Fig. 4
From: Attention and stimulus structure interact during ensemble encoding of facial expression
Image distances between ensemble summary reconstructions and corresponding surround faces, central faces, and other face stimuli across ensemble consistency and attentional task. Summary ensemble representations, as revealed by image reconstruction, are impacted by several factors. (A) The central face outweighs the contribution of surround exemplars as the reconstruction is closer to the former (via an L2 pixelwise metric in arbitrary units); consistency reduces the distance between the reconstruction and both central and surround faces; focused attention increases the contribution of the central face and, for inconsistent ensembles, decreases that of the surround (between-experiment comparisons are shown below the left bar plot). (B) Ensembles sharing the central face with single faces are less clustered in face space during distributed attention (left) compared to focused attention (right). The plots depict two dimensions of face space, averaged across participants, for one facial identity during Experiments 2 and 3. Corresponding central face stimuli, pixelwise ensemble averages and ensemble reconstructions are shown for focused attention (right)—the central face has minimal impact on the pixelwise average of the stimulus, but it appears to dominate the ensemble representation (ns: nonsignificant, *p < .05, **p < .01, ***p < .001, Holm-corrected; error bars indicate ± SE across participants; PVE percent variance explained).
