Copyright Catherine M. Burns Situation Awareness Copyright Catherine M. Burns
Copyright Catherine M. Burns What is Situation Awareness (SA)? Awareness of the meaning of dynamic changes in the environment “Perception of elements in the environment within a volume of time and space, the comprehension of their meaning and the projection of their status in the near future” (Endsley, 1995) Copyright Catherine M. Burns
Copyright Catherine M. Burns Stages of SA Perception Understanding Prediction Copyright Catherine M. Burns
Copyright Catherine M. Burns Power Grid Failure “Lack of Situation Awareness” as a root cause Copyright Catherine M. Burns
Copyright Catherine M. Burns Measuring SA SAGAT (Situation Awareness Global Assessment Technique Operator performs a complex task Periodic interruptions Asked SA questions Location of other people Location of hazards Copyright Catherine M. Burns
Copyright Catherine M. Burns Decision Making Perception Understanding Prediction DECISIONMAKING Wickens Ch7 Copyright Catherine M. Burns
Decision-Making Tasks A decision making task is a task where a person selects one choice from a number of choices there is some information available on the choices the time frame is relatively long (1 sec) there is some uncertainty Copyright Catherine M. Burns
Copyright Catherine M. Burns Tasks medical diagnosis flight judgements process control fault diagnosis safety-related behaviour Copyright Catherine M. Burns
Normative vs Descriptive Models specify what people should ideally do, utility Descriptive describe what people actually do descriptions of what can influence decision making Copyright Catherine M. Burns
Copyright Catherine M. Burns Normative Methods Multi-attribute utility theory Expected value theory Subjective expected value theory Copyright Catherine M. Burns
Multi-Attribute Utility Theory U(v)= S i=1 n a(i)u(i) The sum over all attributes of the magnitude of each attribute multiplied by its utility A way of comparing alternatives that have many different dimensions Copyright Catherine M. Burns
Copyright Catherine M. Burns Car Example in Text You’ve weighted the various car attributes as follows: Engine performance 8 Brake performance 5 Styling 4 Seat comfort 2 Floor mats 1 Copyright Catherine M. Burns
Copyright Catherine M. Burns Car example You are considering the following cars 2004 Mazda 3 2002 Honda Civic 2003 Ford Focus 2004 Honda Civic Copyright Catherine M. Burns
The cars rate as follows: Styling Brakes Seats Mats Engine Score Utility 4 5 2 1 8 M3 3 9 56 H2 60 FF 120 H4 118 Copyright Catherine M. Burns
Copyright Catherine M. Burns Expected Value Theory determines the value of various outcomes under uncertainty assumes people should pick the highest value outcome probabilistic “gambling” type questions p=0.2 of winning $50 E=0.2x$50=$10 p=0.6 of winning $20 E=0.6x$20=$12 People should pick option two Copyright Catherine M. Burns
Problem with Expected Value Theory People don’t actually make the “optimal” choice Values on outcomes can be subjective, different for different people Subjective Expected Utility Theory Copyright Catherine M. Burns
Copyright Catherine M. Burns SEUT Assigns a subjective utility value, instead of the objective value used in EVT Basic idea is still probability (p) x utility (u) highest expected utility is the best decision Copyright Catherine M. Burns
Copyright Catherine M. Burns Descriptive Models Mostly isolate effects that can influence decision making Understand why human decision making doesn’t follow normative models Copyright Catherine M. Burns
Copyright Catherine M. Burns Descriptive Models Satisficing: People look for the first solution that meets the criteria, not the optimal solution Used when there are a large number of potential alternatives, limited time Simplifications, heuristics and biases People create easier ways of thinking about things Copyright Catherine M. Burns
Copyright Catherine M. Burns Heuristics and Biases Three stages: 1. Getting information input (cues) 2. Generating hypotheses 3. Plan generation and action choice Copyright Catherine M. Burns
1. Cue Perception and integration Input or Cue Biases : People pay attention to a limited number of cues, Early information is more influential (cue primacy), Late information is underweighted, Very visible cues are given more weight, Overweighting of unreliable information. Copyright Catherine M. Burns
2. Hypothesis Generation and Selection Biases: People only generate a small number of hypotheses Most frequently seen hypotheses are preferred (availability heuristic) Minimal information gathering if there is a strong cue match (representativeness) Overconfidence. Tend to think they are more correct than they really are. Cognitive Tunneling: Reluctance to change from a hypothesis Confirmation bias: Tendency to only look for confirming information. Copyright Catherine M. Burns
Copyright Catherine M. Burns 3. Action Choice Biases in Action Choice: Retrieving a small number of actions Retrieving the most frequent or recently used actions Evaluation or estimation of the likely outcomes of actions Framing bias: presentation of problem affects the decision Copyright Catherine M. Burns
Special Case of Framing Bias “Sunk Cost Bias” Investors who resist selling losing stocks even when the long term better choice is to sell Copyright Catherine M. Burns
Copyright Catherine M. Burns The SRK Framework Suggests people have to make three categories of “decisions” 1. automated or skill based decisions 2. procedural or rule based decisions 3. knowledge based decisions Depends on familiarity of the situation and experience of the person Copyright Catherine M. Burns
Copyright Catherine M. Burns Decision Support Often aim to reduce decision making biases make relevant cues more salient suggest alternative courses of action provide additional information, especially disconfirming information show realistic simulations of decisions Simulation, expert systems and displays Copyright Catherine M. Burns