Abstract
A number of theories suggest that one can bias visual search toward goal-relevant stimuli by representing a template of the search target in working memory (WM). Recently, Woodman and colleagues used ERP evidence to argue that the search template was handed-off from WM to Long-Term Memory (LTM) when the target remained constant for a number of trials. Here we provide behavioral evidence of this hand-off. Participants’ eye movements were monitored as they performed a difficult visual search among oriented bars. Cues indicated the orientation of the bar that would contain the target (if present). Eye movements were biased toward bars of the cued orientation. When the cued orientation remained constant for a block of trials, this bias appeared to be based on a categorical (steep/shallow) representation of the target orientation rather than a visually specific representation. For example, when the cue was +22.5° from horizontal (shallow) the eyes fixated bars that were -22.5°degrees from horizontal (shallow) much more frequently than bars that were -22.5° from vertical (steep), even though both these orientations were 45° from the cued orientation. When cued orientations were randomly interleaved, this categorical bias was greatly reduced (but did not completely disappear), suggesting that the guidance was based on a more precise visual representation. These results provide a potential behavioral marker of the shift from WM to LTM guidance, and suggest that the nature of the representation is different between these two memory systems. The LTM system seems to maintain a more abstract, categorical representation, while the WM system maintains a more specific, visually-based representation.
Meeting abstract presented at VSS 2013