Abstract
When attempting to understand where people look during scene perception, researchers typically focus on the relative contributions of low- and high-level cues. Computational models of the contribution of low-level features to fixation selection, with modifications to incorporate top-down sources of information have been abundant in recent research. However, we are still some way from a model that can explain many of the complexities of eye movement behaviour. Here we show that understanding biases in how we move the eyes can provide powerful new insights into the decision about where to look in complex scenes. A model based solely on these biases and therefore blind to current visual information outperformed popular salience-based approaches. Our data show that incorporating an understanding of oculomotor behavioural biases into models of eye guidance is likely to significantly improve our understanding of where we choose to fixate in natural scenes.
Acknowledgements
We thank Jim Brockmole, Aude Oliva, and Simon Liversedge for their comments on the submitted version of this manuscript. We also thank Roland Baddeley, Mark Bennett, Tim Dixon, Martin Fischer, Gesche Huebner, Casimir Ludwig, Wayne Murray, and Nick Wade for their comments and guidance during the preparation of this work.
Notes
1Free viewing is often taken as a task that is free from higher level “baggage”. However, this is not the case and free viewing comes with its own set of issues (Tatler et al., Citation2005). We chose this “task” for comparability with other recent studies that have evaluated low-level factors in eye guidance.
2It should be noted that some studies using different methods have reported higher ROC AUC values for Itti's salience map (e.g., Gao, Mahadevan, & Vasconcelos, Citation2008). However, our values fall in the range of previous studies and where in this range we fall does not undermine any comparisons of the relative predictive power of salience and motor biases in our dataset.