| Literature DB >> 1837304 |
Abstract
Strings of 4 unrelated letters were backward masked at varying durations to examine 3 major issues. (a) One issue concerned relational features. Letters with abnormal relations but normal elements were created by interchanging elements between large and small normal letters. Overall accuracy was higher for letters with normal relations, consistent with the idea that relational features are important in recognition. (b) Interpattern relations were examined by mixing large and small letters within strings. Relative to pure strings, accuracy was reduced, but only for small letters and only when in mixed strings. This effect can be attributed to attentional priority for larger forms over smaller forms, which also explains global precedence with hierarchical forms. (c) Forced-choice alternatives were manipulated in Experiments 2 and 3 to test feature integration theory. Relational information was found to be processed at least as early as feature presence or absence.Mesh:
Year: 1991 PMID: 1837304 DOI: 10.1037//0096-1523.17.4.924
Source DB: PubMed Journal: J Exp Psychol Hum Percept Perform ISSN: 0096-1523 Impact factor: 3.332