Stress and Human Error

1
Stress and Human Error

• chapter 12

2
Stress

• A stressor is any thing which
• is not an inherent characteristic of the
  human/machine system
• is not inherent in the information to be
  processed
• is (usually) expected to degrade system
  performance

3
Stressors

• Environmental stressors may include noise,
  extreme temperature, vibration, poor light, drugs
  alcohol
• Psychological stressors might include anxiety,
  frustration, fatigue, incentive

4
Stress

• Stress may have effects on
• phenomenology/affect (emotions)
• physiology
• sensory physical performance
• For example, difficulty manipulating tools with
  cold hands
• info processing/cognition
• For example, difficulty concentrating, difficulty
  performing mental arithmetic

5
How does stress affect performance?

• Yerkes-Dodson law
• Some stress is good, too much is bad. Optimal
  level depends on task complexity.

Good
Easy Task
Performance
HardTask
Bad
Low
High
Stress
6
Specific effects of stress

• Stress modulates arousal/alertness
• arousal general energy level of operator
• higher levels of arousal are generally better,
  i.e., lethargy does not produce good performance.

7
Effects of stress on attention

• Stress narrows attentional selectivity
• stress can increase attentional selectivity
• narrows attention to focus on info that is
  perceived as most task relevant
• under low stress, attention can be too
  non-selective
• under high stress, attention can be too narrow
• focusing errors
• Stress is distracting
• disrupts attention to relevant info tasks
• selection errors

8
Specific effects of stress

• Stress reduces working memory
• can distract attention from verbal or spatial
  rehearsal
• external noise can degrade info in a WM store
• consequences are worse for complex tasks, which
  tend to rely more on WM
• can prevent effective training and learning
• training in a less-stressful simulated
  environment can enhance learning
• Note that retrieval from long-term memory is
  spared!

9
Effects of stress Decision Making

• Stress affects speed/accuracy of performance
• can degrade perceptual-motor decision making
  performance
• can encourage people to trade accuracy for speed
• Stress can encourage use of different decision
  making strategies
• may encourage shift away from slow, deliberative
  decision making toward faster more intuitive
  strategies
• will make DMs more risk-averse
• will degrade performance more for novices, whose
  rapid DM skills are poorer

10
Specific effects of stress

• Stress produces response perseveration
• encourages people to rely on familiar or
  recently-used actions.
• People may continue to make the same response
  even though it has detrimental consequences.
• discourages creative thinking, generation of new
  strategies or action plans (cognitive narrowing)

11
Effects of specific stressors

• Intermittent Noise
• effects are worse if bursts occur randomly
  instead of predictably
• distracts observer from an ongoing task
• distracts attention from rehearsal in WM
• effects may last a short time or a relatively
  long time
• For example, a short burst of noise interrupts
  manual tracking for 2-3 secs, interrupts visual
  scanning for as long as 30 secs
• Example (visual noise)
• Flash bulbs while speaking

12
Effects of specific stressors

• Continuous Noise
• encourages people to selectively attend to
  sources of info that they think are most relevant
• affects attention to spatial non-spatial
  stimulus aspects
• For example, noise increases detection rates for
  signals at central locations, decreases
  detections rates for signals in the periphery
  (attentional narrowing)
• Also reduces the Stroop effect
• Also encourages people to remember some aspects
  of stimuli better than others

13
Effects of specific stressors

• Continuous Noise
• reduces WM, degrading stored info distracting
  from rehearsal manipulation
• For example, complex but not simple mental
  arithmetic is more difficult in a noisy
  environment
• continues to degrade performance for 20-30
  minutes after noise is terminated, especially if
  people believe they have no control over the
  noise

14
Effects of specific stressors

• Anxiety
• narrows attention
• For example, when led to believe that they were
  undergoing conditions of a 60 ft dive in a
  pressure chamber, a group of divers shifted
  attention away from peripheral signals in a
  dual-task study and focused on central signals
• reduces WM
• For example, inexperienced public speakers show
  WM loss shortly before making presentation in
  front of audience
• can lead to perseveration
• For example, when threatened with shock for
  failure, problem solvers tend to get stuck on an
  unsuccessful solution

15
Effects of specific stressors

• Time Pressure
• may encourage people to accelerate performance
  through inadequate info processing, omitted
  processing steps
• speed accuracy tradeoffs are likely
• For example, asked to decide which of two groups
  of numbers has a higher mean subjects under time
  pressure will base judgments on sum
• 3, 8, 10 vs. 8, 10

16
Effects of specific stressors

• Time Pressure
• may encourage shift away from slow, deliberative
  decision making toward faster intuitive decision
  making
• Although the faster method might not be as
  accurate, it is less affected by stress and might
  be more accurate than the deliberate methods in a
  stressful situation.

17
Effects of specific stressors

• Time Pressure
• because experienced DMs are more likely to be
  able to retrieve solutions from LTM, their
  performance is less affected by time pressure
• For example, skilled pilots show better decision
  making than novices under time pressure, and are
  more likely to carry out their first retrieved
  solution
• Similarly, experienced chess players are less
  degraded under time pressure than novices, and
  are more likely to generate a proper move
  immediately

18
How can operators cope with stress?

• Remove the stressor.
• Put up with it.
• realise your weaknesses
• Try harder.
• may itself be stressful
• may entail speed-accuracy tradeoff
• Change goals/methods
• change DM strategies
• Use a simple heuristic rather than a detailed
  analysis

19
Remediating effects of stress

• Design tasks displays to avoid exacerbating
  cognitive effects of stress
• minimize WM demands
• make relevant information attentionally salient
• as much as possible, give operator control over
  stressor
• Provide knowledge of control over stressors
• make users aware of possible stressors and how
  they should cope
• make users aware of when they can expect
  stressors to begin and end

