Psychological Foundations
590H -- Spring 1995
Warning: these notes are really sketchy ...
Card, Moran, and Newell paper: "The Human Information-Processor
[In addition to its discussion of human performance as it is relevant to
HCI, this paper is striking as an exemplar of the cartesian/rationalist
tradition.]
The human information processor is modelled as a set of memories and
processors, and a set of principles of operation.
subsystems:
- perceptual system (sensors and associated buffer memories)
- motor system
- cognitive system (including working and long-term memory)
memory parameters: storage capacity; decay time of an item; main code type
Model Human Processor - principles of operation (see chart on page 27)
Human Performance Examples
Limits based on Perceptual Processor Cycle Time:
- Moving Picture rate -- calculate how many frames per second are needed
- Morse Code listening rate -- calculate the fundamental limits of how
fast a human can read Morse code (based on the fact that stimuli that fall
within one cycle time fuse together)
- perceived causality (pool ball simulation)
- reading rate
Motor Skill - Fitt's Law
examples of key layout on a calculator; different typewriter keyboard
layouts
reaction times
searching long-term memory
rationality princples (really this seems to be a description of
goal-directed problem solving -- see cryptarithmetic examples)
GOMS Model (Goals, Operations, Methods and Selection rules) -- further
family of models, which enables various qualitative and quantitative
predictions to be made about user performance
a problem with such models is that their development takes a long time.
Just about the a cognitive model of performance with line editors was
developed, line editors were superceded by screen-based editors.
other interesting example (not in readings) -- perception as it relates to
color graphics and 3d graphics
Additional Aspects of Human Memory
(from Don Norman, The Design of Everyday Things)
categories of information stored in long term memory:
- memory for arbitrary things
- memory for meaningful relationships
- memory thorough explanation
other categorizations:
- procedural vs. declarative representations
- knowledge in the head vs. knowledge in the world
Kinds of Human Error; Designing for Error
- capture errors
- description errors
- data-driven errors
- associative activation errors
- loss-of-activation errors
- mode errors
detection takes place when there is feedback
Preece distinguishes slips and mistakes
Lessons for Design:
- reduce errors by minimizing the number of modes, or using physical
constraints, or at least making the modes visible
- better to eliminate irreversible actions than to request confirmation
Cognitive Frameworks for HCI (from Preece book)
various moves away from this information processing framework in HCI:
- compational approaches
- connectionist models
- situated actions
- distributed cognition (examples from nuclear power plant control,
piloting an aircraft, interactive video)
visual perception - affordances (see scrollbar example on p 81; Gaver paper)
coding methods (see page 89)
color pollution