| |
Abstract:
Behavioral and neuroimaging experiments suggest that letter
recognition is mediated by a second stage of processing that
follows initial processing of simple features. To dissociate these
two stages behaviorally we used a modified crowding paradigm.
Crowded, or closely positioned, letters are more difficult to
identify than widely spaced letters. In our experiment, subjects
were presented with either a standard character string (crowded
condition) or a modified string in which characters were displaced
up or down from their original location in an alternating fashion
(uncrowded condition). Displays were in the periphery while
subjects maintained central fixation. Subjects were instructed to
identify which of two presentations contained either a target
letter or a target feature. Contrast thresholds were measured for
each condition using an interleaved staircase procedure. Subjects
performed equally well in the uncrowded letter, the uncrowded
feature, and the crowded feature conditions. However, subjects were
much worse in the crowded letter condition showing near-chance
performance. Comparing the large crowding effect for letters to the
nonexistent effect for features suggests that letter identification
occurs at a distinct stage of processing. This stage may be where
the relative locations of line segments are determined.
|
