161 final
Human error probability
=Errors / chances for error E.g. 1 typing error per 100 letters typed
Anchoring
Anchor onto first answer you find Stick with initial diagnosis even as contradictory evidence begins to arrive
Crew resource management
Authority gradient Feedback and avoidance of ambiguity Tenerife accidence involved ambiguity and no feedback CRM trains nontechnical skills Briefing, monitoring, decision making, leadership Assertive, concise messaging Application to other domains with strong authority gradients Surgeon-nurse, supervisor-driller/technician
Authority gradients
Can the co-pilot or lieutenant speak up to the captain? Co-pilot knew fuel was running out; lieutenant knew the Vincennes was about to shoot an airliner Surgeon-nurse, supervisor-driller/technician
Affordances
Clues on how to use an object signs/ restrictions
Overconfidence
Examples Average driver estimates to be in the top 25% of safe drivers Sports and political predictions − Give a prediction and rate confidence, if confidence is high for bad predictions, then there is overconfidence
Icons
Good for International communication Realistic icons require little/no training Pictorial realism / Naïve realism - they look like the real thing Bad for confusability Military ships look similar, many aircraft look similar − 3D orientation adds more complexity, confusability Vs. Military standard symbols − Designed to be not confusable − Require training
Chatty copilot would affect automization design
Have automation announce mode changes, system status, other key factors, especially as they change
Prospective memory
Intention to remember in the future Longer delays lead to poorer memory • Good memory from related tasks − And salient/unusual cues • Motivation and richer encodings (images) can help Manufactured cues − Creating a cue to be noticed later − Sticky notes, marks, coffee cups on controls
Response compatibility
Map stimuli to responses Example: same organizational layout of burner to burner controls You can increase speed by increasing compatibility between stimulus and response. ex- location, rules (even backwards rules are better than no rules)
Mental models
More likely to be associations and memories of state transitions Some limited declarative knowledge of operations that can be reasoned over Can be wrong or incomplete Mental models for human factors design The assumption is that consumers can apply common mental models to how a consumer device operates Visibility: state and possible actions should be clear from the looking − Affordances
CHUNKING
Multiple elements that are bound together through long term memory Dog, cat, rat, goat are as easy to remember as 2,6,3,8 even though they have many more letters Old-time phone numbers, e.g. Riverside 4,3,7,9 - 5 chunks rather than 7 in a modern phone number − 4 1492 8 314 1865, "for Columbus ate pie at Appomattox" - One image chunk
Situation awareness
Perception, comprehension, projection (what? So what? Now what?) E.g. Air France 447: crew did not understand the situation and did not work to address it • Attention/detection • Selective attention can cause missed information • Change blindness
Out of loop problem
Reduced time to respond once alarms sound; "what's going on?"...boom!
Team situational awareness
Requires coordination, sharing, cross-checking Not just the sum of individual SA Saying the right thing to the right person at the right time Requires understanding the team's information needs throughout a task or situation Pushing information rather than waiting for it to be pulled − Recall Gricean Maxims Good displays can reduce the need for comms Threat assessment and management Briefings and sharing information, especially across shifts Vigilance to losses of SA among the team
Slips
Right intention is wrongly carried out
Skill retention
Skill type Perceptual-Motor skills: well maintained over time − E.g. flying/piloting • Cognitive skills: erode over time − E.g. Navigation − Related to time with an autopilot! − Retrieval cues, practice, consistency across applications help Sequence of practice Train perceptual components first Individual differences • Fast learners retain more; perhaps due to chunking
Speed accuracy tradeoff
Speed increases errors Experimenter can ask participants to emphasize accuracy or speed, or both − Different points on the trade-off curve
Data context tradeoff
Very fast or degraded presentations limit bottom-up processing (Top-down context contributes more) Top-down context requires some predictability and redundancy of message Task compatibility If random strings or digits are needed (no redundancy, no context) use larger font If there is plenty of context, a smaller font can be used
Experience and expertise
Very fast, go with the gut, requires good experience Experts have many situations that they apply the reasoning from to the current situation Eg. lightning chess
Word superiority effect
Words are read faster than nonwords Common words are read faster than rare words Word reading speed is a function of length and frequency
Understand cognitive task analysis
decision-centered task analysis Walk through a problem with an expert, discuss each step Focus on key decisions Add "what if" questions to explore alternative situations Prone to same problems as interviews, but the problem provides grounding and structure
Lapses
forgetting an action or step in a sequence Post completion error/ interruption/multi tasking/ Prospective memory
General idea of automization w/ compliancy and reliabilbity
tasks performed by machine which were previously performed by humans Compliance - respond to alerts Lowering the alert threshold leads to users ignoring the alerts (more false alarms) and lower compliance; they carry on trusting the automation Complexity makes reliability harder to check and manage − Organizational failure rather than automation, per se
Cell phone interference
• Cell phone use leads to more accidents • Equivalent to blood alcohol of 0.08% Conversations are more distracting than just listening E.g. more distracting than the radio Cell phone conversations are more distracting than in-person conversation • In-person talker can modulate due to workload conditions, e.g. upcoming freeway change Hands-free phones eliminate motor workload and visual feedback • But do not eliminate conversation factors
False alarms
• False alarms lead to a "cry wolf" response
Post completion errors
− Forgetting the last step to complete a sequence Often a "tidy up" step − E.g. ATM card, surgical sponge, aircraft maintenance