Thursday, October 6, 2011

Levels of Reasoning in Goal Setting

The setting and reaching of a goal is most often pictured as a deliberately planned and executed act. But in reality this is extremely rare. Most often we can see things around us happening that are just following the flow, without any special considerations or thoughtful actions. That creates a disconnection between goals and everyday life. In C2/GOMA we are trying to automate the most common everyday situations using a goal-oriented approach. In order to do that, we have to bridge that gap somehow. We need to change the way we think about goals and accept the fact that goals may not always be deliberately planned. Jens Rasmussen came up with a model called “3 levels of human behavior/reasoning,” which gives us a totally new approach to goal setting.

In the paper “Skills, Rules and Knowledge; Signals, Signs, and Symbols, and Other Distinctions in Human Performance Models”, Jens Rasmussen writes: “The introduction of information technology based on digital computers for the design of human-machine interface systems has led to a requirement for consistent models of human performance in routine task environments and during unfamiliar task conditions. A discussion is presented of the requirement for different types of models for representing performance at the skill-, rule-, and knowledge-based levels...”.

Further down he defines 3 levels of human behaviors:

1. Skill-Based: “The skill-based behavior represents sensory-motor performance during acts or activities, which, following a statement of an intention, take place without conscious control as smooth, automated, and highly integrated patterns of behavior.”

2. Rule-Based: “At the next level of rule-based behavior, the composition of such a sequence of subroutines in a familiar work situation is typically controlled by a stored rule or procedure which may have been derived empirically during previous occasions, communicated from other persons’ know-how as instruction, or a cook-book recipe, or it may be prepared on occasion by conscious problem solving and planning.”

3. Knowledge-Based: “During unfamiliar situations, faced with an environment for which no know-how or rules for control are available from previous encounters, the control of performance must move to a higher conceptual level, in which performance is goal-controlled and knowledge-based. In this situation, the goal is explicitly formulated, based on an analysis of the environment  and the overall aims of the person. Then a useful plan is developed...”

You shall remember that C2/GOMA defines goal as “a future state of affairs that person or system plans or intends to achieve.” But the definition does not explicitly say how that state is set and achieved. It could be deliberately planned, intuitively figured and carried out, or, even, could be a preset function.

Following Rasmussen’s reasoning loosely, we can apply his 3 levels of human behavior to Goal Setting:

  1. Knowledge-based Level. For new and unfamiliar goals, or goals approached in a new way, the person will use his or her knowledge and creativity to come up with a unique plan to achieve them.
    • Examples: researching and deciding on which school to choose for one’s child, acting at a new job position for the first few days, living in a new unfamiliar country, or planning to rob a bank for the first time.
  2. Rule-based Level. For familiar goals, which a person has had previous experience with or learned about from available sources, a pattern-based (or template/rule-based) plan is used. He or she may apply minor tweaks to make that plan a better fit to the current situation.
    • Examples: interacting with a cashier at a store, cooking breakfast, or following a known business-process at work.
  3. Skill-based Level. For goals that are habitual, or anticipated by human nature, a person may act automatically without any conscious effort.
    • Examples: walking down the stairs, saying “hello” to a person who says “hello” to you, driving a car, or pretending to do your work.

As you can see, most of our goals (or future states) are set and achieved at an unconscious, skill-based level. When we set a goal to walk to a nearby place (future state: take a certain position in space) we don’t plan every step, our body simply acts automatically. When we need to type something on a computer (future state: enter specific text) our fingers just press buttons by themselves. We don’t think about it much.

Sometimes, we may use our brain to deal with minor changes in a situation to tweak patterns we acquired over our lives, picked up from family members, learned at school, or watched on TV. The human brain is a perfect pattern-based machine. Its associative memory can store an enormous amount of different patterns and quickly find something, even if it’s loosely related and based on some weird association. In this regard, there is no computer available yet that can match a human’s brain. Even when meeting something new, a person will most likely find something similar in his or her past experience and will quickly apply it to the new situation.

To tell the truth, it is actually very rare that we sit and sweat and try to invent something totally new. It is rare that we try to take on unfamiliar situations and goals. Most of us don’t like that. Working at the knowledge-based level is an area for weird creative individuals. Most of us prefer to stop or walk away from a situation when there is no suitable pattern found.

The majority of our goals are set and carried out at the rule and skill-based levels, and only a few--many times the most important--are performed at the knowledge-based level.

There is one more aspect I want to mention here. C2/GOMA doesn’t make a strong distinction between people and automated systems. That’s one of the strengths of the theory--a person can be replaced by a computerized system or robot and nothing will really change. The management model will stay the same.

How is this possible?

Actually, it’s quite simple. Consider the following: any automated controller or computer program works deterministically. It generates a predictable response to given inputs. Thus, if the goals (future states) of an automated system are preset, and plans to achieve those goals are hardwired into the system design, the system acts at a skill-based reasoning level much like a human would.

Some automated systems, powered by Artificial Intelligence, can act nondeterministically, or may alter their behavior based on achieved results. For instance, rule-based expert systems use predefined rules to generate responses, neural networks are trained on known datasets to be able to recognize familiar inputs, and statistical learning systems are able to process historical information collected in databases and then discover statistical patterns. These methods are still not as sophisticated as the human brain. But they are already able to achieve quite a lot... You’re absolutely right, if you recognize reasoning in these systems at a rule-based level.

But what about the knowledge-based level of reasoning? Most people don’t want and don’t like to operate at the knowledge-based level, and, so far, automated systems are not able to reach that level yet. Therefore, even though C2/GOMA is trying to automate as much as possible, there are still areas with new unfamiliar goals where creative people cannot be replaced with technology. However, considering the rapidly expanding knowledge of our information age, such creative and special areas are shrinking quickly. The three levels of reasoning and goal achievement may soon all be carried out by management automation systems.





4 comments: