Abstract
Recent research with bird experts indicates that their subordinate-level categorization of species is influenced by image color (Hagen et al., 2014) and spatial frequency characteristics (Hagen et al., 2015). The present research sought to examine how these same characteristics influenced learning and subsequent discrimination of species in novice participants trained at the subordinate species level (e.g., “Nashville Warbler” vs. “Wilson’s Warbler”). Six training days included unaltered images (color congruent, all spatial frequencies) of 10 species of birds. During the pre- and post-training tests, images were manipulated to examine the influence of color (congruent, incongruent, or grayscale) or spatial frequency (all frequencies, high spatial frequencies above 8 cycles/image (cpi), or low spatial frequencies below 8 cpi) during a serial species-matching task. Results suggest that, prior to training, congruent and incongruent color did not differ and generally improved discrimination relative to grayscale images. Presumably these pre-training effects reflect the importance of color in part segmentation for novices. However, after training, participants performed best with congruent colors relative to the incongruent and grayscale images. These findings highlight the importance of color knowledge in correct subordinate-level identification. Unlike color, the effects of spatial frequency did not change with training. Training improved performance for all image types. However, accuracy before as well as after training was ordered: all special frequencies > high spatial frequency > low spatial frequencies. In general, post-training results were similar to real-world bird experts who perform best with congruent colors and all spatial frequency images, and who rely on high spatial frequencies more than low spatial frequencies for discrimination. The present results suggest that color congruency effects were dependent on learning, whereas the effects of spatial frequency did not change with learning.
Meeting abstract presented at VSS 2015