Function allocation principle

Mechanize whatever functions that

Principles of HMS Design machines can do.
Mechanize functions with which
machines are better in abilities, leave
others to humans with which humans
K. Furuta (Univ. of Tokyo) are better.
Share functions between humans and
machines in a proper manner.

Fitts list Drawbacks of Fitts list

Humans are better
• Sense low levels of certain
kinds of stimuli.
• Detect stimuli against high
noise level background.
• Recognize patterns of
complex stimuli.
Machines are better Now there are overlaps of functions
• Sense stimuli outside of that both humans and machines can do
the human sensitivity. well.
• Deductive reasoning such
as classification of stimuli. Humans are flexible but inconsistent,
Monitor prespecified
•

(infrequent) events. while machines are consistent but

• Sense unusual and • Store coded information inflexible. The both are complementary
unexpected events. quickly and accurately. rather than substitutive.
… …
— H. Jordan

New principle of allocation Progress of automation

We can mechanize a function when Operator Operator Operator Operator

The function can be mechanized well

enough with the current technologies. Display
devices
Operation
devices
Display
devices
Operation
devices
Display
devices
Operation
devices
Display
devices

The reliability of machines is better than

humans in terms of the function. Automatic
controler
Automatic
controler
Automatic
controler

The reliability of the total human-machine

system is better if the function has been Process Process Process Process

mechanized.
Automated Supervisory
Manual Autonomous
manual control

1
Degree of automation Abstraction Hierarchy
1. The computer offers no assistance, human must do it all.
Total Sub- Function Sub-As- Com-
2. The computer offers a complete set of action alternatives, and System System Unit sembly ponent
3. narrows the selection down to a few, or why
Purposes
4. suggests one, and
Information what
5. execute that suggestion, if the human approves, or Measures
6. allows the human a restricted time to veto before General how
why
automatic execution, or Functions
7. executes automatically, then necessarily informs the human, or Physical what
Processes
8. informs him after execution only if he asks, or Material Form how
9. informs him after execution, if the computer decides to. Configuration
10.The computer decides everything and acts automatically,
ignoring the human. J. Rasmussen
T. Sheridan

Task and task step Description of task step

Task
Goal include may interfere
A set of well-structured actions that is expected to
achieve a designated goal. Execution achieved by Effects Side-effects
conditions
Task step
A group of actions that can be thought of as an require Task step produce

elementary unit of a task. Pre-

require
Function conditions
imply
Resource Time
Behavior of the system that is implemented to
Subgoals Task body constraints constraints
achieve a designated goal.

Goals-Means Task Analysis Goals-Means Task Analysis

begin Paint the roof end Prepare paint

begin Climb on the roof Paint the roof end
Prepare tools
(a) Generation of the main task.

Concurrence constraint:
(c) Order parallel task steps.
Achievement of the step makes
Climb on the roof it impossible to execute other steps.

begin Prepare paint Paint the roof end Prepare paint Climb on the roof
Prepare a
begin Climb the ladder Paint the roof end
ladder
Prepare tools Prepare tools

(b) Expansion of subtasks. (d) Embody task step.

2
What should be assessed? Workload
Number of concurrent goals Physical workload
Goals that should be pursued in parallel. Level of physical-mechanical activities of
Available time the skeletal-muscular system.
How long will it take to finish the task. Measured by metabolism, oxygen
Workload consumption.
Level of physical and mental activities of humans. Mental workload
Cooperative relations Level of mental-cognitive activities of the
Points that require task cooperation and central nervous system.
communication between individuals.
Objective measure VS subjective measure

Yerkes-Dodson’s law Assessment of MWL

Subjective Dual task Physiological Task analysis Observation

Task performance

Selectivity ○ ◎ △ ○ ×
allowable level Reliability ○ ○ × ？ ◎
Sensitivity ○ △ ○ ？ ？
Theoretical △ ○ × ○ ○
Interference ◎ × ○ ◎ ◎
Real time × ○ ◎ × ○
too low appropriate too high Easiness ◎ ○ △ × △

Mental workload

Optimization of MWL Optimization of MWL

Automation Restricting the amount of information
Only important information is to be
Cognitive support tools displayed.
Procedure manuals, check-lists, and Prioritization of information.
various computer based support tools.
Changing the style of presentation
Training
Massive information can be delivered in a
Change the mode of behavior from few chunks of stimuli.
knowledge-based or rule-based to skill-
based. Multi-modal presentation.

3
Example of display style
A B C D
Convensional SSSI (5 chunks)
Types of S-R compatibility
A Degree of correspondence between
stimulus (S) and response (R)
E B Movement compatibility
Spatial compatibility
Modality compatibility
Conceptual compatibility
D C
E
Plygon display (1 chunk)

Examples of S-R compatibility Prioritization of information

up
Display of unimportant information should be
down
suppressed or done by request.
Important information should be displayed in
the central position to draw attention.
Movement Compatibility The color, size, or form of display should be
changed depending on the importance.
B B
A C
D A C
D
In CRT, information should be ordered in
accordance with the importance.
A B C D A B C D

Spatial Compatibility

Alarm suppression Layout design

Alarm avalanche Physical-perceptual features of a human
Too many alarms may annunciate at the Environmental factors
same time so that a human cannot Task related characteristics
recognize what actually has happened.
Individual characteristics
Alarm suppression rules Important or frequently used elements should be located
at the central and accessible position.
Suppression by mode Relations between different items
Suppression by propagation Related elements should be located close together.
Suppression by level

4
 Eye-movement link between
Link analysis aircraft instruments

Assessment of affinity among interface elements. 2%

Temporal affinity 10%

Topological or positional affinity
Structural affinity 16% 21% 4%

Etiological affinity Cross pointer Air speed Directional gyro Gyro hirizon
Engine
instrument
Teleological affinity

Mapping the conceptual space of affinity on an 9% 3% 4% 3% 11%

Euclidian space.
Manual operation of a link graph.
Altimeter Tun and bank Rate of climb
Analytical methods of systems engineering. 2% 2%

2%

Consistency of mental models Example of model mismatch

Control A
Design Model User’s Model
Freezer
Designer User
Cold Air
Valve Cooling Unit

System
Fresh Food
System Image Control B