Aeronautical Decision Making I. Introduction A. Scholars frequently write about prescriptive (what people should do) and descriptive (what people actually do) models of decision-making. B. Better to think of these as styles of decision-making, each of which is used and each of which may be optimal under the right conditions. C. McGuire: “All good ideas are true.” The question is “Under what conditions are they true?” II. Models/Styles of Decision Making A. “Rational” Decision Maker 1. Utility Theory i. Based in economics and philosophy of logic 2. Subjective Expected Utility (SEU) Theory i. Adaptation of utility theory to allow for subjective costs/benefits 3. Process i. Goals: what do you want? a. Examples Safety: Get there safely Time: Get there fast Money: Get there cheaply Comfort: Get there comfortably ii. Utility: how much is it worth to you? a. Examples Objective: Price of ticket/ Price of fuel Subjective: Value of time/Value of comfort iii. Actions/Options: each option has costs/benefits related to one or more goals a. Examples Fly direct route through turbulence Safety: depending on degree may be relevant or not Time: quicker than deviation Money: cheaper than deviation (in short run) Comfort: more uncomfortable Fly direct route through thunderstorm Safety: extremely hazardous Time: quicker than deviation Money: cheaper than deviation (if succeed) Comfort: very uncomfortable iv. Probabilities: each action will incur costs/achieve benefits with some probability v. Expected Values: function of probability and benefit/cost a. Examples Probability of encountering severe turbulence may be quite small (e.g., .001) If encounter severe turbulence, cost in safety/comfort may be very large (e.g., $1,000,000) Partial expected value = probability*cost = .001(-$1,000,000) = -$1000 vi. Optimal Choice: choose that option that maximizes total expected value 4. Considerations i. Cost of information: getting information is itself costly a. Time Example: Time to get enhanced weather briefing b. Resources Example: Equipment, dispatchers, service subscriptions ii. Opportunity costs a. Taking one action now may preclude taking other actions now or later Example: Look/see departure may preclude taking ground transportation to destination on time 5. Problems i. Only optimal under Full Information. If have limited information (uncertainty) about the components, the rational model may not produce optimal solution. ii. Example: If you don’t know how important time is or how likely it is to encounter turbulence, can’t produce expected value. B. Bounded Rationality 1. Human as “Cognitive Miser” – trying to save mental work whenever possible i. Use satisficing instead of maximizing 2. Leads to use of “heuristics” that “bias” the decision outcome in predictable ways which in turn lead to not-quite or “bounded” rational decisions. 3. Examples i. Elimination by Aspects for selecting options ii. Framing Effects (cf. Prospect Theory; Kahneman & Tversky, 1973) a. People tend to be risk-averse in the domain of gains but risk-seeking in the domain of losses b. Example: Asian Disease Problem (Kahneman & Tversky, 1973) c. Example: VFR pilot Go/No go decision Loss (risk seeking) If drive will miss beginning of meeting If fly may encounter IMC and could crash Gain (risk averse) If fly may make meeting on time or not at all If drive will be sure to make most of meeting iii. Anchoring & Adjustment a. People tend to under-adjust from anchor Example: Countries in Africa Example: PIREP altitudes on climb/descent iv. Availability heuristic a. People tend to overestimate likelihood of things that can be easily brought to mind Example: Number of words that start with “R” or have “r” as 3rd letter Example: Terrorism vs. accident risk b. Hindsight bias People tend to believe that what happened was more likely and could have been foreseen Example: Accident investigation results v. Representativeness heuristic a. People tend to judge likelihood by how representative outcome is to category Example: Linda the feminist bank teller Example: Likelihood of encountering severe icing 4. Considerations i. Stable/Known environment vs. Unstable/Unknown environment a. Use of heuristic may be useful and appropriate when know environment well and the environment remains stable. ii. Experience a. Use of heuristic may be useful when have enough experience in environment to make heuristic-based judgments accurate (e.g., most available can be most frequent). iii. Importance a. More likely to rely on heuristic short cuts when not appropriate if believe that the decision is unimportant (e.g., psychology experiment vs. life/death decision). iv. Motivation a. More likely to rely on heuristic short cuts when not appropriate if not motivated to expend energy