"In a hotel restaurant, when the cheque came, I signed my name to it but couldn’t remember my room number. So I looked at my watch."

This example illustrates a typical slip of action among more than 200 that I have collected in two years. Slips are amusing and happen to all of us; except when they lead to embarrassment, they usually seem like harmless oddities.

But slips that occur in the conduct of certain tasks can be dangerous. An air-traffic controller once told a plane to taxi to the left runway when he meant the right. Such a left/right slip is the most common of verbal confusions, but in this case it could have led to tragedy (fortunately, it didn’t). Forgetting to turn on headlights when driving a car at night is also common. The slip is usually caught before an accident occurs, but not always. Today, human error is one of the largest causes of accidents.

The founder of psychoanalysis, Sigmund Freud, analysed errors primarily to discover a person’s true beliefs or intentions. One case he examined involved the president of the Lower House of the Austrian parliament, who opened a meeting by shouting, "I declare this meeting closed!" That act does seem to reveal hidden motives; as Freud noted, "The president secretly wished he was already in a position to close the sitting, from which little good was to be expected."

But I believe that the human mind is an exceedingly complex computer, and that slips can also occur when stray information throws off human information-processing systems. During Freud’s era, the science of computation, information processing and control was primitive, and Freud was not led to think about the nature of a human processing system that could contribute to errors. I interpret Freud in modern terms as saying that slips result from competition and intermixing among underlying mental-processing mechanisms, often working parallel to one another.

What about the pilot who lands a plane with the wheels still up—does he have a hidden wish to kill himself? It’s possible, but simpler explanations are possible, too.

Most actions, I propose, are carried out by subconscious mechanisms. We 'will' an action. The intention, once specified, releases control processes, or "scheme," that lead to the exquisitely timed, complex motor actions involved in manipulation of mind and body. When I drive home from work, the appropriate schemes are activated by previous actions. I need not plan the details; I simply decide and act.

Do I wish to detour to the fish store? I must have "fish store" actively in mind at the time I pass the critical choice point between work and home. Let it lapse from my memory at the critical junction, and I am apt to find myself at home shortly thereafter, fishless.

In different slips, different parts of the information-processing sequence go awry. For purposes of convenience, I categorise them according to which part of the human machinery is involved. As an example of a description error, consider the following:

A chartered airliner flying from Houston to Montreal crashed exactly on the border between the United States and Canada. A major political issue developed over the following question: In which country should the survivors be buried?

Most people puzzle over the choice of country. In fact, the story was made up to trap the unwary: it is the dead who should be buried, not the survivors. The story sets us up to be lazy mentally—to except the critical leftovers to be the dead ones—so we don’t process the word "survivors" deeply enough.

In day-to-day activities, selection errors are similarly common. For instance:

In getting ready for a party, one person carefully prepared a cake and a salad, then put the cake in the refrigerator and the salad in the oven.

Such inadvertent ambiguity can also lead to eating your friend’s sandwich from a plate that looks like yours, or to putting the top of the sugar bowl on top of a coffee mug of the same size. One of the sugarbowl on top of a coffee mug of the same size. One of my graduate students returned to his home after a track workout, pulled off his sweaty T-shirt, and tossed it into the toilet. It was not an aiming error—the laundry basket, his intended target, was in another room.

Activation & Triggering Errors. Once an intention is selected, executing it can easily misfire. We can, for example, forget the initial intention while some of the schemes it ordered run their course. One colleague reported that before starting work at his desk at home he headed for his bedroom, only to realise after getting there that he’d forgotten why he had gone. "I kept going." he reported, "hoping that something in the bedroom would remind me." Nothing did. He finally went back to his desk, realised that his glasses were dirty and, with a great sense of relief, returned to the bedroom for the handkerchief he needed to wipe them.

Yet another category was analysed by William James in his classic textbook on psychology; he reported that "very absent-minded persons in going to their bedrooms to dress for dinner have been known to take off one garment after another and finally to get into bed, merely because that was the habitual issue of the first few movements when performed at a later hour."

Such errors, called capture errors, constitute one of the most fascinating categories in the field of error. They involve a simple principle: pass to near a well-formed habit and it will capture your behaviour. A colleague in England who studies errors reports that someone told him this story:

"I meant to get my car out, but as I passed through the back porch on my way to the garage I stopped to put on my Wellington boots and gardening jacket, as if to work in the garden.

Our errors are easy to detect, of course, when they have such obvious consequences. Sometimes the consequence materialises several minutes after the event: for example, when you plug in the coffee-pot in the morning and forget the coffee, you learn of your slip only when you start to pour.

The likelihood of human error increases when there is stress. Consider a situation in which innocent errors can have serious consequences—as, say in unclear-power plant. The total length of all control panels in these complex systems may be more than 100 feet, with some 700 instruments and controls that require testing or servicing from time to time.

The difficulty of controlling reactors is simply poor systems design, indeed, modern systems of all sorts are inconsiderate of human beings. In my research laboratory, we are attempting to understand the basic properties of human information processing, with emphasis on how skilled people select and guide their action—and make their slips. Although the basic thrust of the research is theoretical, we share an important subgoal, to give designers the guidelines for designing systems that work with people, not against them. We hope not to eliminate any disastrous effects of error, but also to make human interaction with machines pleasurable, efficient and creative.

Think of what a salesperson must do in a department store. Even for a simple purchase, a huge array of numbers must be entered into the computer cash register. The numbers are supposed to make the accounting system work, but the priorities are backward. Instead of forcing people to act like machines for the benefit of machines, why not make it the other way around? Machines can be made to do the translation into whatever hidden, laborious codes they require. The cash register should "know" automatically in which department it is located. The sales person would simply punch a button for "shirt"; if the computer needs to know more, it could be programmed to ask for specifics, which the salesperson could then provide by punching other buttons.