, are designed by humans. This is an over-simplistic
view, however, and would lead to an equally simplistic solution, i.e., that removing
the human would remove a major source of failures, and hence eliminate many
accidents. The idea of humans being accident prone in a volitional way (i.e., of
their own free will) dominated early thinking in human error research.
The cause of an accident is often attributed to human error (pilot error, driver
error, operator error, and so on). This is a judgment that is built on several
underlying assumptions that, at best, usually only represent a partial view of the
true situation. There are many examples of such a view. Arnstein ( 1997 ), for
example, found that the number of problems that could be attributed to human
error in anesthetics ranged from 64 to 83%, whilst in aviation the range was
40-88%. Johnson and Holloway ( 2007 ) also noted the tendency to over-emphasize
human error as the reported cause in transportation accidents for the years
1996-2003. Whilst it was still found to be the main attributed cause, the levels
were somewhat lower at 37% for the US National Transportation Safety Board
(NTSB), and 50% for the Canadian TSB.
One of the main reasons that accidents end up being attributed to human error is
because of the limitations of causal analysis. It is difficult to continue the analysis
through the human when we do not have direct access to what was going on in the