=====Harris & Aguirre 2010: Behavioral data=====
This page discusses additional behavioral results associated with this paper:

  * A Harris, GK Aguirre. (2010) [[http://www.ncbi.nlm.nih.gov/pubmed/20505126|Neural Tuning for Face Wholes and Parts in Human Fusiform Gyrus Revealed by fMRI Adaptation]]. //Journal of Neurophysiology//. 104(1):336-45.

====Introduction====
In our 2010 paper entitled //Neural Tuning for Face Wholes and Parts in Human Fusiform Gyrus Revealed by fMRI Adaptation// we examined the habituation of neural responses within the FFA to part changes. In Experiment 3, a "kite" arrangement of stimuli within the face-part space was used:

|{{:public:papers:kite_stim_space.png?200|}}|

We found that responses within the Right FFA suggested neural tuning for face parts for this configuration of face stimuli.

We hypothesized that the reason for our imaging findings was that the asymmetric distribution of part-changes in this space led to part-based perception of the stimuli. This would suggest that markers of holistic perception, such as the face-inversion effect, would be reduced for stimuli from this space.

====Materials and Methods====
//Subjects//: Thirteen subjects between the ages of 18 and 35 with normal or contact-corrected vision were recruited from local universities for the behavioral experiment.

//Stimuli//: Whole face stimuli from the original (Experiment 1) and asymmetric (Experiment 3) stimulus spaces were used in the behavioral experiment.

//Procedure//: Subjects completed 2 sessions (1 original, 1 asymmetric) of 4 blocks each (alternating upright and inverted, 2 blocks each), separated by a 5-minute break. The order of the original and asymmetric sessions was counterbalanced across subjects. Each trial consisted of two stimuli presented simultaneously side-by-side until the subject responded, with a 250 ms inter-trial interval. The two stimuli could be identical or vary either along one or both axes of the stimulus space. Half of the trials were “same” trials, and the same and different trials were randomly interleaved in each block. Subjects performed a same/different judgment with visual feedback. There were 180 trials per condition for a total of 2880 trials. Because the face pairs that differ by a single feature (Pure stimulus transition) are the same between the two stimulus spaces, these stimuli are most informative regarding differences between the two stimulus spaces in terms of processing. Therefore, analysis focused on the inversion effect for the Pure stimulus transition in the two stimulus spaces.
====Results====
In Experiment 3, we found that the use of an asymmetric stimulus space produced a part-based pattern of neural tuning in the right FFA. Here we tested whether the asymmetric stimulus space is also associated with a behavioral marker of reduced holistic processing. We examined the face inversion effect (Yin, 1969), a differential reduction in recognition performance for faces versus other objects when the image is turned upside-down.

Impaired performance for the inverted face is often attributed to a disruption of holistic processing (Rossion, 2008), possibly in conjunction with a switch to part-based processing (Bartlett and Searcy, 1993; Searcy and Bartlett, 1996). Therefore, if our fMRI data for the asymmetric stimulus space truly reflect part-based neural tuning within the right FFA, we would similarly expect a reduced behavioral inversion effect for these stimuli.

We examined the face inversion effect for the original and asymmetric stimulus spaces in 13 subjects (3 of whom also participated in fMRI scanning). This comparison focuses on the Pure stimulus transitions—that is, stimuli that differ only in eye or mouth identity. Because the physical distances between these stimuli are identical across stimulus spaces, differences in the degree of inversion effect between the stimulus spaces can therefore be attributed to their differential geometry.

Behavioral inversion effects for the Pure stimulus transitions are displayed in the figure. Consistent with our fMRI findings from Experiment 3, the same subjects show a smaller inversion effect for the asymmetric stimulus set compared to the original, symmetric stimulus space. This was confirmed by a repeated-measures ANOVA with stimulus space (original/asymmetric) and orientation (upright/inverted) as factors, which found a significant 2-way interaction effect (F(1,12) = 5.04, p = 0.04).

|{{:public:papers:harris_aguirre_kite_behavior.png?200|}}|

====Discussion====
These behavioral data suggest that the asymmetric stimulus space used in Experiment 3 is processed in a less holistic manner than the stimulus set from Experiment 1. Within a single group of subjects, there is a significant reduction in the face inversion effect, a standard index of holistic processing, for the asymmetric ("kite") stimulus set relative to the original faces.