Abstract
When a 2AFC preference task is performed within the same object category; face (FC) or natural scene (NS), a distinctive segregation occurs - a familiar FC is chosen more than a novel FC across trials, whereas a novel NS tends to be chosen more than a familiar NS (Shimojo et al., VSS'07, '08). How would the preference choice change if a FC is presented with a surrounding NS (just like a typical commercial advertisement
Thus we combined a central FC [old (o) or new (n)] and a surrounding NS (o or n) as an integrated single stimulus, for which subjects were asked to rate attractiveness in a 7-point scale. We selected 21 pictures, excluding too high- and too low-rated, from a pre-rated set in each category and used the 11th as a repeating (o) stimulus. There were two experiments with different instructions: 1) Rate the attractiveness of the entire picture, or 2) Rate the central face only, neglecting the surrounding
The results of Exp 1 clearly betrayed the prediction by simple summation of attractiveness across objects where the FCo-NSn combination would have the highest slope of accumulated attractiveness. Instead, the results were: FCn-NSo [[gt]] FCo-NSn [[gt]] FCo-NSo = FCn-NSn in the order of slope of the rating X trial number plot. The results of Exp 2 were more consistent with the simple summation model: FCo-NSn = FCn-NSn [[gt]] FCo-NSo [[gt]] FCn-NSo.
The 1) results suggest that attractiveness is integrated in a non-additive way across object categories and figure/ground. The 2) results can be accounted for by two factors; Familiarity preference in FC at an explicit level, and novelty preference in NS at an implicit level. These results together indicate a high degree of nonlinearity and implicit processing in memory-based attractiveness integration.