.

Thinking Like A Human Factors Expert

Marc Green


The rat is always right- B. F. Skinner

It has become increasing clear over my years in human factors that few people understand exactly what human factors is. This includes nonexperts, but also the majority of people who try to pass themselves off as human factors experts. Human factors does not exist so that people can appear in court and testify about accidents. Human factors exists to design a better world. A person who has never done human factors design can no more be a human factors expert than a person who has never done surgery can be a surgery expert. Stating that standard perception-reaction time is 1.5 seconds or parroting some rule or regulation from a handbook does not a human factors expert make. Here, I describe the fundamental nature of human factors and explain how a human factors expert ought to think, both in court and out.

A human factors expert must of course have extensive knowledge about perception, cognition, response and about human abilities, limitations and predispositions. Perhaps more importantly, however, a human factors expert must see the world in a manner best summarized by Skinner's Law, which I first heard early in my graduate school days 40 years ago. Over the years, I've increasingly learned to appreciate the profundity of its insight. Its message is the fundamental underpinning of the human factors profession. If you don't get it, then you don't get human factors. If you still don't get it after reading this article, try looking at "The Psychology Of Everyday Things" (Norman, 1988).

B. F. Skinner, the great behavioral psychologist, originally meant his statement to guide his fellow experimental researchers, especially those who studied learning in animals. He was saying that if you perform an experiment, and the rat does not do what you want him to do, it isn't because the rat was stupid, lazy, or inattentive. It was because you set the experiment up incorrectly. Rats can only be rats. It's your fault, so don't blame the rat. (Or as another fundamental principle, The Duke's Dictum, says, "A rat's gotta do what a rat's gotta do").

Behind Skinner's Law lies a deeper worldview, namely that human behavior is a natural phenomenon like gravity or light and that understanding people is an empirical endeavor. This is why a real human factors expert incessantly reads research on human abilities. In contrast, the engineering (and often legal) view is that people "ought to" or "should" behave in some prespecified, idealized way. It doesn't matter whether or not it is how real people are likely to behave or whether it is consistent with human nature and our evolved survival skills. The engineering worldview is that there is automatically something wrong with the rat if he fails to act as we want him to act, and rat nature be damned.

What has this got to do with human factors? The answer is everything. The goal of human factors is to use scientific knowledge about perception, cognition and response to design artifacts and environments that are easy, highly intuitive, errorless and safe to use. If you design something and many users find it difficult, unintuitive, error prone and unsafe, it is not because those contrary users are stupid, lazy or inattentive. It is because of your misdesign. The rat is always right.

Often, however, designers and companies automatically blame users when accidents occur. They say that the product would be perfectly safe if people would use it as intended and follow the instructions. If the users don't and are injured, then it's their problem. The rat is always wrong if he doesn't do what you want. (This is not hyperbole. A representative of the Federal Railroad Administration once said to me that anytime a train collides with a car at a crossing, it is automatically the car driver's fault.)

This attitude reflects naiveté and wishful thinking about human behavior, and is completely unrealistic. If normal users do not always use the product as the company intends and suffer injuries, then q.e.d. there is likely something wrong with the product. It or its instructions (or training) are faulty, too difficult to use, offer unintended affordances or run contrary to some other aspect of human nature. In any event, the product must be fixed or, in the extreme, the product must be withdrawn from the market. It doesn't matter whether users failed to follow the instructions, or to use it as intended. Harm is being done. The purpose of human factors design is to make an intuitive, safe product or environment and not to assign blame when something goes wrong. Human factors adapts to human nature. It does not make the completely unrealistic expectation that human nature will adapt to the designer's notions of how people ought to behave. In human factors, we call this the principle of "user centered design". We design around the user. We don't demand that the user behave in some unnatural way because it is convenient (i.e., cheaper and more lucrative for us) if he does. This isn't just because human factors professionals are especially public spirited. We just know that it is futile to do otherwise. We are not going to change human nature any time soon.

I was recently involved a case that perfectly highlights Skinner's Law. A person was riding on the back of an all terrain vehicle (ATV) going up a hill and fell off, suffering an injury. The ATV was almost plastered with numerous stickers saying not to carry passengers on the back. The same warnings were prominently featured in the instruction manual. It took little investigation, however, to discover that the warning is largely ignored or unnoticed by ATV owners and that carrying passengers is routine. Despite the large number of stickers, many owners never even noticed the warnings because it was such routine behavior. Moreover, some companies actually advertise and sell seats to be attached to the back of ATVs for carrying passengers. In this case, the seat was sold as an extra by the same retailer who sold the ATV. The passenger was riding in such a seat when the mishap occurred.

Of course, the ATV company blamed the driver and passenger for the injury, saying that they ignored the warnings. However, the company was aware that the warnings were largely ignored. The seat manufacturer made the same claim because the manual said that it was to be used only when stopped! No, you read that correctly: the company made a seat for the back of an ATV and then told the user that it should only be used when the vehicle is stopped. Naturally, the seat company blamed the user for ignoring the warnings and riding in the seat while the ATV was moving.

This is not realistic thinking and it is certainly not thinking like a human factors expert. If people ignore the warnings, and ignoring the warnings leads to injuries, then something must be done. The goal of design is safety, not to assign blame. Moreover, it was highly predictable that users would ignore the warning. People routinely ride as passengers on other vehicles like motorcycles. ATVs are commonly fitted with long seats that are big enough to carry two people. In fact, they offer an affordance that suggests carrying a passenger. The seat attached to the ATV was even a clearer affordance. As explained elsewhere, moreover, warnings are generally not very effective and people will inevitably use affordances provided to them.

The companies' defense was a common one - they cannot control what people do after they have purchased the produce. This is completely false. What they mean is that they have downloaded responsibility for safety on to the user. The user, not the company is supposed to provide the safety. In other words, the product is perfectly fine until the user mucks it up. It is true that a company relying on warnings exerts minimal control over user behavior, but that is why warnings are such a weak safety mechanism and should not be relied upon.

In fact, the ATV company could control user behavior. The solution to the problem would be obvious to anyone familiar with the safety hierarchy - change the design to make some behaviors difficult or impossible. In this case, it means changing the design to physically prevent people from riding on the back. This could be done by shortening the seat, molding it to a shape that fits only one rider, putting a structure immediately behind the seat that prevented sitting, attachment of third party seats, etc. The seat that fitted on back must be withdrawn from the market or some physical method must be devised for preventing its use in motion. The more cynical side of me says that the companies never took any real steps to actively discourage the carrying of passengers on a moving ATV because it would decrease sales since people like riding together.

Finally, I have one caveat. Skinner's Law should not be taken to its limit. There are certainly bad rats. Rats who come from broken rat homes, join rat gangs, sell rat drugs, etc. Some rats can and do act outside the norm of rat behavior. Skinner's Law applies to the normal, typical rat. When a number of rats act in a certain way, however, then we are not talking about aberrant bad rats. We are talking about predictable and foreseeable rat behavior. If you find many rats riding on the back of ATVs, then the rats are not the problem.