Supplement to: Objects do not predict fixations better than early saliency; A re-analysis of Einhäuser et al.’s data

We use three prevalent scoring methods to test the object-map hypothesis, since validity of this model boils down to fair model comparison with bottom-up saliency models. Please see Appendix B for explanation of state-of-the-art saliency models used here. We report results using the previously proposed Normalized Scanpath Saliency (NSS) (Peters et al., 2005; Parkhurst et al., 2002), Correlation Coefficient (CC), and Shuffled AUC score (AUC Type-3) (Parkhurst & Niebur, 2003; Tatler, 2007). Our emphasis is more on the Shuffled AUC score as it is the only score that tackles center-bias at the data level. Please see Appendix C for detailed explanation of scoring methods.