Speaker
Description
When we observe an object, our visual system identifies its general shape and adds specific details to form a coherent representation. This coarse-to-fine approach involves quick processing of low spatial frequency content (LSF) to generate a basic template, which aids the integration of the more detailed high spatial frequency information (HSF). In three experiments, we explored how LSF and HSF are integrated, leveraging on the face inversion effect, whereby inverted faces are more difficult to recognize than upright ones. In Experiment 1, eight participants determined the gender of unfamiliar faces with different orientation. Results showed that identification accuracy for HSF faces was lower than LSF or unfiltered faces. In Experiment 2, fourteen participants matched two familiar faces displayed in succession. The template and the sample shared either the same SF (congruent) or had complementary SF (incongruent). In congruent conditions, HSF templates tend to be better than LSF templates. However, in the incongruent conditions mapping a LSF sample on an HSF template was more effective only for upright faces. In Experiment 3, six participants performed the same task but with unfamiliar faces. The overall accuracy was better for congruent than incongruent conditions, and it was not dependent on template’s SF. According to the coarse-to-fine model, LSF content provides a template to integrate HSF information. However, our results challenge this view, suggesting instead a flexible encoding of SF information, depending on task and image contingencies.
Figures
Experiment-1:Unfamiliar-faces:gender-discrimination-task
Experiment-2:Familiar-faces:match-to-sample-task
Experiment-3:Familiar-vs-Unfamiliar-faces:match-to-sample task
| If you're submitting a poster, would you be interested in giving a blitz talk? | No |
|---|---|
| If you're submitting a symposium, or a talk that is part of a symposium, is this a junior symposium? | No |