20
Remediating effects of stress

• Train for effective performance under stress
  ensure that skills are overlearned
• responses become automatic
• information can be retrieved from LTM
• Design train for emergencies
• be aware that overlearning of emergency skills
  will not happen during normal operations
• That is, emergencies are rare during normal ops.
• make emergency procedures as consistent as
  possible with normal procedures

21
Human Error

• Human error is an important topic for study
  because
• provides knowledge of form and causes of error
• errors are consequential
• For example, a study by the Institute of Medicine
  estimated that medical errors kill 44,000 or more
  patients annually. A newer study found that
  medication errors occurred on 5.7 of all orders
  in a pair of childrens hospitals

22
Errors from Engineering Psychologists View

• Some errors are inevitable
• In SDT, perfect detection/discrimination is rare
  because of in Signal Noise curves often overlap
• Perfect decision making is impossible in a
  probabilistic environment.
• e.g. given x, 90 of the time y happens, but
  sometimes z happens instead
• Time stress often precludes optimal human
  performancefor firefighters, soldiers, police,
  E.R. doctors, etc., a mediocre choice in time is
  better than an optimal choice too late (Klein,
  1996)

23
Errors from Engineering Psychologists View

• Some errors are only errors in hindsight
• Errors can be the result of a long chain of very
  unlikely events.
• Hindsight bias (belief that I knew it all
  along!) makes cause of errors look obvious in
  retrospect.
• Second guessing is easier than actually
  predicting errors.

24
Classifying Errors

• Mistakes intended behavior is inappropriate
  under the circumstances
• knowledge based mistakes result from poor
  information, poor situation awareness, etc.
• For example, when the USS Vincennes shot down and
  Iranian airliner,
• USN had been expecting hostility
• Vincennes was being fired on by Iranian navy
• airliner was not on commercial airline schedules
• airliner was mistakenly identified as military
• airliner was not responding to communications

25
Classifying Errors

• Mistakes intended behavior is inappropriate
  under the circumstances
• rule based mistakes result when info and
  situation awareness are good, but operator lacks
  experience/knowledge to choose correct action
• For example, deciding to let a flight take off
  despite knowledge of potentially dangerous
  weather

26
Classifying Errors

• Slips intended behavior is appropriate, but is
  not carried out correctly
• Turning on the wrong burner
• Grabbing the wrong bottle of medication
• Hitting Save function instead of Save As
• Audi unintended acceleration
• are especially likely to occur when a familiar or
  automatic behavior must be inhibited
• Writing the wrong year on a check
• Putting car into drive when it should be in
  reverse

27
Classifying Errors

• Lapses an intended action is forgotten, not
  carried out
• Failing to push Timer button on VCR
• Failing to put gas cap on after filling up
• Failing to attach files to email

28
Classifying Errors

• Mode Errors action is appropriate in one system
  mode, but is carried out in a different mode
• typically results because correct mode is not
  reinstated when it should be
• Failing to put car back into drive after backing
  up a little bit
• Typing in Overwrite mode instead of Insert mode
• Failing to put remote control into mode
  appropriate for device (TV, VCR, stereo)

29
Classifying Errors

• Errors can be distinguished from violations.
• Violations intentional choice to disobey rule or
  procedure, not meant to cause harm
• Flying too close to another aircraft
• Surfacing a submarine without doing full acoustic
  or visual search
• Releasing bombs from fighter jet without
  permission
• Setting yourself on fire because someone on MTV
  did it
• can occur because of organizational emphasis on
  costs or productivity instead of safety
• For example, failing to meet OSHA standards

30
Remediating Errors

• Equipment design
• avoid perceptual confusions, ensure that stimuli
  are easily discriminable
• Ensure that medicine bottles are clearly labeled
• Ensure that ports on computer are labeled,
  discriminable
• Bad design USB ports
• make consequences of actions visible to user so
  that slips can be detected
• Display message onscreen to indicate if VCR is in
  Play, Record, FF,

31
Remediating Errors

• lock-out inappropriate behaviors
• Dont allow automatic transmission to be moved
  from park without the brake pedal depressed.
• Dont allow car windows to be raised with one
  touch of button
• Dont allow key to be removed from vehicle that
  isnt in park
• Give warning before deleting files
• give reminders
• Ask user if file should be saved before program
  shuts down

32
Remediating Errors

• avoid multi-mode systems, make mode easily
  visible
• Make overwrite/insert cursors distinct
• know that errors will occur build
  error-tolerant systems avoid systems where
  consequential actions are irreversible
• Include undo functions in software

33
(No Transcript)
34
Human Reliability Analysis

• Reliability probability of making an error
• Serial vs. Redundant Systems
• Serial systems
• only as good as the weakest link
• Redundant systems
• only fails when all systems at a certain level
  fail
• more robust than a serial system

35
Human Reliability Analysis

• Serial System
• only as strong as the weakest link.
• probability of the system failing is the product
  of the probability of any part not failing.

.90
.81 probability of success
.90
36
Human Reliability Analysis

• Redundant System
• only fails when all systems at a certain level
  fail.
• probability of the system failing is the product
  of the probability of the parts at one level
  failing.

.90
.99 probability of success
.90
37
Human Reliability Analysis

• Theoretically, HRP could be used to successfully
  predict failure rates of a system
• Would make it easier to chose between different
  designs.
• Would spotlight weaknesses in the system.

38
Human Reliability Analysis

• However, there are major problems
• Lack of a database
• the data for the frequency of cognitive errors is
  rather thin.
• Error monitoring and non-independence
• If Humans make an error
• they may change their behavior.
• Humans are complex
• If they detect a weakness (dim display) they may
  change their strategies.
• Humans take machine reliability into account
• if the machine is not reliable, people will make
  adjustments.

39
(No Transcript)