• The Nature of Decisions
  • Decision-makers' Biases
  • What is a "Good" Decision?
  • The C2 Decision Circle
  • Satisficing
  • Decision Paradigm Clash
  • Decisions and Synthetic Environments
  • Decision-making - Man or Machine?

Understanding how military groups make decisions is obviously important; it can cost lives to one side or the other, or both. As such, it is a legitimate and indeed important subject for study.

Surprisingly, how we humans make decisions is remarkably poorly understood. Exploring how the military approach the issue in relatively well-defined situations which they have, in the past, faced might help us understand how things work - or don't work! And, perhaps, we can apply any new knowledge to less extreme situations! A sort of intellectual "swords into plough shares"!

The Nature of Decisions


"...a choice between options..."

  • Utility Theory proposes that we make decisions to maximize utility or usefulness
  • Statistical, Catastrophe and Chaos Theories indicate future may be predictable/unpredictable as weather
    • long-term stable but short-term unstable, according to climate
  • How can we choose between options in dynamic situations if the future is unknowable?
  • Can we/should we avoid making decisions if the future is unknowable?

How do we go about making decisions anyway?

Decision-makers' Biases

It seems that humans, complex as ever, are prey to a wide variety of behavioural influences and pressures when they make decisions. Ian White has identified some of the biases which affect decisions:-

  • Adjustment & Anchoring. Decision Maker selects a Datum and fits other data to it improperly
  • Availability. Uses only freely-available data. An event is believed to occur frequently if it is easy to recall similar events
  • Conservatism. Failure to revise estimates as frequently as necessary
  • Data Saturation. Reaching premature decisions on too small a sample and then ignoring further data
  • Self-fulfilling Prophecy. Values certain outcomes and acquires and analyses only data that supports that outcome
  • Attribution Error. Associates success with inherent personal ability and failure with bad luck. "When you are wrong, you screwed up, when I'm wrong it was just bad luck"
  • Gambler's Fallacy. Assumes the occurrence of one set of events enhances the probability of an event that has not yet occurred. "I have smoked for 10 years without getting cancer&endash;clearly I am immune, so I can go on smoking
  • Habit. Familiarity with one rule results in its excessive use
  • Law of Small Numbers. Confidence in predictions based on small samples with non-discomforting evidence, than in predictions based on large samples with discomforting evidence
  • Order Effects. Order of information presentation affects retention and weighting
  • Outcome Irrelevant Learning. Use of an inferior rule leads to belief in results because of inability to evaluate choices not selected
  • Panic. Under stress, facing many options which cannot be evaluated, either selects at random or fails to act at all

What is a "Good" Decision?

The Causal Loop Model above shows the dilemma of judging whether a decision can be deemed good or not. It really all depends on your viewpoint. You can take either of two viewpoints:-

  1. A decision is "good" if the decision maker has taken account of all the relevant data, has weighed up the pros and cons, and has developed a balanced judgment
  2. A decision is "good" of the out-turn is that which the decision-maker intended

For instance, is a good gun one made of the best metals, finely made, with a well machined bore and a smooth, well-oiled breech?


Is a good gun one which puts the shell on the target?

Put more succinctly, is the quality of a decision in the process or in the outcome?

For my part, I have to fall down on the side of outcome. However, the current paradigm in Command & Control thinking centres on process.

The C2 Decision Circle

Classically, decisions are reached by a repeated sequential process, shown above. Start at Assess Situation, and follow the arrows. Since the Action taken changes the situation, Assess Situation is necessary after the change, and so the process continues ad infinitum.

Another way of looking at this process, one which omits the cyclic nature, is shown above. This model is employed in:-

  • The Estimate-a formal decision-making process in wide use by various militaries around the world
  • Systems Engineering
  • Current C2 Systems Technology


The process shown above has a distinguishing characteristic. The decision-maker(s) try to consider all the possible options, and to trade-off between them in order to find the preferred option. While this might seem logical, it is not necessarily how human's evolved to make decisions under pressure.

For instance, who considers all the options when the tiger is only five steps behind you?

We evolved an innately-human ability to make fast decisions under fight-or-flight pressures. If we had not, we would not have evolved as a species. Under such conditions, we do not consider and weigh up all the options - that would be naive. Instead, we use our experience of what has worked before in similar situations. We might mentally generate options, but we will stop doing so as soon as we identify and option that will serve the purpose. At that point we will go with the selected option.

This process is called "satisficing".

