Fig 1.
Examples of stimuli used in Experiment 1.
Each of the four face-scene congruency combinations are represented: Male face in a male scene (top left); female face in a female scene (top right); male face in a female scene (bottom left); female face in a male scene (bottom right).
Fig 2.
Mean inverse efficiency scores, response times and accuracy for congruent and incongruent face-scene conditions in Experiment 1 (face categorisation), Experiment 2 (scene categorisation), Experiment 3 (face detection), and Experiment 4 (face detection with scene context preview).
Above each plot, the results of paired t-tests between conditions are reported. Error bars represent within-subject variability via 95% Cousineau-Morey confidence intervals [38]. *p < .05, ** p < .01, ***p < .001.
Fig 3.
Image saliency for face-present and face-absent scenes (left-hand images) were compared using course-grained (central images) and fine-grained (left-hand images) saliency maps.
Face saliency scores were computed by subtracting the sum of pixel values for the face region (red box, lower images) from the corresponding region in face-absent scenes (red box, upper images). Face saliency was highly correlated between congruency conditions for both course-grained (a) and fine-grained (b) methods. Furthermore, face detection performance in Experiments 3 and 4 (Inverse Efficiency Scores [IES]) showed negative correlations with face saliency scores for both course-grained (c and d) and fine-grained (e and f) saliency methods, indicating that salient face regions were detected more efficiently.