Introduction to Industrial Engineering

By Jane M. Fraser

Chapter 9


Return to the Table of Contents.

9.3 Cognitive engineering

At least some of the events at Three Mile Island (recall our discussion in Chapter 8) can be attributed to difficulties workers had in figuring out what was going on in the reactor. One problem is that the workers’ normal job largely consists of monitoring a smoothly running reactor (yes, picture Homer Simpson). Such a job is boring and can quickly lead to lack of vigilance. When a problem occurs, the person is “out of the loop” because the computer controls have been running the plant. The workers must spend time figuring out what has happened. A second problem at Three Mile Island was that the design of the control room at that reactor did not convey crucial information to the workers, particularly the level of coolant in the reactor; they had to infer that level from other indicators.

Cognitive engineering builds on knowledge from psychology about human abilities in memory, perception, reasoning, and attention to design tasks that a human can do with efficiency, quality, and safety. Again, the focus is on adapting the workplace to the human. A human who must remember certain tasks in a specific order can be given a checklist. A human who has to perceive a change in an array of displays can be aided by a computer that detects the changes and alerts the human (for example, cockpit alarms for loss of altitude). A human who has to do a complicated set of reasoning can be supported by a computer system (for example, an immunohematologist who must interpret blood tests to identify antibodies in a patient’s blood). A person who must pay attention to several sources of information can share the task with computers and with other humans.

A balance must be achieved between understimulating the human, leading to boredom, and overstimulating the human, leading to stress. Both can lead to losses in efficiency, quality, and safety. Generally, the human performs better when the worker clearly has control of the environment, including work pace. Shifting control to the computer can lead to boredom, stress, inattentiveness, and overreliance on the computer.

The design of controls, including computer hardware and software, to support human tasks requires careful analysis of usability, which is affected by screen layout, task sequence, and many other factors. NASA's Human Systems Integration Division

advances human-centered design and operations of complex aerospace systems through analysis, experimentation, and modeling of human performance and human-automation interaction to make dramatic improvements in safety, efficiency, and mission success.

This FAA analysis of an airliner accident in 1993, in which 2 people were killed, shows the interplay of the design of the controls, the training of the pilot, and the behavior of passengers.

Flight 583 was level at 33,000 feet when the leading edge slats deployed inadvertently. The autopilot disconnected and the captain was manually controlling the airplane when it progressed through several violent pitch oscillations and lost 5,000 feet. ...

The National Transportation Safety Board determines that the probable cause of this accident was the inadequate design of the flap/slat actuation handle by the Douglas Aircraft Company that allowed the handle to be easily and inadvertently dislodged from the up/ret [retracted] position, thereby causing extension of the leading edge slats during cruise flight. The captain's attempt to recover from the slat extension, given the reduced longitudinal stability and the associated light control force characteristics of the MD-11 in cruise flight, led to several violent pitch oscillations. Contributing to the violence of the pitch oscillations was the lack of specific MD-11 pilot training in recovery from high altitude upsets, and the influence of the stall warning system on the captain's control responses. Contributing to the severity of the injuries was the lack of seat restraint usage by the occupants.

The root cause of that accident was identified as poor design of a handle. This paper analyzes three cases that the author identifies as human error. The author makes several recommendations, including that all pilots be required to take a course in pyschology.


Learning a task involves cognitive and physical improvement. Increased familiarly with the task and improved dexterity lead to a reduction in the time to do a job. Various learning curve equations are used to describe empirical relationships about what improvement can be expected from learning.

The Wright learning curve is described by stating that a doubling of the cumulative number of units produced leads to a (100-)L% reduction in the cumulative average production time. For example, with a 80% Wright curve, if the first unit takes 100 minutes, then the first 2 units will take an average of 80 minutes, the first 4 units will take an average of 64 minutes, and so forth. It can be shown that the Wright curve must have an exponential equation, where a and b are constants.

Y = aXb

Y = the cumulative average time (or cost) per unit.

The learning curve can be used to estimate production times for new parts, based on the planned production quantity. This page gives typical values for the learning curve parameter. For example, repetitive welding operations have a 90% learning curve.