analysis must be followed. This detailed information is found in sources specifically devoted to multidimensional scaling (e.g., Shiffman, Reynolds, & Young, 1981).
For nearly 150 years, the goal of measuring sensation magnitude has been fundamental to psychophysics, yet many problems remain today. For example, different scaling techniques often produce different results. Either all or some of the methods produce invalid results, or the different methods are validly measuring different aspects of perception. It has been argued that the solution to this problem depends on the development of fundamental psychophysics defined by Ward ( 1992) as an attempt to find "a core of concepts and relations from which all the rest of psychophysics can be derived" (p. 190). We have a tremendous amount of experimental data accumulated since the birth of psychophysics, and we have many concepts such as spatial and temporal summation, adaptation, masking, contrast, stimulus variability, sensory-system variability, to name but a few, to explain them. The basic question is whether there are a few fundamental concepts that, when properly interrelated, will explain all of psychophysics. In recent years, several proposals have been made for such a fundamental psychophysical theory. For example, for Norwich ( 1993), the fundamental principles of psychophysics came from information theory in which the critical event is the flow of information rather than stimulus energy as in more traditional psychophysical theories. From a few basic principles of information flow, Norwich was able to explain sensory adaptation, Fechner's law, and Stevens' law. A somewhat different approach can be seen in Link ( 1992), Wave Theory of Difference and Similarity. According to this theory, on each trial, the observer samples the stimulus continuously over time and the result is the envelope of a time-amplitude waveform. At any point during the sampling process the value sampled from the stimulus wave is subtracted from a value sampled from a referent wave, and the result is a comparative wave. The differences between the sampled wave and the referent wave cumulate over time until the sum exceeds the observer's threshold. The response to the stimulus starts as a quantized action of sensory receptor modeled by a Poisson process, which allows for the prediction of sensation magnitude functions. This theory, in contrast to that of Norwich, focuses on stimulus energy rather than stimulus information. Finally, the nonlinear nature for the process of judging sensory stimuli as they are first converted to sensory