Abstract
In visual search, the tuning of attention depends not only on the target feature but also on the nontarget ("context") features. How do target and nontarget features interact? The relational theory proposes that when target and nontarget features remain fixed, attention is tuned to relative feature properties (Becker, 2010). In Experiment 1, we sought evidence for the theory using previously untested colors. For example, the target was greenish-yellow and the nontargets yellow. Following the theory, attention should be tuned to "greener". In line with the theory, spatially unpredictive precues captured attention only when they were greener than the context. For instance, a green cue in a greenish-yellow context captured attention, as evidenced by spatial cueing effects on response times. Simultaneously, a greenish-yellow cue in a green context (yellower) did not capture although it shared the target feature. However, target and nontarget colors were very similar. Thus, attention might not have been tuned to relative attributes, but to exaggerated feature values to improve internal target-nontarget discrimination (e.g., Navalpakkam & Itti, 2007). For example, when the target was greenish-yellow and the nontargets yellow, attention might have been tuned to green instead of greener. Following such an alternative account, attention should not be tuned to exaggerated features when target and nontargets are dissimilar, but to the exact target feature. In Experiment 2, the target was again greenish-yellow but the nontargets were red, for example, which are highly dissimilar colors. The cues were the same as in Experiment 1. In line with the relational theory, cues only captured attention when they had the same relative attribute as the target, independently of whether they had the exact same color as the target. The results indicate that attention can be tuned to the relative target-nontarget attributes even when target and nontargets are relatively easy to discriminate.
Meeting abstract presented at VSS 2016