In the present study we manipulated two top-down information types: target template and scene context. Previous studies have focused on the relationship between salience and scene context (Torralba et al.,
2006), or salience and target template information (Kanan et al.,
2009), or all three information types at once (Ehinger et al.,
2009), but have not focused specifically on the target template–scene context relationship. Understanding this relationship is important: there is growing evidence that top-down information dominates real-world image search processes, such that the influence of low-level salience information on search guidance is minimal (Ehinger et al.,
2009; Einhäuser, Rutishauser, & Koch,
2008; Einhäuser et al.,
2007; Foulsham & Underwood,
2007; Henderson, Brockmole, Castelhano, & Mack,
2007; Henderson, Malcolm, & Schandl,
2009; Tatler, Baddeley, & Gilchrist,
2005; Tatler & Vincent,
2009; Turano, Geruschat, & Baker,
2003; Zelinsky,
2008; Zelinsky, Zhang, Yu, Chen, & Samaras,
2006). Even the high intersubject agreeability of the first saccade—which had been thought to be a result of the visual system selecting low-level salient areas to fixate (Carmi & Itti,
2006; Parkhurst, Law, & Niebur,
2002)—has since been suggested to be derived from common high-level knowledge strategies (Tatler et al.,
2005). Low-level salience's minimal influence on real-world image search is most likely due its independence from a task goal while the placement of eye movements is heavily task dependent (Castelhano, Mack, & Henderson,
2009; Foulsham & Underwood,
2007; Land & Hayhoe,
2001; Land, Mennie, & Rusted,
1999; Rothkopf, Ballard, & Hayhoe,
2007; Yarbus,
1967). For these reasons, we concentrate on top-down information types in the current study as they will reveal the most about the guidance of attention and eye movements in real-world search.