ISSN 0439-755X
CN 11-1911/B

›› 2011, Vol. 43 ›› Issue (09): 983-992.

Previous Articles     Next Articles

The Mental Representation on Three-Dimensional Object Recognition: Dependent or Independent on the View

LIAN Ling;YOU Xu-Qun   

  1. (1 School of Psychology, Shaanxi Normal University, Xi’an 710062, China)
    (2 Ideological and Political Department, Xi’an Polytechnic University, Xi’an 710048, China)
  • Received:2010-10-25 Revised:1900-01-01 Published:2011-09-30 Online:2011-09-30
  • Contact: YOU Xu-Qun

Abstract: Object recognition refers to a decision about whether an object has been seen before. Recognition involves generalization across shape, size, spatial relation, location, view and illumination. The view has been the focus of the debate in many recent object recognition studies. The object-centered theory assumed that object recognition is a qualitative representation of an object in terms of its three dimensional primitives (for example, geometrical ion) and their spatial relationships, independent on the view. By contrast, the view-centered theory posited that object recognition is dependent on the view. It is view-specific, or represented in an egocentric frame of reference. It might contain information about illumination, colour, material and depends systematically on experience with specific views of an object.
We decided to investigate the roles of structural information and view information in object recognition, with specific emphasis on the potential separability of the shape and the rotational ways of the view in object perception, as predicted by the view-centered theory.
In the current study, made 3D objects as experimental materials according to the idea of Biederman’s small geons; used the priming paradigm; adopted identify task; participants were required to judge whether two objects are the same or not. The effect of the 3D object recognition on the PC (part-changed) or RC (relation-changed) distractors is studied in Experiment 1. A 6 (view: -30°, 0°, 30°, 60°, 90°, 180°) × 4 (compound mode: DoDv, DoSv, SoDv, SoSv) × 2 (task: indirect, direct) mixed design is utilized in Experiment 2. The rotations were clockwise by 90°and 180°in picture-plane level, and the rotations were counter clockwise about the vertical axis by -30°, 30°and 60° in depth. In the indirect task, those participants were told to respond ‘same’ if two pictures appeared at any view. In the direct task, those participants were told to respond ‘the same’ if two pictures appeared only at the identical view.
The results confirmed our hypothesis. In Experiment 1, the speed of object recognition in repeat prime was faster than those in the part prime and the spatial relation prime, there was a significant difference between the RTs of the part and spatial relation prime. In Experiment 2, ANOVA revealed the main effect of the RTs of the rotational view, compound mode and task type, as well as a significant view-by-mode, mode-by-task interaction. Especially, the RTs in the SoSv mode are shortest. That is, the view prime effect was found in two tasks. The size of priming effect is affected by the rotational ways of the object.
Based on these results, we concluded that (1) the speed of 3D object recognition was influenced by the shape of part and the categorical spatial relation, which suggested that the object-centered theory was rationality. (2) The view information was independent on the shape information, and the depth rotation was better than the picture-plane rotation, which suggested that the depth rotation phenomenon can be better explained by the view-centered theory. The findings not only provided experimental evidence for object recognition, but also shed light on the object theory.

Key words: view, three-dimensional object, the view-centered theory