Abstract
Distinct orientation subbands (like spatial frequency subbands) make differential contributions to facial recognition and emotion identification in faces. In particular, horizontal orientation subbands are essential for facial emotion recognition in adults and child observers (5-10 years of age). In prior work, however, we have shown that emotion recognition from body images depends on different orientation information: Adults recognize emotion more effectively from body images containing vertical orientation energy. Given that we have also found that the tuning of facial emotion recognition to diagnostic and non-diagnostic subbands changes during middle childhood, we chose to investigate children's sensitivity to diagnostic orientation subbands for body emotion recognition. We recruited 30 children (5-10 years old) to perform a simple emotion categorization task using body images with the horizontal or vertical orientation energy information filtered out. We compared younger children's performance (5-7 years) to older children (8-10 years) to determine how the use of vertical vs. horizontal subbands differed with age. Our older group of children performed better at the task in general (p< 0.001), but both age groups exhibited poorer sensitivity to happy vs. sad emotion in the horizontal condition (Mean d' = 1.63) versus the vertical (M = 2.40) and both subband (M=2.41) conditions (p < 0.001), and we observed no interaction between filter orientation and age group. Likewise all participants were slower at the task when given only horizontal information relative to the other two conditions (p< 0.001), suggesting that the sensitivity results do not simply reflect a speed-accuracy trade-off. We propose, based on these results, that an adult-like bias favoring vertical orientation energy for body emotion recognition is present during early childhood. Also, compared to our prior work describing children's abilities to recognize emotion in filtered face images, young children appear to be better able to recognize body emotion using non-preferred orientation information.
Meeting abstract presented at VSS 2016