ABSTRACT
Most theories of visual search maintain that attention is selectively tuned to the attributes of the search target (e.g., orange). Conversely, according to the relational account, attention is biased to the relative feature of the target (e.g., redder). However, previous studies that supported the relational account mainly measured mean response times. Hence, the results might not reflect early, perceptual mechanisms (e.g., signal enhancement) but later, decision-based mechanisms (channel selection). The current study tested the relational account against feature-specific theories in a spatial cueing task, in which the targets were backward-masked, and target identification accuracy was measured. The first experiment corroborated earlier results, demonstrating that relational effects are due to signal enhancement. In the second experiment, we chose highly discriminable colours along the blue–red continuum, and obtained results that were more consistent with broad feature-specific rather than relational tuning. The implications of these findings for current theories of attention are discussed.
Disclosure statement
No potential conflict of interest was reported by the author(s).
ORCID
Josef G. Schönhammer http://orcid.org/0000-0002-6635-2100
Notes
1 We display the percentage of errors instead of the usual percentage of correct responses to make the results more easily comparable to those from RTs experiments.
2 The validity effect for the cue with the more extreme colour (Cue 1) tended to be larger than that for the less extreme cue (Cue 2), F(1, 18) = 3.8, p = .064, ηp2 = 0.17. It is currently unclear why these validity effects differed in size; however, these differences in the magnitude of validity effects appear to be unreliable across studies. In studies that measured validity effects on mean RTs, no such differences occurred (Becker et al., Citation2013; Harris et al., Citation2013), or the validity effects for the more extreme cues were smaller than for the less extreme cues (Schönhammer, Grubert, Kerzel, & Becker, Citation2016). Moreover, inverse validity effects tended to be larger for Cue 3 than Cue 4, F(1, 15) = 3.4, p = .073, ηp2 = 0.16, but also this difference was unreliable across studies.