We are not random Brownian particles.  At any given decision point, we do not truly have freedom to choose.  Instead, our policies form constraints that guide and inform our decisions.  Explicitly, a policy is a set of rules.  Implicitly, that set of rules describes past decisions.  Like legal or medical precedence, our past decisions determine our current policies that in turn determine our future decisions.  But this makes for circular logic.  So how does one make the decisions?

There are any number of texts and seminars on how to make decisions.  One of the better examples is here.

The process divides into steps:

1.decide on the goal

2.collect relevant evidence, i.e. project the outcomes

3.weigh potential projected outcomes, i.e. use gut intuition

4.decide

5.perform a post evaluation

This is sound advice.  But step 1 requires a decision.  Step 4 requires a decision.  Modern Computer Science Data Mining can help collate evidence data for step 2 - but without clear communication on the goal from step 1, often misses the essential keyword, "relevant."   And step 3 transfers the responsibility for the decision to an unknown, ethereal "sense" of intuition.  Thus, a decision still relies on making decisions.  It still begs the question: how does one make a decision in the first place?

To answer this, one must go to the beginning.  How do young learners make decisions?  The following is an excerpted interview with a kindergartener on deciding what is and is not lunch.

Experimenter: "It's twelve o'clock in the afternoon and the sun is shining.  You already ate something today, but you're still very hungry, so you decide to eat pancakes with syrup, orange juice, cereal, and milk.  Could that be lunch?"

Child: "No."

E: "Can you have pancakes for lunch?"

C: "No… no."

E: "Well, how do you know if something is lunch or not?"

C: "If the time says 12:00."

E: "This was 12:00."

C: "Well I don't think so."

E: (Repeats story) "Is that lunch?"

C: "I know… that one is not lunch… you have to eat sandwiches at lunch."

E: "Can you have anything else?"

C: "You can have drinks, but not breakfast."

(Keil, 1989)

Is this lunch?  Is that a good investment?  Should I stop or should I go or pull to the curb?  Was that a polite greeting or a rude one?  What do we tell the staff?  The question is not how to make a decision.  It is how to decide how to make a decision.  The question is why we make the decision. 

The two main answers to this are definition logic and probabilistic associations. 

Definition logic refers to building a decision concept based on a canonically correct answer.  Lunch is by definition any meal consumed on or about 12:00 noon.  A good investment by definition generates a positive net present value return (or generates an internal rate of return above the required rate).  Green light and all clear means go, red light or any obstruction means stop. 

Inhelder and Piaget (1964) explored when this definition logic manifests in young children.  They asked several children to classify or group together several objects from among stuffed animals, toy vehicles, and furniture.  The older school children, aged 7+ grouped the objects via definition.  That is, they grouped animals with animals, toys with toys, and vehicles with vehicles.  For younger, preschool children aged 3-6, it was a different matter entirely.  They grouped dogs with cars, cats with chairs, and toys with the shelf.  This violates all manner of definitions logic.  When asked, the younger children explained that dogs like to ride the car, cats like to curl up on chairs, and games belong on the shelf when done.

Bauer and Mandler (1989a) also explored definition logic among very young infants (1-year olds).  The study paradigms were much simpler, however.  First, they showed infants a toy monkey.  Then they asked the infants to find another object just like that toy monkey.  The two choices were a banana and a bear.  85% of the infants chose the bear.

This raises an anomaly.  Why can 1-year olds and 7-year olds exhibit decision making by definition logic but 3-6 year olds cannot?  Smiley and Brown (1979) and Blaye and Bonthoux (2001) explored this anomaly.  Their findings indicated that 3-5 year olds can decide by definition logic if explicitly asked to consider their choices in this manner.  Cole and Scribner (1974) found the same pattern of decision behavior among African tribespeople.  The tribespeople would make decisions using non-defining logic until asked specifically, "Now how would a stupid person do it?"  Whereupon they shifted towards more definition logical decisions.  Apparently, people have a choice on how to choose decisions.  Perhaps we people prefer to use approaches that we find more rewarding - which means at certain stages in life, we prefer the more amusing, entertaining, or just plain more interesting viewpoints.   

