PSYCH 120A Lecture Notes - Lecture 6: Bigram, Subjective Constancy, Visual Search
Document Summary
For literate adults, word recognition is trivially easy. Single letters are more easily broken down to its features but they are not actually processed as quickly/well as whole words. Word superiority effect: words are easier to perceive than letters alone or nonword strings. Feature detectors: tiniest unit, features that make up a letter. Each bigram has a resting baseline level of activation depending how common it is in the english language. More frequent higher level of activation. Over-regularization: the network is biased to favor frequent letter combinations. When in doubt, stimulus is degraded, ambiguity assume the more regular pattern is the one that you are seeing. Helpful most of the time despite some errors faster processing for more frequent words. Distributed knowledge: knowledge of the whole stimulus is represented across many units in the network. Need to combine pieces of info that each detector has to get a holistic sense of the word.