Errors: An Inherent Part of Human-System Performance - Foundations for Designing User-Centered Systems - page 281

operators' heads when the accident happened. So once the human is reached in the
sequence of attributable causes, the analysis frequently gets terminated, and we are
left with human error as the result.
As we have noted several times, system performance is the result of people
interacting with technology in a particular context (organizational and physical
environment). The importance of context should never be underestimated. Very
often, when we look at accidents we find that the users were working in a context
constrained by time pressures and limited resources.
Nowadays, there is a greater awareness of the influence of the context in which
work takes place, and the fact that human attentional resources are limited (see
Chap. 5 ) . Many events that previously were typically attributed to humans being
accident prone would now be analyzed and categorized differently.
In aviation, for example, where multitasking is an inherent part of flying a
plane, distractions are recognized as being a particular problem. Dismukes et al.
( 1998 ) noted that nearly half the reported NTSB accidents attributed to crew error
involved lapses of attention associated with interruptions, distractions, or an
excessive preoccupation with one task to the exclusion of another that had to be
performed within a similar time frame. The vast majority (90%) of competing
activities that distracted or preoccupied pilots fell into four categories: commu-
nication; head-down work; searching for other traffic in good weather (visual
meteorological conditions or VMC); or responding to abnormal situations. Flight
crews have to work as a team, but this has to be done in such a way that it does not
detract from the individual tasks that have to be performed as the following
excerpt from incident report #360761 from NASA's Aviation Safety Reporting
System (ASRS) illustrates:
Copilot was a new hire and new in type: first line flight out of training IOE. Copilot was
hand-flying the aircraft on CIVET arrival to LAX. I was talking to him about the arrival
and overloaded him. As we approached 12,000 feet (our next assigned altitude) he did not
level off even under direction from me. We descended 400 feet before he could recover. I
did not realize that the speed brakes were extended, which contributed to the slow altitude
recovery.
Here the Pilot Not Flying (PNF) was trying to help the co-pilot (the Pilot Flying
or PF), which led to problems on two levels. First, the combination of flying the
plane and trying to heed the PNF's advice simply overloaded the PF. Second, the
fact that the PNF was focused on making sure that he gave the PF appropriate
assistance meant that he was distracted from his task of monitoring the ongoing
status of the plane. Both flight crew members were trying to do the right thing, but
they did not have enough resources to accomplish everything they needed to do.
The distractions in this case were at least partly self-created; such distractions
often lead to incidents in many domains (Baxter 2000 ). This incident would tra-
ditionally have been attributed to pilot error, but a closer examination of the
context suggests that this is an over-simplification.
We noted earlier that there is a fine line between success and failure. In the
aviation incident described above, where the plane descended too far, it seems