An alternative decision making answer follows probabilistic associations.  Lunch is normally sandwiches and juice at 12:00 noon, but likely does not include breakfast foods.  A good investment is one that everyone else also decides to invest in - which causes the economies of scale, enhancement to productivity, profits, and rises in asset prices that generates the positive return.  Politeness is consistent with socio-cultural norms, whereas rudeness is not.  We tell the staff the things we would have liked to have heard.  Probabilistic associations allow for status quo decisions that expand or contract concepts and boundaries with experiences. 

Rosch and Mervis (1975) explore how cue validity can account for strengthening these associations towards one or another decision concept.  Similar to the well known term frequency - inverse document frequency encountered in basic text processing, it basically accounts for how often a cue associates with a given particular concept and nothing else. 

Rosch, et al. (1976) applied an hierarchical framework to decision making.  The goal was to attempt tying both logical and associative decision approaches.  Chairs fall in an hierarchy, with furniture as the superordinate category and kitchen chairs being a subordinate category.  The theory is that children first learn the basic chair category, then use spreading associations to build up and down the hierarchy by assigning chairs to their place below furniture but above kitchen chairs.  When 3-9 month old infants habituated - that is, got used to - grouping together images of various horses, the test was to see which would be more surprising: a giraffe or another horse. 

Colombo, et al. (1987), Quinn, Eimas, and Rosenkrantz (1993), and Roberts (1988) explored the flexibility in hierarchies further.  They exposed infants to images of various horses as before.  The ordering and timing of the initial images showed the infants that they were all of the same theme.  When the infants subsequently observed novel images, they dis-habituated - that is, showed surprise signifying a different category - upon seeing zebras, giraffes, or cats, but not while seeing novel horses.  They formed the appropriate horse basic category. 

The fascinating finding comes when the teams exposed infants to various images of mammals under the same paradigm.  Again, the ordering and timing of the initial images showed the infants that they were of the same theme, what adults would call mammals.  The infants subsequently dis-habituated upon observing birds, fish, or furniture, but not while observing novel mammals.   They formed the appropriate superordinate category.  It appears that infants and young learners at least can form super, basic, or sub ordinate categories at the will and whim of the adults setting up the initial exposures.  Once the child gets the key point or the specific purpose of the objects - as portrayed by the adults - they make decisions on them in a similar manner as adults. 

Implication: technically, repeating the process such that the infants get used to seeing horses, mammals, fish, and furniture forces them to form another super-ordinate category including all the above.  While there would be no published textbook logical definition linking all of these in any sane world, the infant might decide to group these as akin to, "objects encountered in a museum lab test setting" (or wherever they are during the experiment). 

Therein lies the key decision of all - which approach best describes how we make the decisions that form the policies that guide our decisions?  This is the top of the hierarchy.  Do we have a defining logic first that points our way second?  Or do we point the way first and explain it with an artificial compressed descriptor of a logic second? 

In the first,  we all have the inherent logic to make good decisions if only we would listen to it (i.e. Kahnemann's Thinking Fast and Slow).  In the second, we all have the inherent connections with family, neighbors, colleagues, and environment to fit any decision into patterns of behavior. 

In the first, our language logic dictates that we look up the textbook definition of lunch.  In the second, our language association has us eat the meal and then write the textbook definition to describe it. 

Decision making at its root hierarchy depends on the eye of the beholder.  If we view ourselves as children in a big, wide world, then our decision logic is imposed upon us.  We follow it by definition and by the book.  We follow the adults.  We take the test.  If we view ourselves as adults in a small world, then we impose our decision values.  We make the definitions.  We give the test.  Lunch is whatever we wish it to be.

To answer the original questions:

Q: "It's twelve o'clock in the afternoon and the sun is shining.  You already ate something today, but you're still very hungry, so you decide to eat pancakes with syrup, orange juice, cereal, and milk.  Could that be lunch?"

A: "It is if you can understand and accept that I had pancakes at lunch."

Q: "How do you develop a coherent policy that guides our decisions?"

A: "Our policy describes and summarizes our past decisions."

Q: "Then how do you make the decisions in the first place, without a policy?"

A: "We decide in accordance with getting the other person or the environment to do what we want.  The only boundaries are those that follow and describe our decisions.  A decision is nothing more than a form of communication.  As is the policy that follows and describes them."