Visual search is one of the main types of perception. Visual processes are fundamental to objects recognition and perception. Feature integration theory is investigated by a large number of researchers including Waden and Swanston (2001), Duchowski (2003), and Gordon (2004). The main theories of feature integration involve Gestalt theory and neuropsychological approach, Man’s computational approach, and ecological optics. Researchers consider the process of feature integration as a mapping from the measurement space to the feature space (Waden and Swanston 2004).
One approach is to assume that the features used in recognition are abstract, general features that apply to varying instances of the same object. That is, prototype descriptions for objects would be in terms of general features. The well-known feature list includes general binary contrasts such as the presence or absence of “straight-vertical” and “intersection.” However, there has not been much research in the last 20 years that directly tests these features, perhaps because of the very general level at which such features are (necessarily) specified.
One exception is research examining contrasts, in which features are present or absent to varying. Gordon admits that “typically, what they try to explain are basic sensory discriminations: how perceivers process some of the basic information contained in, say, the visual image, how this is coded and in what form it is sent onwards into the higher regions of the visual pathways” (Gordon 2004, p. 74). Recently, the importance of binary contrasts has taken on added meaning in models of object recognition Objects are represented in terms of their parts and are identified by detecting a limited number of easy-to-detect binary contrasts.
They are unlikely to occur by accident, and they remain invariant across changes in viewpoint. These characteristics are important because object recognition must distinguish between image properties that are valid indicators of object properties and image properties that may occur because of an accidental alignment of stimulus features. A second response to differences between instances of the category is to simply focus on a single font and to assume that features correspond to local parts or global properties within that font (Duchowski, 2003).
The hypothesis is that the perceptual processing of objects is greatly facilitated by nonvisual information. The nonvisual information supplements the Feature integration information and improves performance relative to the situation with just visual information. One issue relevant to the two approaches already mentioned is the degree to which the processes underlying recognition become “tuned” to the details that characterize particular fonts.
If features are only abstract and general, then varying the font should not matter. As a result, letter recognition should be more accurate with target sets of n letters of a single font than with n letters of two or more fonts because tuning can be more precise with a single font. However, the rather small effects in the masking task qualified the extent to which such features contribute to recognition; more general features may be more important.
References
- Gordon, J. E. (2004). Theories of Visual Perception. Psychology Press.
- Duchowski, A. T. (2003). Eye Tracking Methodology: Theory and Practice. Springer.
- Waden, N. J., Swanston, M. (2001). Visual Perception: An Introduction. Psychology Press.
- Zwaga, H. J. G. et al (1998). Visual Information for Everyday Use: Design and Research Perspectives. Taylor & Francis.