Аннотация:Recent theoretical developments, such as Reverse Hierarchy Theory (Hochstein & Ahissar,
2002) and reentrant processing theory (Di Lollo et al. 2001) have led to new interest in top-down influences upon visual processing. Here we address the “word superiority effect” (WSE; Cattell, 1886), which actually describes two similar phenomena: it is easier to
perceive letters which are part of words, relative to isolated letters, and it is easier to perceive letters in words relative to nonword strings (e.g. Reicher, 1969). Research on this phenomenon has been concerned primarily with the effect of lexical organization on the
accuracy of visual letter recognition. WSE has been shown to reduce lateral masking, or
crowding (Fine, 2001) and metacontrast masking (Luiga et al., 2002). We propose that the
WSE determines the units that attention can act upon. In the temporal domain, there are data demonstrating that the WSE attenuates the attentional blink (Falikman, 2002). However, there has been no work in the domain of spatial attention.
Here we utilized the visual search paradigm to test for an influence of WSE on spatial
attention. In visual search, observers are asked to find a specified target item in a cluttered background of distracting items (Wolfe, 1998). Visual search is a widely used experimental technique for determining how easy or difficult it is to direct attention to a particular class of stimuli (e.g. Horowitz & Treisman, 1994).
We asked observers to report the presence or absence of a target letter (e.g. “T”) among letters organized into strings which formed either Russian five-letter words (with no letters repeated) or nonpronounceable nonwords, compiled of the same words. The target letter was embedded into a string which was a word or nonword, while distractor strings could also be words or nonwords, thus making four experimental conditions. Set size was varied from 3 to 7 to 10 strings, and targets were present on 50% of trials. Strings were presented on an imaginary circle with a radius of 9,7°. On each trial, strings appeared in a random set of locations spaced equally around the circle. In order to avoid the influence of the left-to-right reading strategy, the target letter was always presented in the 2nd, 3rd, or 4th position within the string. Subjects were instructed to respond as quickly and accurately as possible.
We hypothesized that organizing letters into words would alter the units of the guidance map (Wolfe, 1992), so that only words, and not their constituent letters, would guide attention. We therefore predicted that search would be more difficult when the target is embedded in a word rather than a nonword, because observers must search not for a target letter but for the larger category of “words containing target”.
Search was highly inefficient: target-present RT x set size slopes exceeded 120 ms/item in all conditions. In contrast to our hypothesis, search was not least efficient when the target was embedded in a word. Target type did not affect target present slope (F(1,5) < 1). In general, itwas easier to reject word distractors than nonword distractors (137.5 ms/item vs. 172.4 ms/item, F(1, 5) = 26.7, p = .004), suggesting that words facilitated recognition of component letters, in line with previous research. More interestingly, we observed an interaction between target string type and distractor string type (F(1, 5) = 20.0, p = .007). For a given distractor string type, search was easier when the target was of the opposite type (word among nonword, 154.3 ms/item vs. nonword among nonword, 190.5 ms/item; nonword among word, 124.1 ms/item vs. word among word, 154.3 ms/item). This pattern may reflect a weak preattentive sensitivity to string structure. Words do not qualitatively differ from nonwords in their ability to attract attention. However, the visual system does seem to represent the structure of strings, even outside the focus of attention.