Random Thoughts

Decision Trees

The concept of the decision tree is something I have stolen from games theory. I will not detail how they are used in that area of mathematics, since I hope how I explain my use of the idea here will be clear enough on its own.

A decision tree is a way of breaking down a complex (or simple) problem, or decision process, into smaller manageable pieces that can be solved individually. With enough of them worked out, the entire complex can usually be resolved much more easily.

There are two things which sometimes occur which would difficult to discern without this process: one is that an insoluble part of the problem may turn out to be irrelevant to the outcome, and the other is that certain smaller parts of the complex may lead to the same results whichever way they play out.

Even if every aspect of a complex needs to be resolved prior to taking a clear final action, this process can help tremendously.

This is an ideal way to approach things in life which make your head spin. When a difficult choice or series of choices confront you, often you will find that various aspects of it whirl in and out of focus, and every thought you have about one leads inevitably to problems with the others. This can be very wasteful of emotional energy, and often result in no decision being made, or a poor one you will later regret.

Making clear decisions is very important to personal growth and satisfaction in life. It may be obvious that making the "right" decision does this, but even clearly made "wrong" decisions are very useful. As long as a decision was made with a clear understanding of the process, an undesired outcome can be related to how you made the decision, and perhaps you can determine what part of the process was a mistake - and learn from this when choices confront you in the future.

So, what is a decision tree?

In a decision tree, basically you approach a problem analytically. You break it down into the smallest possible parts, each one being its own fairly simple decision. Any part of the complex with more than two possible choices should be reduced to several binary choices - that is, ones with only two possible outcomes.

For instance, a list of three favorite colors for a purchase could be turned into a two tiered process. The first is between your number one choice and not getting it. The next, only required if the first is not satisfied, is between your other two preferences. If a further series of options must be considered, each step can be between the "next favorite" and it's unavailability.

When dealing with a complex decision, it may be that parts of the process depend on choices made in other parts. The tree is a way to make sure that every choice involved that conditions what other choices must be made eventually leads to it. In other words, if one choice leads to another choice, the second, derivative decision must be on a branch that comes from the prior one.

The next step is to separate emotional from rational considerations.

Certain parts of the tree will essentially say "check how this feels" in order to proceed. It is important not to consider the feelings that result to be the final condition of your decision. If at some point on the tree, for instance, an emotional input is required, and that input is highly pleasant or unpleasant, it only applies that that particular part of the tree. You must not say "oh, that feels good," and proceed to decide at that point to do whatever will get to that part of the tree in real life. This is because there maybe many other aspects of the tree which would make this a bad choice.

Another problem with making your decisions based on how some part of the tree "feels" is that you will not really be able to assess the decision making process later - all you did was "what felt good" (or avoided what "felt bad"), and contradicting your feelings is not a good way to learn how to make decisions. Of course, certain trees will lead to a point where one simple emotional decision is what everything all the way back to the root choice depends upon - and that is just fine, if you have done all the work alogn the way. It is how you determined that the choice was a simple emotional one that matters here.

One thing I often find myself doing when confronted with relatively simple, short term choices of how to act or behave, is to try to formulate a course of action which satisfies the tree in such a way that however the variables (often things out of my control, like others behavior or choices) play out, the result is a pleasant or desired one for me. Sometimes it is possible to make it so my action produces my desired outcome no matter what else happens, and sometimes it is a matter of two or more different things that are fine with me happening. Of course sometimes these "results" are just a matter of how I feel about a situation, so carefully thinking it through is what actually makes the possible outcomes fine with me!

The best place to start your tree is with the most obvious or largest aspect of the choice presented to you. While this may not end up being the "root" choice in the end, you have to start somewhere! Formulate in such a way that it is a straightforward, binary two option choice. Then try to consider the other main aspects of the decision, other smaller choices that are part of the problem, and how they relate, in logical terms, to the one you started with. If they do not naturally seem to follow only one of the main outcomes, it is possible they are prior to the chocie you started with.


  • Main choice (seemingly) = "What car should I buy?"
  • Other things in my mind:
    • What color should it be?
    • What can I afford?
    • Do I really need another car?
    • Will my friends like it?

Now, as you can see, the third "sub decision" should actually come before the others. It may present itself as a consideration in the process of deciding what car to buy, but if you answer the question "Do I really need another car?" with a "No," then all the rest of the process is a waste of time and energy, unless the fantasy of buying a car you don't need is a pleasant one in its own right.

Now, car buying is a great example, since it would seem that except for some small issues like color, it should be a pretty rational process, but in reality the way people buy cars is incredibly emotional - even to the point of ignoring whether they need them or can afford them!

But car buying is boring and trivial in life. They are just machines, and cannot love you back!

My intent in outlining this process is to aid, if even some tiny amount, in the making of decisions that actually matter in life, choices we make about how to treat other people, what to share and when, and when to do that most difficult thing - do what we do not want to do at first, in order that life will hold greater joys and possibilities for us in the future.

You must also remember to include external influences in your tree. If for instance, the only "way" to get to your desired outcome involves another person's decision, which will be an unknown at this point, you should consoder how much of a problem it will be for you if they do not decide in your favor. There may be another set of branches that actually produces better outcomes - perhaps two quite favorable results rather than the risk of a very favorable one possibly ruined and made unfavorable by something not under your control.

So you've tried to think through your problem, and identify various aspects of its dependencies (these are how the various choices relate to each other - a choice is dependent on another if it is on branch leading out from it), and where the emotional inputs come in. You know perhaps what sort of outcome you want. Do you find that outcome to be a likely and reasonable result at the end of any of the series of branchings? If you do, then work backwards from that point, and when you get to the first choice to be made, make it so you are led onto the branch with your favorable outcome on it.

There maybe other requirements of you along the way - but if you have thought it through carefully, you will already know what to do when you confront them. Further choices to be made, or tasks to be accomplished in order to continue the processes of life along your branches of choice until your desired outcome is in place.

This may sound incredibly sneaky, controllig, or even Machiavellian, but it is not. It is just a matter of being clear about when you have choices to make in life and have preferences as to the outcome. You will be happier if you make them well and understand how you did it.

Now what if in the course of defining your tree, you come across many unpleasant (emotionally negative) places? What if they block all the reasonable branching paths to the outcome you want? You can do one of two things (well, three, actually!). You can: either suffer the unpleasantness along the way, knowing that your eventual situation will be favorable (this is called "delayed gratification"); identify another outcome, perhaps less wonderful but still acceptable, that does not require such pain along the way (this can run the risk of what is called "copping out"); or you can attempt to master your emotional reactions in such a way that the unpleasant phases of the process are either not so unpleasant or perhaps not even nasty at all.

If you can pull off the third process listed above well, you will not only have made a careful decision with probably pleasant consequences, but you will have come to know your emotional self better, and probably matured a bit in the process!

2/15/01, bare bones. Fleshed out, 12/11/01

typos? comments? mail me here

© Huw Powell

Printer-friendly version - (no indent)