Purchase this article with an account.
Jun Saiki; Features underlying visual search asymmetry revealed by classification images. Journal of Vision 2005;5(8):780. doi: 10.1167/5.8.780.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The underlying mechanism of visual search asymmetry remains unclear. Many accounts attribute the asymmetry to features used in visual search, but features are usually not clearly defined. For example, in a case of O and Q, asymmetry may reflect different target-defining features (e.g., circle and crossing), or asymmetry in search for presence and absence of a common feature. To overcome this limitation, classification image technique was employed to estimate which visual features were used in searching for O and Q. Three observers viewed displays with 1 target and 3 distractors, either O (1.9 degree diameter ring) or Q (O plus 0.9 degree vertical bar), embedded in white Gaussian noise, each located on a corner of an imaginary diamond at 3.8 degree eccentricity, and were asked to localize the target. Classification images were constructed from error trials, where the target and selected distractor images were treated as miss and false alarm images, respectively. Both Q- and O-target conditions revealed the same feature, the vertical bar, and the asymmetry reflects the amplitude of the feature, which is larger for O-target than for Q-target. This amplitude difference reflects the strength of noise canceling the bar feature. Regardless of the target, only noises in Q stimuli were correlated with response. Thus, 3 Q's in O-target condition is more likely to have a noise strong enough to lead to an error, than a single Q in Q-target condition. Furthermore, singleton search task showed results similar to the target defined version, suggesting stimulus-driven mechanisms. Search asymmetry between O and Q does not reflect different target-defining features. Also, contrary to the idea that search for feature absence is more difficult than search for its presence, the presence of a feature is more vulnerable to noise than its absence. Difficulty in search for O is not due to search for feature-absence itself, but due to many error-prone feature-present distractors.
This PDF is available to Subscribers Only