Poster #3 - Exploring the Role of High and Low Spatial Frequencies in the Development of Infant Selective Attention to Faces

• In Event: Poster Session 15

Abstract

Humans process faces more holistically than other stimulus types (Tanaka & Farah, 1993). The Encoding Switch Hypothesis proposes that before holistic face processing develops, infants and children up to about 10 years of age engage in a more piecemeal or parts-based perception of faces (Diamond & Carrey, 1977). However, recent research suggests holistic processing is present earlier in development than previously thought and could potentially precede parts-based processing (Nakabayashi & Liu, 2013). The current study aims to test this alternative hypothesis. Adult behavioral data suggests that low spatial frequencies (LSF) facilitate holistic face processing whereas featural face processing relies on high spatial frequencies (HSF) (Goffaux & Roisson, 2006; Goffaux et al., 2005). Infant contrast sensitivity at LSF develops rapidly and peaks at 10 weeks of age. In contrast, sensitivity at HSF increases dramatically from 9-32 weeks of age allowing infants to perceive increasingly finer detail. This may indicate that infants initially engage in holistic processing due to the limited HSF visual information available to them. Newborns have been found to rely on LSF for face recognition (de Heering et al., 2008); however, there is a lack of research exploring the role of SF content in face perception as a function of age. The current study examines 4- and 9-month-old infants’ perception and recognition of LSF and HSF filtered face stimuli. A 2x2x3 design is being used with a between-subjects factor of age (4 months, 9 months) and within-subjects factors of identity (Familiar, Novel) and SF condition (LSF, HSF, Hybrid). Filtered stimuli are created using a Fourier transformation and Gaussian bandpass filter in MATLAB to create LSF and HSF images containing only spatial frequencies below 2 cycles/° visual angle and above 6 cycles/° visual angle respectively. Hybrid image stimuli are created by combining the LSF content from one (novel or familiar) identity with HSF content from the other (Olivia & Schyns, 1999). Each trial begins with a familiarization phase consisting of unfiltered faced images of the same identity (Rose, Feldman, & Jankowski, 2002). Infants are then presented with paired comparison trials for novel vs. familiar faces in LSF, HSF, and inverse hybrid conditions in a preferential looking task (Fagan,1970). Eye tracking data is collected at 500 Hz using an Eye-Link 1000 Plus (SR Research Ltd., Mississauga, Ontario, Canada). Preferential looking data will be analyzed using a mixed effects model with fixed effects of age and condition as well as a random effect of participant. Fixation data will be analyzed using the IMap technique (Caldara & Milett, 2011). We predict that 4-month-olds will prefer novel faces in LSF but show no preference in HSF conditions, suggesting they are unable to use HSF content in face processing. We predict that 9-month-olds will prefer novel faces in both LSF and HSF conditions. Additionally, we predict that 4-month-olds will prefer hybrid images with novel LSF content but 9-month-olds will prefer novel HSF content images or show no preference. These findings would support the alternative hypothesis that holistic face processing develops prior to parts-based face processing.

Authors

• Kenna R. H. Clayton, University of Tennessee - Knoxville
  Presenting Author

• Lauren Slivka, University of Tennessee - Knoxville

• Ashley Nicole Barrows, University of Tennessee - Knoxville

• Greg D. Reynolds, University of Tennessee - Knoxville