Decision making, uncertainty and bias

How can a sensible initial plan become a trap?

A plan should remain conditional on the evidence that supported it.

Bias shapes the evidence search

Confirmation bias favours information supporting an existing explanation. Plan-continuation bias can preserve an approach despite deteriorating conditions. Sunk effort makes past time feel relevant even though the next decision should compare future options and their risks.

Ask what would change the plan

Before a demanding phase, identify continuation criteria and a workable alternative. Seek evidence that could falsify the preferred explanation. A conservative decision is not automatically correct, but a decision that ignores a failed safety condition is not repaired by confidence.

Mechanism and limits

Risk perception is influenced by familiarity, social pressure and recent experience. An adverse event that did not occur on previous flights can become easier to discount, even if the current conditions differ. A recent vivid event may instead make one risk seem disproportionately likely. Neither feeling substitutes for the present facts and applicable criteria. Discussing alternatives before commitment helps reveal these influences. Hazardous attitudes describe tendencies in reasoning and behaviour; they should prompt a useful countermeasure and self-reflection rather than become fixed labels attached to a colleague’s personality.

Applied case · Name the failed assumption

A synthetic destination remains reachable, but a new weather report removes the margin assumed in the original approach plan. The crew has already spent time preparing that approach and begins highlighting every favourable detail. Ask which specific condition originally supported continuation and whether it remains true. Previous preparation effort does not restore the missing condition. Compare the feasible alternatives using current weather, fuel, aerodrome and crew information. An explicit trigger established before pressure rises makes a timely change easier to justify and helps the monitoring pilot challenge a plan that has lost its basis.

How does time pressure change a useful decision process?

Different time horizons require different levels of analysis.

Use the time actually available

An immediate hazard may require a trained, approved response before extended deliberation. With more time, compare options, constraints, uncertainties and consequences. Generating many options during an urgent control task can be as harmful as failing to explore alternatives during a long diversion decision.

Evaluate after acting

A decision includes monitoring its effect. Define the expected result, an update point and a trigger to change course. The decision lab distinguishes a satisfactory result from a merely lucky outcome: good process can still face uncertainty, while a poor process may occasionally escape harm.

Mechanism and limits

A useful option must satisfy constraints, not merely have the most attractive single feature. An aerodrome with the shortest distance may be unsuitable because of weather, runway or aircraft condition; a longer option may be infeasible because of time or fuel. Make the constraint explicit before comparing preferences. Uncertainty also needs an allowance: a plan that works only if every optimistic assumption holds is fragile. Once an action is selected, assign monitoring so the crew can recognise when a constraint or expected outcome changes and revise the plan while alternatives remain available.

Applied case · Choose a review point

A synthetic diversion decision is made with adequate time to compare two suitable options. The crew chooses the option with better weather but plans to review conditions before a route point that would reduce flexibility. This is a complete decision only if the expected benefit and review trigger are explicit. When a new report arrives, compare it with the assumptions instead of treating the earlier choice as a commitment. In a different immediate-control emergency, the analysis window would be much shorter and trained approved action would take priority. The same decision method should not be forced onto both time horizons.

Can you explain and apply the mechanism?

Four original questions. Work from the stated facts, then compare your reasoning with the explanation.