Is Experience Really the Best Teacher? A look at Recognition-Primed Decision (RPD) Making

Let’s imagine a group of firefighters attending to an emergency. In a split second, they have to make decisions about whether to organize search and rescue, initiate offensive or defensive measures, or how to manage their water resources. Every second matters, particularly in that moment, and one simply doesn’t have the luxury of time to make choices based on lengthy statistical analyses or cost-benefit analyses. In those time-sensitive situations, people rely on experience to support their decision making. They rely on their ability to quickly recognize situations, analyze them, and develop a course of action. The Recognition-Primed Decision (RPD) model explains this decision-making strategy well.

The RPD model has three steps. It begins when the decision maker experiences a situation. The decision maker then recognizes certain variables which informs the generation of plausible courses of action. The generated courses of action are evaluated by the decision maker’s mental simulation of the outcome of each action. Given time constraints and external pressures, a satisficing action is implemented in the third stage. Typically, the first action which to be validated by the mental simulation is implemented.

This type of decision making strategy is also called recognitional decision making. Without prior experience to a similar event, the decision maker is limited as far as recognition is concerned. During the recognition phase of the model, a decision maker contemplates plausible goals, expectations, while looking out for relevant cues and actors. We therefore expect experts and novices to differ in their response to similar situations. Experts generate plausible courses of action quicker than novices because of the recognition phase. Data analyzed by Klein suggests that compared to analytical strategies, recognitional strategies are less frequent among less-experienced decision makers [1].

RPD strategy has several factors that make it different from formal and normative decision-making methods. Three major differences lie in the desired outcome, order of evaluation, and delay of actions.

First, the desired outcome in a recognitional strategy is not the best solution. Like other descriptive methods of decision-making, the final decision is a satisficing one. Here, the first accepted action will be implemented by the decision maker. The reason is simple, every second counts. Second, the order of evaluations used in recognitional strategy is a serial evaluation rather than a concurrent evaluation of several methods. You can imagine the firefighters in our opening paragraph evaluating their actions as they generate them, rather than generate them and then concurrently analyze each action. No brainstorming sessions are organized since time is of critical importance. Third, there is little to no delay between when decisions are made and when action is taken. As such the decision maker is always ready to act in the given situation.

The RPD model is employed when time is in short supply, situations are not clearly defined, and feasible actions must be taken swiftly. In so many ways, this is exactly what many real-world problems tend to be -ill-defined situations that require immediate action. For such problems, analytical methods are considered to have a downside. They are generally more time-consuming; and they are unlikely to be applied by people in less than a minute [1].

But what about machines? They have come such a long way since the RPD model was developed. In 2016, DeepMind’s AlphaGo defeated the world champion, Lee Sedol in a game of Go. Could machines not suggest better courses of action given that they could sift through vast amounts of data in short amounts of time?

Following a machine learning analogy, and to an extent, the very concept of recognitional decision making, it also means that we can train people to recognize situations even before they happen. Training people with scenarios and role play, project-based learning, and learning by case studies may be potent examples of how to achieve this.

Just as with any other cognitive model, knowing when to make use of it is perhaps the most important. It is essential to maintain a balance between the importance of experience and the nature of the situation. Knowing when to draw on your experiences will prove useful. In certain situations, we must throw our experiences out the window and observe problems with a fresh pair of eyes. In those times, not only will a recognitional strategy impede us from generating feasible courses of action, it will also slow us down.

In the end, the key is to stock up on several tools for decision making and know when to pull out which tool. As the saying goes, when all you have is a hammer, every “problem” you see is a nail. The firefighters from our example employed the recognitional strategy in handling the emergency; but when one of them decides to invest in her first stock, she will most likely use a different approach.

Notes

[1] G. A. Klein, “A recognition-primed decision (RPD) model of rapid decision making,” Decis. Mak. Action Model. Methods, pp. 138–147, 1993.