Academic journal article Perception and Psychophysics

The Repetition Discrimination Task: An Objective Method for Studying Perceptual Grouping

Academic journal article Perception and Psychophysics

The Repetition Discrimination Task: An Objective Method for Studying Perceptual Grouping

Article excerpt

Five experiments are reported that demonstrate the use of the repetition discrimination task (RDT) to study perceptual grouping effects objectively and quantitatively. Experiments 1 and 3 validate the method by measuring grouping based on proximity, color similarity, common region, and element connectedness. Experiment 2 compares the RDT effects for proximity grouping to explicit subjective ratings of grouping strength in identical displays. Experiments 4 and 5 investigate the effects of size and orientation of surrounding ovals in displays in which competing organizations are present. In each case, the RDT produces clear, consistent patterns of response times that are consistent with predictions based on grouping. It thus represents an objective method for studying the full range of grouping phenomena originally described by Wertheimer.

In 1923, Max Wertheimer called attention to one of the fundamental problems of visual perception: How does the visual system organize the complex array of light striking the retinal mosaic into the coherent spatial array of objects that we experience when we view the world? Although Wertheimer did not by any means solve this complex and difficult problem, he did make substantial inroads by identifying a number of stimulus-driven principles of grouping. These include the well-known factors of proximity, similarity, good continuation, and common fate that are discussed in virtually all textbooks on vision (see, e.g., Palmer, 1999). Other principles of grouping have since been added to this list, including common region (Palmer, 1992), element connectedness (Palmer & Rock, 1994a, 1994b), and synchrony (Blake & Yang, 1997), but Wertheimer's original article is still the definitive work.

The method Wertheimer (1923) used to investigate perceptual grouping was phenomenological demonstration. He published displays that illustrated the principles he proposed and left it up to readers to decide whether or not their phenomenology on viewing the displays agreed with the descriptions he gave in the accompanying text. Obviously, most people agreed, for Wertheimer's principles of grouping are among the oldest, best-known, and most widely cited findings in perceptual psychology. The implicit agreement among his readers was sufficient for his purpose, which was simply to demonstrate the effectiveness of isolated stimulus factors in organizing individuated elements into groups.

There are two aspects of Wertheimer's method that limit its usefulness for more sophisticated scientific purposes, however. One limitation is that it is qualitative rather than quantitative. That is, it generates no numbers that can be used to measure the strength of a single grouping factor or of the combined effect of multiple factors. This limitation is important because Wertheimer's principles are essentially ceteris paribus rules-rules that hold with certainty only when all else is equal in the sense that no other grouping factor differentially influences the outcome-and ceteris paribus rules provide no way to predict the combined influence of multiple conflicting factors. Quantitative theories of multifactor integration are needed to overcome this problem, and such theories can be tested only with methods that allow grouping strengths to be quantitatively measured in both single-factor and multifactor designs.

A different limitation is that Wertheimer's demonstration method is phenomenally based, or subjective, rather than physically based, or objective. The distinction we intend to make between objective and subjective methods is simply that the former requires that there be observer-independent (i.e., physically defined) correct responses to the stated task, whereas the latter requires only that there be observer-dependent (i.e., phenomenally defined) correct responses.1 At least for certain purposes, such as determining the accuracy of observers' performance and any response biases that may exist, it is preferable to have a method that measures performance against a physically well-defined (objective) criterion rather than a purely phenomenal (subjective) one. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.