Please enable JavaScript.
Coggle requires JavaScript to display documents.
MODELS OF PATTERN RECOGNITION (feature analysis / detection (evidence for…
MODELS OF PATTERN RECOGNITION
the template approach
templates
: stored models of all categorical patterns
problems
equiv patterns are not identical
can recognize degraded patterns
have to learn lots of templates
feature analysis / detection
patterns = features (distinct, separate parts)
small # of features w / great combo potential
ex. pandemonium lol
evidence for feature detectors
neurological
behavioral
problem
: we need top down processing which isn't in template models
connectionism
: neural net models, parallel distribution processing
contains a system of interconnected nodes w / knowledge represented by interconnections b/t units
massively parallel
top - down + bottom up
interactive activation model
word superiority effect
: when words or letters are briefly flashed, Ps find it easier to ID a letter in a word context than in nonword or indiv.
recognition by components
geons
: building blocks of visual objects
recognize geons + where they join
data on non recoverable objects are hard to ID
continuous line > deleted junctions
junction formation = impt
problems
super bottom up, limited
doesn't include influenced of embodied cognition
whole vs part perception
components + global patterns parts of brain are different
