@MartinSpecken#CognitiveEngineering

Decision-making models, decision support, and problem solving

Mark R. Lehto, Fiona Fui-Hoon Nah, Ji Soo Yiin Human Factors Fundamentals (2012)

@MartinSpecken#CognitiveEngineering

Integrative model of decision making

• Single DM• Time pressure• Multiple stages• Risk attitudes• Knowledge• Decision tools• Multiple DMs

@MartinSpecken#CognitiveEngineering

(Sometimes overlapping)

1. Group

2. Dynamic

3. Routine

4. Conflict

Categories of decision making (1/5)

@MartinSpecken#CognitiveEngineeringGroup decision making

•Debate•Bargain•Vote

Categories of decision making (2/5)

@MartinSpecken#CognitiveEngineeringDynamic decision making

• Decision 1: Take medical test• Decision 2: What to do with the result?

Categories of decision making (3/5)

@MartinSpecken#CognitiveEngineeringRoutine decision making

Categories of decision making (4/5)

@MartinSpecken#CognitiveEngineeringConflict-driven decision making

Categories of decision making (5/5)

Various forms of conflict must be resolved beforean alternative action can be chosen.

@MartinSpecken#CognitiveEngineering

Decision making models (1/3)

1. Normative

2. Behavioral

3. Naturalistic

@MartinSpecken#CognitiveEngineering

Normative decision models (1/6)

Classic decision theory with 4 elements:

• Potential actions to choose between [A]

• Events/world states [E]

• Probabilities [P] from combination [A] & [E]

• Consequences [C] from combination [A] & [E]

@MartinSpecken#CognitiveEngineering

Normative decision models (2/6)

1. Dominance

2. Maximizing expected value

3. Subjective expected utility theory

4. Multiattribute utility theory

@MartinSpecken#CognitiveEngineering Dominance

Occurs between two alternativeactions, Ai and Aj , when:

Action Ai is at least as good as Aj

– for all events EAND

– for at least one event Ek

then Ai is preferred to Aj

Normative decision models (3/6)

@MartinSpecken#CognitiveEngineeringMaximizing Expected Value

Select the alternative with the greatest expected value.

• 3 kittens are stuck in trees• Decision:

– Save 3 kittens is 3 timesas good as saving 1 kitten.

• Choises:– 100% chance to save 1 kitten

Expected value = 1 * 1 = 1– 50% chance to save all 3 kittens + 50% to save 0 kittens

Expected value = ( 3 * 0,5 + 0 * 0,5 ) = 1,5

Normative decision models (4/6)

@MartinSpecken#CognitiveEngineeringSubjective Expected Utility Theory

Basic axioms (Dutch: grondstelling) of rational choise:

• If A > B and B > C , then A > C

• People’s preferences can conflict with the axioms.• Movement toward less restrictive standards.

Normative decision models (5/6)

@MartinSpecken#CognitiveEngineeringMultiattribute Utility Theory

Normative decision models (6/6)

Consider attractiveness of• Economic benifits• Social benefits• Environmental benefits

@MartinSpecken#CognitiveEngineering

Decision making models (2/3)

1. Normative

2. Behavioral

3. Naturalistic

@MartinSpecken#CognitiveEngineering

Behavioral decision models (1/5)

Normative decision models don’t always work when compared to human behavior.

1. Statistical estimation and inference

2. Preference and choise

3. Adaptive decision behavior

4. Behavior economics

@MartinSpecken#CognitiveEngineeringStatistical estimation and inference

1960s: People behave as ’intuïtive statisticians’. Many examples of findings in the paper. A selection:

• DMs tend to ignore the reliability of the evidence.

• DMs tend to seek out confirming evidence rather than disconfirming evidence.

• DMs are overconfident in their predictions.

Behavior decision models (2/5)

@MartinSpecken#CognitiveEngineeringPreference and choise

People’s preferences changewhen the outcomes areframed in terms of costs,as opposed to benifits.

Behavior decision models (3/5)

Research: observing human preference and choise and comparing it with the SEU theory

@MartinSpecken#CognitiveEngineeringAdaptive decision behavior

Behavior decision models (4/5)

Individual DM uses different strategies

in different situations.

@MartinSpecken#CognitiveEngineering Behavior economics

Behavior decision models (5/5)

@MartinSpecken#CognitiveEngineering

Decision making models (3/3)

1. Normative

2. Behavioral

3. Naturalistic

@MartinSpecken#CognitiveEngineering

Naturalistic decision models (1/5)

In dynamic and realistic environment, actions by decision maker are made sequentially in time.

