Object/Pattern Recognition- Theories:o Template theories Templates stored in memory, math input to template, whichever matches best Too unwieldy, too much variance for this theory to be trueo Feature theories Object perception is based fundamentally on simple primitive features Whatever the features are that the perceptual system registers, we work from the primitive base, work up from there Certain set of features together= object perceivedo Feature Integration Theory Triesman and Colleagues First feature detection operates, then focal attention Feature detectors- Detects and identifies all features present in the array- Operate in parallel (automatic)- Operate without attention- Doesn’t identify where the features are present, doesn’t connect anything together Focal attention- Identifies features in space, connect features that are present in the same space- Operates serially (control)- Demands attention Empirical evidence- use visual search task (which is able to discriminate between serial and parallel processes) Constructs stimulus materials, differentiates targets and distracters,features that distinguish the 2, difference is 1 simple primitive feature- color If target is black and all distracters are white, can do visual search test without having to rely on feature Single feature search: searching for the presence of a single feature, don’t have to wait for a process to finish before u can respond- flat parallel function Focal attention is serial but unconscious (Shiffrin and Schneider can’t handle) Conjunction search: look for black circle, being able to put the blackness and circleness together demands focal/ serial attention- linear increasing 2:1 slope ratio Masking- replace stimulus with random noise, flash a visual array,try to disrupt processing so that feature detection has operated by focal has not, then ask what was present in the array Seems to be a drive in the human info processing situation that makes sense of the info we have Simple perceptual level- if experimental paradigm is successful at distracting, we would have floating pieces, but people strive to make sense of them so they report seeing actual features She flashes a visual array, people reported seeing s’s, |s, $s, then angles, /s, and not triangles- Why don’t they see triangles?- There must be a feature triangles have that’s not present in the array- Triangles are closed features, maybe the property of being aclosed figure is a simple primitive feature- Add circles into the mix, people report seeing: angles, /s, circles, and TRIANGLES- World recognitiono World superiority effecto Trace- Shape recognitiono Biederman’s theoryo Empirical evidence9/25/12- Some features functioned as primitive is some environments and some don’t - Some reduce reaction time in visual search tests, depending on what other features were present- Have 2 processes: feature detection, locates features in space- Conjunction feature search: features distinguished targets from distracters- Focal attention joins all features into an object- Feature integration theory- building up from features to more complex objects- If in order to identify the letters you need knowledge of word, how can you make knowledge of word if you haven’t identified the letters yet- Reicher (1969)- 2AFC (2 alternative forced choice). o Show frame 1 and 2, frame 1 flashed very quickly, subject couldn’t processo Present string of 4 letters in first frame (CORD)o Frame 2 comes up next, subject is asked to choose which of the two letterswas in frame 1 (DK). o Letters appear right over letter they’re being asked to choose. Circles are “visual masks” which disrupts the visual sensory storeo Subject has to respond whether they saw D or Ko Letters appear equally throughout tests, cant predict which its going to beo What if stream of letters doesn’t create word? Performance shouldn’t be asgood as if it did make up a wordo Independent variable: whether or not letters make up word, whether or notuse higher up word level informationo If its all bottom up- doesn’t matter if word thereo Dependent measures- reaction time of identifying letter, accuracy- probability of a correct responseo Letter recognition is a completely stimulus driven process- from features to letters to words, before word level information has come into playo Word level information plays a role- information about the more complex object is used to identify the parts of the objecto NEXT OCRD is flashed o Word provides advantage that letter doesn’t haveo Single letter provides advantage that the word doesn’t haveo Words work in the identification in letters- make use of word level information in identifying letters Performance is better when part of word because word informationis present in this case, words better than non words, individual letters better than non wordso Words don’t work- all bottom up Words and non-words should be the same Should do better with individual letters because all attention is thereo Single vs. non word Individual letter better (which both theories say)o Word vs. non word Words are faster than non words (words work theory)o Word vs. single Words