Abstract
Attentional orienting in our complex visual environment is supported by statistical learning of regularities in the ambient array. For instance, in visual search, detection of a target is faster when the spatial configuration of nontargets is repeatedly encountered, suggesting that learned contextual associations can guide attention (contextual cueing; Chun, 2000). However, scene layouts sometimes change, requiring observers to adapt previous memory representations. Recent work has demonstrated that contextual cueing is initially rather efficient to extract invariant statistical regularities. Yet, a subsequent change of the target location within an otherwise unchanged nontarget layout completely abolishes contextual cueing and the benefits deriving from the invariant context recover only slowly with extensive training (Zellin et al., 2014). This suggests that contextual learning is quite effective initially, but inflexible to incorporate changes. Here, we explored how memory-based attentional guidance is reflected in lateralized event-related EEG potentials. Our results revealed reliable contextual cueing (of 117 msec) during an initial learning phase, which was associated with an increased EEG amplitude for repeated, relative to random layouts in lateralized attention-related components, starting with an early posterior negativity, the N1pc (100–160 msec). In a subsequent test phase, the target location swapped hemifield within an otherwise unchanged context. This target location change led to a substantial reduction of contextual cueing (21 msec). Moreover, the N1pc component was now reversed in repeated layouts, showing a positive-going deflection after the target location change (and no reliable amplitude differences between repeated and random layouts in subsequent lateralized components). The N1pc may therefore be interpreted in terms of an automatic matching process that compares the incoming sensory information with a contextual memory template during early visual processing to guide attention effiently. However, after the change, template matching persists and still provides a bias to the initial target location, thus resulting in attentional misguidance.
Acknowledgement: German Research Foundation (DFG, grant: FOR 2293/1 and GE 1889/4-1)