Computational Approaches To Vision Research Flashcards
What did Alhazen find out about the visual system?
Had similar view to what we know now, how information goes from one eye to the opposite hemisphere, and how large proportion of brain is to do with vision
What did Hering suggest?
Introspection - can’t see reddy green or bluey yellow
Opponent process theory - predicted 3 colours from the cones must somehow be processed later to be either blue or yellow, or green or red - can’t be both
- > cones first stage of colour vision
- > opponent cells come after
What are the problems with introspection?
Introspection only works for the really insightful/lucky
Also needs verification from other methods
Not everyone’s perception is the same eg. colour blind people, stereoblind people (can’t see 3D)
What is a good test of vision?
Forced choice - no problems eg. demand characteristics or criterion (some people have different levels of certainty)
Make people make a decision - 2AFC (2 alternative forced choice), 4AFC etc
Why are models of behaviour important?
Predictions about what kind of behaviour you’ll get
Tend to be mathematical in vision - precise estimates of how performance will change
Predict reaction times, where people look (saliency map) -> can measure these
Why should we use eyetrackers?
Eye movements tend to give insight to attention
What do uses of models show about humans?
Can show people are bad at recognising shadows but computers don’t ignore them -> can recognise fake/doctored images
