Title: PSY205s: The psychology of aviation - Situational Awareness
1PSY205s The psychology of aviation -
Situational Awareness
Dave Nunez, MPhil Department of
Psychology University of Cape Town
2Admin info
- I am in room 4.22 (?650 4606)
- Paper Endsley 1999 from Work Return Room
- Slides for the time being
- http//www.cs.uct.ac.za/dnunez/teaching
- (on the course web page later)
3Why study aviation in psychology?
- Airplanes get more complex.
- .but the people who fly them remain the same.
4Why study aviation in psychology?
How safe is flying? Deaths per year by
cause (USA 1981-1994) Commercial
flight 100 Electrical current 850 Bicycle
riding 1000 Pedestrian 8000 Falling 12000 A
uto accidents 46000 So flying to Joburg should
be about 460 times safer than driving there.
- Goal Increase safety (purely applied)
- Aviation is a potentially dangerous activity
- Increase safety by engineering (better engines,
etc) - What about the people doing the flying?
5A little history
- During the psychometrics boom (1920s-1930s),
psychologists get involved - Pilot selection tests
- During WW2, psychologists begin to look at
aviation loses - More aircraft lost to accidents than the enemy!
- Bartlett, Chapanis, Craik, Gibson and other big
names got involved
6Pilot error vs. Human error
- General rule was If the plane didnt fail, it
was pilot error - Pejorative phrase laid blame
- Implication not good enough
- Evidence from Chapanis and others showed it was
actually human error - Acknowledge limits of human beings
- Certain system features create situations where
an error is more likely - Problem becomes worse under certain environmental
conditions
Alphonse Chapanis (1917-2002), was a leading
figure in the psychology of aviation safety since
the 1940s
7Example Human error (and a solution)
- Problem Pilots shut down engines in flight by
pulling the wrong lever
Beechcraft Duchess (late 1970s)
Douglas DC-4 (early 1940s)
- Chapanis solution shape coding the lever
handles (also color coded by function)
8Human Factors (Ergonomics)
- Understand patterns of errors
- Many errors can occur regardless of experience
- What about being human makes us likely to commit
errors? - Examine cognitive processes to identify danger
spots - Goal To create systems which reduce the
probability of errors - Pilot friendly aircraft which reduces errors
and creates a better working environment
9A major contribution SA
- A recent important contribution Situational
awareness - Combination of mental models, working memory and
situated cognition theory - Tries to predict how and when errors can happen
- Applied to operation of complex systems (nuclear
powerplants, ships, aircraft, cars) - Much research, and is taught to pilots
- Increase their awareness of when things can go
wrong
10A quick recap of the info-processing model
- Herbert Simons model
- Attention, WM (STM), LTM.
- Attention filters irrelevant (unexpected) stimuli
out - Stimuli are transformed in WM according to active
rules and schema (from LTM) - Contents of WM in turn activate rules and schema
as required by the data - Behaviour/consciousness is based on the contents
of WM and active scripts
11SA definition
- the perception of the 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 1998) - Cognitive task (probably expertise bound)
- Dynamic (over time and space)
- Several levels of processing (perception,
comprehension, prediction)
12Levels of SA (1)
- Level 1 Perception of the environment
- Other aircraft, terrain, own aircraft systems,
navigation, radios - Deals with the present loads perceptual
buffers, attention and working memory
Mica Endsley (of SA Technologies, previously of
MIT) is one of the leading experts on situational
awareness in aviation
13Levels of SA (2)
- Level 2 Comprehension
- Synthesis of disjoint level 1 elements (big
picture) - Assign importance to each element (goal-directed)
- Forms a holistic understanding of what is
happening now - Reaching this level requires experience
- It is mostly a top-down task loads working
memory, and requires LTM
14Levels of SA (3)
- Level 3 Projection (predicting)
- Requires both Level 1 and Level 2 SA
- Also expertise bound More experienced pilots
spend more time predicting what will occur - Effectively gives the pilot more time for
decision making
The heavier an aircraft, the longer it takes to
respond to a pilots input. In such situations,
Level 3 SA is essential. This is partly why
airlines pick their most experienced pilots to
fly such aircraft.