to solve problem (e.g., psychology experiment vs. life/death decision). C. Rule-based/Codified/Procedural Decision-making 1. Organizations frequently impose rules on individuals and require selection of a particular option when conditions are met. i. Examples: FAA FARs; Airline SOP 2. Can decide to obey self-generated rules i. Examples: Don’t date the boss ii. Examples: FAA Personal Minimums program 3. Considerations i. Effective when the environment is known and stable as are the individuals. ii. Under these conditions, rule-based decisions are fast, efficient, and independent of characteristics and experience of individuals iii. However, difficult to write rules to cover all situations a. Example: According to FARs, VFR pilot cannot legally depart controlled airport in Kansas with 950 ft ceiling and 10 mile visibility but can depart airport in Colorado with 1000 ft ceiling and 3 mile visibility. iv. Rarely have rules to change rules when individuals or situations change a. Example: Beginning IFR pilot may decide not to fly approach with less than 1000 ft ceiling and 1 mile visibility, but legally can fly approach with 200 ft ceiling and ¼ mile visibility. When is it reasonable to change “personal minimum”? D. Associative Decision-making 1. Recognition Primed Decision-making (Klein) 2. Make decisions by recognizing situation and implementing previously associated solution and then monitors for expected outcome. If unexpected result, quickly alters actions. i. Examples a. Fire Captain recognizes type of blaze; directs fire fighters appropriately b. Pilot recognizes Pacific cold front and chooses route/altitude appropriately c. Chess masters recognize patterns; can play multiple games simultanteously 3. Qualities i. When environment is stable, experienced individual can form appropriate associations make fast and efficient decisions that are relatively immune from physiological and emotional interference ii. Requires appropriate experience and possibly sensitivity to internal cues a. Example: advantage of appropriate “gut-level” feeling in handling emergencies b. Example: FBI agent examples of failure to rely on “gut-level” feelings c. Questions: Is virtual or vicarious experience enough to form associations? III. Adaptive Decision Maker (Payne) A. Novice can only use rule-based decisions or analytical decision strategy B. Domain expert can use associative decision strategy C. Expert decision maker adapts decision strategy to situation 1. When in known environment, uses associate decision strategy 2. When in novel environment or when associative strategy fails, reverts to analytical strategy IV. Factors that affect Decision-making A. Individual Differences 1. Individual differences can be arrayed on a stable (slow to change) - unstable (quick to change) continuum 2. Personality traits (stable) i. Big-Five (Goldberg); macro-level factors a. Openness b. Conscientiousness c. Extraversion d. Agreeableness e. Neuroticism ii. Temperament (Rothbart); lower-level factors a. Attention focusing b. Attention shifting c. Effortful control d. Impulsivity e. Etc. 3. Hazardous Attitudes (slow change) i. E.g., Machismo, invulnerability, etc. ii. Poor psychometric properties; no evidence of validity 4. Experience (slow-moderate speed change) i. Total a. Little relation between total amount of hours and any outcome measure; but used extensively by employers. ii. Recent a. In general, recent experience predicts better outcomes iii. Type a. Experience with particular environment predicts better outcomes (e.g., experience with winter in Northwest predicts decision outcomes in Northwest; experience with winter in the Southeast does not) 5. Training (relatively quick change) i. Structure a. More highly structured training (Part 141) may be better but confounded with variety of other variables (e.g., consistency, motivation) b. Sometimes additional skills only allow for additional problems Example: Instrument rating c. Skills learned and retained better with spaced rather than massed practice Example: Problem with “10-day instrument course” ii. Decision Training a. Little evidence for effectiveness of general training in logic, statistics, rational decision theory b. Some evidence for domain-specific decision training when combined with domain-specific practice 6. Physiological State (relatively fast change) i. Fatigue reduces attentional focus; lessens motivation ii. See physiology notes 7. Emotional State i. Anxiety reduces attentional shifting; short-term memory capacity ii. See emotion notes B. Environment 1. Characteristics of environment affect decision style and effectiveness i. Familiarity ii. Complexity iii. Time 2. B=f(p, s, p*s) V. Internet-based Decision Research System A. Pilot research conducted at UO