1. Levels of task performance

2. Recognition-primed decision making

3. Dominance structuring

4. Explanation-based decision making

@MartinSpecken#CognitiveEngineeringLevels of task performance

Most decisions are made on a routine basis. The levels of performance tasks are based on:

• Skills• Rules• Knowledge• Judgement

Naturalistic decision models (2/5)

@MartinSpecken#CognitiveEngineeringRecognition-primed decision making

80% of firefighters make decisions based on some sort of situation recognition

Naturalistic decision models (3/5)

@MartinSpecken#CognitiveEngineeringDominance structuring

Nonroutine decisionsinvolve sequence of:

1. Screen alternatives.2. Select alternative.3. Check dominance of selection to

other alternatives.4. No dominance? Restructure info

to force dominance.

Naturalistic decision models (4/5)

@MartinSpecken#CognitiveEngineeringExplanation-based decision making

Jurors seem to organize massive amount of data in terms of stories describing cause and intent.

Naturalistic decision models (5/5)

@MartinSpecken#CognitiveEngineering

Group decision making (1/5)

1. Ethics and social norms

2. Group processes

3. Group performance and biasses

4. Prescriptive approaches

@MartinSpecken#CognitiveEngineeringEthics and social norms

Four dilemma’s of right versus right (Kidder, 1995)

1. Truth versus Loyalty

2. Individual versus Community

3. Short term versus Long term

4. Justice versus Mercy

Group decision making (2/5)

@MartinSpecken#CognitiveEngineering Group processes

Tuckman (1965):

1. Forming (initial orientation)

2. Storming (conflict)

3. Norming (develop cohesion, express opinions)

4. Performing (obtain solutions)

Group decision making (3/5)

@MartinSpecken#CognitiveEngineeringGroup performance and biasses

Strong opposition to usefull products

Group decision making (4/5)

@MartinSpecken#CognitiveEngineeringPrescriptive approaches

Group decision making (5/5)

@MartinSpecken#CognitiveEngineering

1. Decision analysis

2. Individual decision support

3. Group and organizational decision support

4. Problem solving

Decision support and problem solving (1/5)

@MartinSpecken#CognitiveEngineering Decision analysis

Decision support and problem solving (2/5)

C11

C12

C21

C22

E1

E2

E1

E2

A1

A2

P1

1 - P1

1 – P2

P2

A = ActionE = EventP = ProbabilityC = Consequenses

@MartinSpecken#CognitiveEngineeringIndividual decision support

Decision support and problem solving (3/5)

@MartinSpecken#CognitiveEngineeringGroup and organizational decision support

Negotiation Support Systems assist people in activities that are competitive of involve conflicts of interest.

Decision support and problem solving (4/5)

@MartinSpecken#CognitiveEngineering Problem solving

Psychologist work togetherwith computer scientist to

understand the human mind

Decision support and problem solving (5/5)

@MartinSpecken#CognitiveEngineering

Book Malcolm Cook et al. (2007)

http://tinyurl.com/boekdecisionmaking

@MartinSpecken#CognitiveEngineering

Paper Sarah Sharples et al. (2007) (1/5)

http://www.ncbi.nlm.nih.gov/pubmed/17499573

@MartinSpecken#CognitiveEngineering

Key components in airspace systems:

• Air traffic controllers

• Airline management

• Aircraft pilots

• Aircraft systems

Paper Sarah Sharples et al. (2007) (2/5)

@MartinSpecken#CognitiveEngineering

Different goals:

• Air traffic controllerssafety of all the aircrafts in the system

• Airline managementexpediency & efficiency (profit)

• Aircraft pilotssafety & expediency of aircraft (local)

Paper Sarah Sharples et al. (2007) (3/5)

@MartinSpecken#CognitiveEngineering

Different available information:

• Air traffic controllersmore: global traffic patternsless: global weather

• Airline managementless: traffic patternsmore: current weather

Paper Sarah Sharples et al. (2007) (4/5)

@MartinSpecken#CognitiveEngineering

Increase of air traffic: How can extra aircrafts operate economically, efficiently and safe?

• Freeflightsome transfer of responsibility from air traffic controller to pilot for determining flight paths

• Datalinkprovide electronic exchange of information between pilots and air traffic controllers

Paper Sarah Sharples et al. (2007) (5/5)

@MartinSpecken#CognitiveEngineering

Finally

The integrative model ofdecision making showshow the various approachesfit together as a whole.