The figure shows satisficing. Note the iteration indicated by the feedback from "initiate" to "recognize situation cues". The initial satisficing decision might have been wide of the mark, even for an expert decision-maker - he or she may have read the cues wrong. However, this is a cyclic process. As the action unfolds, the decision-maker has expectations which can be verified by the merging cues. If the cues are as expected, fine. If, on the other hand, the cues are out of line with expectations, then the expert decision-maker will reevaluate the situation and satisfice again.

So, satisficing is a decision-making process employed instinctively by an expert under time pressure, and it "homes in" on a final solution (as opposed to making a "big-bang" "this is the only way" choice and sticking to it in the face of mounting contrary evidence).

Decision Paradigm Clash

On occasions, these two styles of decision-making can come into conflict, as shown above.

  • At left, naive decision-making might be undertaken by a group of staff officers in an HQ, and they might take anything up to 24 hours or more to go round the circle - referred to as the planning cycle.
  • At right, the "boss" is briefed, sometimes using CCTV, so placing him or her under psychological pressure to make on-the-spot decisions in full view of many subordinates. The boss is, therefore, likely to "satisfice" or employ Recognition-primed Decision-making as Gary Klein calls it.
  • The result?
    • disagreement, of course
    • the boss, using his experience may reject the advice of the staffs and select a particular course of action. the boss probably also decides that his staff are not doing too well.
    • the staff, having spent 24 intensive hours coming up with a recommendation, feel rejected and undervalued
  • These military situations would find their counterparts in boardrooms around the country, with MDs and CEOs corresponding to bosses

Decisions and Synthetic Environments (SE)

  • Great potential
    • until now, C2 systems designed by engineers to meet what they believe are operators needs
    • operators not sure what they need, unable to describe it effectively in writing, engineers have different viewpoint...
  • C2 technology based on Naive Decision-Making Model
    • slow, methodical, comprehensive, predictable, uninspired
    • each process supported by technology as though it were alone
    • i.e. no concept of system, team or learning in the design
  • SE/VR. (Virtual Reality)
    • New concept.
    • Engineers create SE within which operators may form teams, employ synthetic "technology", tackle variety of situations
  • Result. Operators:-
    • a) design own, team-based systems
    • b) develop/train cohesive operational team

Decision-making - Man or Machine?

  • For set-pieces, prescribed decisions are feasible:-
    • area air defence, known incoming weapon systems
    • saturation bombing, non-retaliating enemy
  • In volatile situations, short time horizons of predictability,
    • chain from cause to effect unclear
    • less amenable to machine decision
  • In the following 3 examples, all derived from, or based on, newspaper accounts of Desert Storm, are decisions best suited to man or machine (i.e. software algorithm)?

This diagram shows how different parties to a conflict might represent the same potential target. So, a military target of Roads and Bridges might be intended to deny the enemy freedom of action and/or weapon replenishment. Enemy propaganda could easily, even perhaps justifiably, represent attacks on Roads and Bridges as an attack on refugees, bullied innocents who would be unable to flee the conflict. With CNN listening for every opportunity, that sort of propaganda can wreak havoc with military strategies and tactics.

This figure shows how military planners might try to reconcile the issue of politically, or environmentally, sensitive sites which are also military targets. In particular the approach might consider the employment of precision weapons with little or no collateral damage risk.

Finally, in this group of three, we have a classic decision tree. This one compares the value of standoff Precision Guided Munitions (PGMs) i.e. laser-guided bombs launched at a distance, with old-style iron bombs delivered from close to the target. The comparison starts with establishing a comparative outcome measure in the right hand column of boxes.

  • For instance, the worst outcome, -100, arises when a PGM equipped aircraft is lost having failed to reach the target. This is a total lose-lose of an expensive asset.
  • The next worst outcome, -95 is for an iron bomb aircraft lost to enemy fire. This is also lose-lose, but of a less expensive asset - hence the slightly better score
  • The top score, 50, is for a successful attack by an iron-bomb - win-win using the cheapest weapon

In this simple example, for which I made up the numbers of course, the outcome is strongly in favour of standoff PGMs - surprise! (What use is a complex method of 'deciding' what you already knew? and, on what basis would you select the numbers to put in the right-hand column boxes?))

So, have you made your choice?

  • Which of the three decision-making methods is best suited to humans, and which to machines?
  • Come to that which, if any, is valuable, or would you rate them suspect?

And that is one of the main issues with decision-methods

  • Individuals and groups make decisions differently
  • Experts make decisions differently from naive beginners
  • Experts make decisions in different ways according to the urgency
  • Experts under pressure "satisfice"
  • Homo sapiens evolved by satisficing&endash;remember, who trades off options when the tiger is five steps behind?

We do not understand yet how we comprehend and make decisions

Last updated: Feb 2005

© D K Hitchins 2016