英文摘要 |
A fundamental question for models of object recognition is why human recognition can be so efficient (Biederman, 1987). To clarify this question, research endeavors in the past have focused on the specific format of object representation, while overlooking the differential contributions that different levels (in terms of spatial scales) of object representation make. In the current study we looked into the contribution of global representation on the recognition of an object, i.e., global diagnosticity. It is suggested that global representation is prioritized by the recognition mechanism for a number of reasons and that objects high in global diagnosticity are easier to recognize than objects low in global diagnosticity. In addition, the effect of complexity on recognition was examined. Results of three experiments indicate that both global diagnosticity and complexity affected the naming performance of briefly-exposed and backward-masked objects in systematic ways. Globally diagnostic (GD) objects were recognized faster and more accurately than globally non-diagnostic (GN) ones (Experiment 1-3). Recognition of GN objects was found to be more susceptible than GD objects due to manipulation by the duration of exposure (Experiment 2), as well as location uncertainty and presentation location (Experiment 3). U-shaped complexity effects were further found in the RT data of GD objects and the error rate data of GN objects (Experiment 1-3). Implications for models of object recognition and researches adopting object stimuli were also discussed. |