perception Flashcards
what is a local level of processing
details and parts of the whole
what is the global level of processing
the whole picture
what is viewpoint invariant
object recognition does not depend on viewpoint
doesn’t matter what angle looked at from, processes the same
what is viewpoint dependent
object recognition does depend on viewpoint
within category discrimination
what are the 3 different types of pattern recognition theories
template
prototype
feature
template theories
selfridge and neisser 1960
selfridge and neisser 1960
template theories
template theories
explanation
patterns in real world recognised but matching those stored in templates
prototype theories
rosch 1975
rosch 1975
prototype theories
prototype theories
explanation
we store prototypes instead of templates
the most typical member of a category
feature theories
jaun and duin 2004
jaun and duin 2004
feature theories
feature theories
explanation
patterns consist of set of features of attributes
feature theories
neisser 1964
find the letter z in 2 lists
list 1 was more difficult to identify the letter z because it shares more features with the other letters e.g. straight lines
navon 1977
global and local
global and local
navon 1977
navon 1977
study
performance speed was slowed when asked what the small letters was when the large letter differed
however
decision speed with the large letter was not influenced by the small letter
i.e. when the small letters differed, decision speed for the large letter was not slowed
global affects local but not visa versa
dalrymple, kingstone and handy 2009
replicated navon
replicated navon
dalrymple, kingstone, handy 2009
replicated navon
study
when the small letters were densely packed, it replicated Navon but when the spacing between the letters were bigger, processing was faster at the local level than the global letter
biederman 1987
recognition of components
recognition of components
biederman 1987
recognition of components
explanation
we can construct everything in our world form a set of 36 shapes called geons
geons are view-invariant
they are easily discriminable from each other
complex shapes are more than one geon connected
4 steps pf object recognition
edge extraction
detection of non-accidental properties/parsing of regions by concavity
determination of components
matching components to object representations
convex vs concave
the invariant properties can still be detected even why only parts of edges are visible
provided the concavities are preserved, the object can still be identified
what is apperceptive agnosia
a failure in recognition due to a failure of perception
cannot copy shapes when asked to do so