They chronicle a number of well known characteristics, but this article brings it together in a convenient table:
• Human Performance:
• Human Error:
- -Varies nonlinearly with several factors
- -Follows an inverted U-curve relative to stress
- -Excessive cognitive complexity can lead to task shedding and poor performance
• Human Adaptivity:
- -Lack of inspectability into system operation can induce human error
- -Incompatibility between human processes and machine algorithms can lead to human error
- -Sustained cognitive overload can lead to fatigue and human error
• Multitasking:
- -Adaptivity is a unique human capability that is neither absolute or perfect
- -Humans do adapt under certain conditions but usually not quickly
- -Human adaptation rate sets an upper bound on how fast systems can adapt
- -Tradeoff between human adaptation rate and error likelihood
- -Need to define what is acceptable error rate (context-dependent)
• Decision Making Under Stress:
- -Humans do not multitask well
- -Stanford University’s research findings show that so-called high multi-taskers have difficulty filtering out irrelevant information, can’t compartmentalize to improve recall, and can’t separate contexts
• User Acceptance:
- -Under stress humans tend to simplify environment by disregarding/under weighting complicating factors
- -Reduced ability to process multiple cues or perform tradeoffs
• Risk Perception and Behavior:
- -Overly complex system design can lead to rejection of the system
- -Humans do not have to really understand software/system operation to develop confidence and trust in system
• Human-System Integration:
- -Humans accept greater risks when in teams
- -Humans have a built in target level of acceptable risk
- -Humans are creative but rarely exactly right; however, human errors usually tend to be relatively minor
- -Software/system solutions tend to be precisely right, but when wrong they can be way off