15Cognition and SA levels
- All three levels require some WM and attention
- Level 1 Speech comprehension, decoding system
interfaces attentional filtering (mostly
bottom-up) - Level 2 Activating mental models and schemata
(mostly top-down) - Level 3 Take LTM information and apply it to
active models (mostly top-down) - So WM and attention loads can be high during SA
- But Experts will use less (better chunking
strategies, better at filtering out irrelevant
elements)
16Information load Example
Data Sources Outside (terrain) Outside
(weather) Outside (aircraft) Inside
(gauges) Inside (maps) Inside (checklists) Inside
(crew) Aural (crew) Aural (control) Aural
(aircraft) Aural (alarms) Haptic
(controls) Haptic (buffets, etc)
17Individual factors in SA (1)
- Limits of attention
- Novices or experts in novel situations require
more attention to be placed on the environment - Information overload can exceed the capacity
miss important relevant information - Giving more attention to one SA task reduces it
on another - Serious problem NTSB review 31 of human
errors due to problems acquiring relevant data
18Individual factors in SA (2)
- Can be overcome by sampling information
- Learn a way to scan the world (avoids fixation)
- Strongly trained patterns become habitual
- Sampling can fail
- Non-optimal strategy (focus on the wrong things)
- Visual dominance (forget other inputs)
- Memory failures (forget relative importance of
elements) - In overload, leave out certain elements
19Individual factors in SA (3)
- Attention limits can be helped by expertise
- Top-down knowledge creates expectations, which
can increase processing speed - BUT if unexpected information occurs, more
likely to make an error (superiority effects) - WM is used mostly for integration and projection
(Levels 2 3) - If much new info is being processed, little WM
will be left for integration (and vice-versa!) - Projection places a particularly heavy load on WM
(need to store multiple states)
20Coping mechanisms (1)
- All is not lost cognitive strategies/structures
exist to deal with this - coping mechanisms (not really)
- Normal info sorting/learning structures
- Generally use previous knowledge to order the
world - Some trained (automaticity) some developed
(mental models) - Generally automatic, subconcious processes
21Coping mechanisms (2)
- Structured knowledge from experience (LTM)
- Schemata, scripts mental models
- fill in missing info (default values)
- Help with structuring comprehension (reduce WM
attention used) - Increase accuracy of predicting the future
- Can be a fuzzy fit
- Almost essential for higher levels of SA
22Coping mechanisms (3)
- Goal-driven processing
- Goals determine how resources are allocated also
- Goals provide a structure in which to process
(allows higher levels of SA) - Automaticity (habitual responses) scripted
- Allows processing with minimal attention
- Can miss novel stimuli
- Safe for routine situations (is there such a
thing?)
23Putting it all together (Endsley 1995)
24Where the problems can occur
25Factors which reduce SA Stress (1)
- Physical stressors
- Noise, vibration, lighting, temperature, fatigue,
jet lag - Social/psychological stressors
- Anxiety/fear, uncertainty, self-esteem, career
advancement, time pressure - Stress effects are complex a little can help
(yerkes-dodson law)
Stress produces many physical and psychological
effects which can reduce SA and undermine a
pilots ability to act correctly.
26Factors which reduce SA Stress (2)
- Why does stress affect SA?
- Attentional narrowing (high arousal/anxiety)
- Oversampling of dominant cues
- Scan patterns disrupted
- Premature closure (hasty decisions)
- Reduction in WM capacity / LTM retrieval (affects
Level 2 3 most) - Training reduces these effects
- Automaticity reduces attention and WM
requirements - More cues, better associations improve retrieval
27Factors which reduce SA under/overload
- Mental overload
- WM and attention limits reached
- Incomplete/erroneous perception
- Stressor (being behind the plane)
- Mental underload
- No active search for info
- Low vigilance/motivation
Air traffic controllers (ATCs) also require high
levels of SA. In busy sectors (such as London,
Atlanta or Tokyo) the volume of traffic can lead
to mental overload.
28Factors which reduce SA bad systems
- The aircrafts interface can present information
poorly - Presenting too much can lead to overload
- Hiding too much can lead to unawareness
- The layout of information can interfere with the
scan - Recently Smart planes (glass cockpit)
- Aware of the information required in a flight
phase - Show what is necessary, but watch for problems in
the background - Alert the crew if a problem exists (speech,
icons, etc)
29Improvements in interfaces
Boeing 737-200Adv (late 1960s)
Boeing 737-800 (Early 2000s)
30Factors which reduce SA - Complexity
- Aircraft keep getting more complex
- Technology demands
- Increases workload more system components, more
interactions - Effectively increases number of goals and tasks
- An expert in these systems will be protected (a
little) - Pilots vary widely on their self-reported
understanding of the systems - A difficult road to becoming an expert!
31Factors which reduce SA Automation (1)
- Habitual procedures can take crews out of the
loop - Reduce vigilance, increase complacency
- Become a passive recipient of information
- Automatic states have bad cognitive consequences
- Pilots are slower to detect problems
- Slower to re-orient after realizing the problem
(schemata de-activation/re-activation)
32Factors which reduce SA Automation (2)
- But automation can help SA also
- Computers can monitor many variables
- Remove unnecessary manual work (navigation)
- Can present many variables already integrated
(for Level 2) - The trick is create systems which aid but do not
promote complacency
33How serious is a failure in SA?
- Jones Endsley looked at accidents in the USA
over a 4 year period (major carriers) - 77 had a substantial human error component
- Of those, 88 due to a failure of SA
- SA failures not even distributed among levels
- Level 1 72
- Level 2 22
- Level 3 6
34Further specific cause
35Teamwork SA in CRM
- Most aircraft are flown by a team
- Do other people increase or decrease SA?
- Spread the work Effectively have more WM and
attention - But Is that enough for collaboration?
Can too many cooks spoil this broth?
36Sharing data
- To work together, people must share info
- Keep mental models etc. aligned
- Each must know what information the other needs
- Must also share higher level understanding, and
projection (level 2 and 3) - Essential shared mental models
- High functioning crews communicate less than low
functioning crews