Training Fast and Frugal Heuristics in Military Decision Making

This tested decision making (DM) performance by teaching junior officers in a fast & frugal heuristic.

A heuristic is “a strategy that arrives at a satisfactory solution with a modest amount of computation” (p3). Heuristics allow adequate strategies to be found particularly when time is limited.

Providing background, they note:

·????????While heuristics simplify and reduce efforts in decision making, concerns surround how they can lead to biased thinking. However, more recent evidence has “identified a number of heuristics that are not less accurate than complex strategies and may even be more accurate” (p3)

·????????DM in the military is said to be taught formally as a sequence of analytical steps. It involves assessing the situation, developing several courses of action, comparing them & developing orders. The military process emphasised a rigorous process of logical deduction to complex and ill-structured situations; but there are many instances when this process isn’t suitable, like under time pressures.

·????????Instead, here, they used a fast & frugal decision tree heuristic, which draws on only a few cues. A key facet was a simple structured fast & frugal heuristic which was “plausible, fast, and simple to apply” (p4)

·????????They discuss Naturalistic Decision Making (NDM) and Recognition Primed Decision Making (RPD) in this study. Other work found that 60% of military decision strategies during an exercise were coded under RPD compared to just 2% of strategies coded as ‘classical decision making’

·????????They note NDM shares some fundamental concepts with fast & frugal heuristics, e.g. “Both areas rejected the heuristics and biases programme which implied that people were irrational and flawed decision makers” (p5)

·????????Further, both areas “share a view of bounded rationality that is constrained by the environment and cognitive limitations rather than comparing people to an unbounded rational standard of decision making” (p5)

·????????There are differences, though. They highlight that while RPD and fast & frugal models “both describe responses to cues, typically recognition-primed decision making relies more on cues from experience and fast and frugal heuristics rely more on cues in the environment” (p5)

·????????They argue that fast & frugal heuristics are “modelled more formally and with greater precision than NDM”, but most work in this area hasn’t taken place in real-world settings. In contrast, NDM focuses on real experts in real settings but, in their view, has largely enhanced knowledge around how experts make sense rather than novices without experience

Method

Authors identified the most important cues used for DM making in some military scenarios. 14 cues were identified & 32 scenarios created based on this. Using multiple regression, cues were analysed based on how influential they were in the decisions.

The expert heuristic was then made by finding “the minimum number of cues that could effectively classify the experts’ decisions”. Only 3 cues incorporated in the heuristic was needed to predict ~80% of expert decisions.

58 junior officers were divided into a training group (heuristic) or control (no training). DM performance was compared pre/post training & between groups.

Results

Found was that making decisions via the heuristic leads to decision accuracy similar to decisions based on existing, more thorough analytical methods. No difference in overall workload was found between methods, although mental demand was lower in training group compared to control.

That is, the heuristic allowed decisions as accurate as control but with less mental demand. Authors note that their findings support other research that supports fast and frugal heuristics allowing as effective performance as more complex decision methods.

They explain that a common explanation for heuristics is the trade-off between accuracy and effort, where heuristics reduce effort but at the cost of accuracy. This study found mental effort to be reduced but *not* at the expense of accuracy – accuracy was maintained.

Interestingly, they mention that whereas the heuristic met the criteria for frugality, it wasn’t rated as fast – being similar in speed to the analytic procedure. They reason this could be due to the analytical method being highly trained in the officers whereas the heuristic was novel. Further training may improve heuristic speed.

Nevertheless, they do argue that a key benefit of the heuristic may be frugality rather than speed in trained decision makers. However, given that this was a classroom setting with few competing mental demands, the speed benefits may be realised under realistic settings with time and stress.

Authors note some circumstances where fast and frugal heuristics may not be as suitable. E.g. in courtroom or in capturing highly emotional aspects of decisions such as medical decisions. Here, authors reason that heuristics may complement existing analytical methods rather than replace them. [** However, Gerd Gigerenzer argues why heuristics can be very powerful in medical decisions.)

Overall, findings suggest that heuristics allow decisions as accurate as those of existing analytical methods but with less mental demand. Importantly, heuristics can be learnt.

Another challenge is that this method relied on developing scenarios and extracting cues. Thus, it may favour specific instances or problems.

For those interested, there’s some cool studies in the Resilience Engineering space exploring how to develop these types of scenarios and adaptive capacity of teams. I see a lot of value in teaching heuristics during fault finding and problem solving during unexpected conditions (e.g. high-hazard process faults).

Link in comments.

Authors: Banks, A. P., Gamblin, D. M., & Hutchinson, H. (2020). Applied Cognitive Psychology, 34(3), 699-709.