Preparing for the Unexpected: A Psychologist’s Case for Improved Training Although design and automation help reduce human error incidents and accidents, training shortcomings must be corrected to enhance crew performance improvements. by H. Clayton Foushee, Ph.D. U.S. National Aeronautics and Space Administration (NASA) Ames Research Center, Moffett Field, California (Foushee is now with the U.S. Federal Aviation Administration, FAA, in Washington, D.C.) (Presented at the International Airline Pilot Training Seminar conducted by VIASA Airlines and the Flight Safety Foundation in Caracas, Venezuela, on January 23-25, 1989) The large percentage of aircraft incidents and accidents incidents involving near stalls caused by advanced flight attributed to human error has focused increasing atten- guidance systems left engaged in the vertical speed tion on the performance characteristics of the individ- mode. ual pilot. Traditionally, human factors specialists have channeled their research energies toward exploration of The point is that design and automation are not the only the human information processing and perceptual as- answers. They will only be effective to the extent that pects of the pilot’s job, with the important goal of new training approaches are developed to keep pace, designing equipment best suited to the characteristics and I do not believe that training has kept up. I will of the human operator. focus upon what I believe to be three major shortcom- ings of current training. First, the training environment Since it is commonly agreed that humans will always is preprogrammed so that everyone knows what to ex- be fallible creatures, our goal has been to design and pect. We do an excellent job of training procedures to automate systems in areas where human operators are handle predictable situations, but as we all know, most notoriously unreliable. Thus, human factors psycholo- accident scenarios are combinations of very unpredict- gists and aeronautical engineers have continually re- able events. Psychological and learning theories sug- fined designs such that each generation of aircraft promises gest some potentially serious drawbacks to current training not only better performance and reliability, but also approaches and I would like to discuss some of these reduced pilot workload. Despite this effort, between 65 along with potential solutions. and 75 percent of accident causes (depending on whose statistics you listen to) continue to fall into the human Second, almost all training has been oriented toward error category. perfecting the technical skills of individual pilots. Yet, most incidents and accidents are not attributable to the Unfortunately, improved design and increased automa- errors of a single crewmember — they are often the tion are not the only answers. If they were, we should result of breakdowns of crews. be seeing a steady decrease in the number of human error incidents and accidents, but we are not. My col- Third, the regulatory environment may not be able to leagues at NASA have been conducting evaluations of handle some of the new training innovations that we the introduction of new generation transport aircraft believe will address current and future operational problems. into service. These studies have not been published Contrary to popular belief, this is not the fault of the yet, but we are seeing some potentially troubling signs. rulemakers. To a very large extent, the regulations are One of these “clouds on the horizon” is a reduction in merely a reflection of current operational practices. some types of errors along with the creation of a new However, if we are to do a good job of handling prob- class of human error. A good example is a rash of lems one and two, it is likely that regulations governing FLIGHT SAFETY FOUNDATION • F LI GH T SA F E TY DI G E S T • MARCH 1990 1 training will have to evolve to take best advantage of plane). However, recent research has found that the new training approaches. overlearning phenomenon extends to complex behav- ioral sequences involving verbal interaction and deci- sion making. Problem Number 1 — Dangers of How many times when someone asks you how are you Preprogrammed Training doing, do you respond, “fine,” when actually if you Environment thought about it you couldn’t be worse. This is an example of a mindless behavior. I have seen numerous The first problem area is concerned with the dangers of examples of mindless behavior in air carrier cockpits a totally preprogrammed training environment. Most such as checklists accomplished with items incorrectly flight training today is accomplished through the use of or not completed because the normal overlearned re- a highly structured training syllabus which is designed sponse is “checked” or “set.” Bells, buzzers and whistles to cover everything considered important for individual are ignored because you hear them too frequently. Clear- pilots to accomplish their jobs safely and efficiently. ances are incorrectly executed because the last 27 times There is almost no question that most advanced flight you flew in to airport X on a nice day, you got the same training does an excellent job of providing pilots with clearance. These are all examples of mindless, highly adequate technical knowledge in most areas. It is also overlearned behaviors. doing a good job of standardizing procedures so crew- members can theoretically work together and, of course, This phenomenon suggests that training programs, whose pass flight checks. underlying philosophies suggest that performance can be improved through the repeated practice of predeter- The basic methodology consists of the classroom pres- mined standard operating procedures, are unrealistic. entation of technical knowledge coupled with extensive What such approaches may be contributing to is a rather drills in procedures trainers and simulators until profi- mindless adherence to a set of operating procedures. ciency is achieved in all areas defined by the syllabus. The problem is that you simply cannot develop a stan- Everyone knows what he or she will be expected to do. dard operating procedure for every situation. An ex- In short, every procedure is very well learned, or in amination of the incident and accident record confirms psychological terms, pilots are programmed or condi- that these events almost always occur as a result of a tioned to respond according to the appropriate stimulus number of highly unusual circumstances for which there cues that are defined by the training process. simply is no standard operating procedure. It is important to point out that the training environ- We often talk about crew errors caused by compla- ment is usually a highly controlled situation where one cency, but I would argue that what we are really talking stimulus at a time is presented, the appropriate condi- about in many situations are instances of mindlessness tioned response is eventually acquired with practice, caused both by training programs that are procedures and thus can be accomplished very reliably over and oriented and by an operational environment where equip- over again. It is also important to note that this is a ment is so reliable that nothing ever happens. The very effective technique for learning discrete behaviors distinction between mindlessness and complacency is in isolation. important because there are two very different psycho- logical processes involved. Such techniques are effective, but they may contribute to a psychological phenomenon known as mindlessness The complacent pilot may be one suffering from low (Langer, 1980). Mindlessness refers to the fact that motivation, and training may be of limited effective- when certain behaviors are overlearned, humans tend to ness in dealing with motivational problems. On the react automatically when any stimuli similar to those other hand, I would argue that mindlessness can be present when the behaviors were learned triggers the decreased through various training processes, which dominant or overlearned response. When a “mindless” induce cognitive processing and awareness. behavior sequence is initiated, the behavioral process is often not at the level of conscious awareness. There is another danger associated with overlearning procedures. Psychological research overwhelmingly More often than not, an individual cannot recall or indicates that in highly stressful situations, humans tend verbalize the precise sequence of events leading up to to exhibit a narrowing of perceptual attention, or tunnel the behavior or even remember components of the ac- vision. Cognitive processing is more difficult under tual response itself. These overlearning effects are, of such circumstances, and we tend to react instinctively. course, well-known with regard to various motor be- This is simply another way of saying that we tend to fall haviors (e.g. walking, riding a bicycle, driving a car, back on well-learned or dominant responses. For pilots even the mechanical acts involved with flying an air- operating in stressful circumstances, this phenomenon 2 FLIGHT SAFETY FOUNDATION • F LI GH T SA F E TY DI G E S T • MARCH 1990 has significant advantages and disadvantages. for ice accumulation during a simulator session. On the positive side, if the dominant or overlearned If done well, LOFT can serve as a substitute for valu- response is appropriate, it can be summoned quickly able experience that pilots no longer get because air- and precisely in situations where there is little time to craft and their systems