2.3.1. Introduction to Psychophysics
Gustav Theodor Fechner was the first to specify psychophysics as “an exact science of the functional relations of dependency between body and mind”
(Torgerson, 1958 p.v). Psychophysical methods are used in order to put subjective attributes into units of measurement (Fairchild, 2005 p.36). For the investigation of colour appearance, visual experiments are necessary so as to examine human perception.
Humans tend to judge an aroused stimulus of a reproduction copy by comparing it with a memory stimulus of an original (Hunt, 2004 p.32). In Figure 2.3-1, Hunt explains diagrammatically this procedure of comparison.
However, there are variations in the memory stimuli due to different viewing conditions at each observation situation. Such variations can affect hue, lightness and colourfulness.
Figure 2.3-1 Visual appreciation mechanisms of human perception (Hunt, 2004 p.33)
Visual adaptation is called the function by which “the visual system adjusts its operating level to the prevailing light level” (Norton and Corliss, 2002 p.77). The visual system adapts to the light conditions because of the eyes’
characteristics: (1) the photoreceptors’ operation, i.e. rods and cones function; (2) the ability to modify the concentration of photopigments; and (3) the ability to alter its sensitivity, i.e. neural responsiveness (Norton and Corliss, 2002 pp.76-77). In particular, dark adaptation is the “decrease in threshold luminance [increased sensitivity] as a function of time in darkness”
(Norton and Corliss, 2002 pp.77-78). Experiments have shown that after changing from an adapted lighting condition to a dark one, there is a fast reduction of threshold luminance in the cones which tends to stabilise within the first 3 minutes. Then, until the 11th - 12th minute, there is little reduction and afterwards “the cones approach their lowest threshold level”. This is also illustrated in Figure 2.3-2.
Figure 2.3-2 Typical dark adaptation curve (Norton and Corliss, 2002 p.78)
The visual system has the ability to detect and identify objects from patterns of light and dark (Norton et al., 2002 pp.138-141). This indicates that in terms of spatial vision, humans can easily spot differences in luminance.
Relative luminance describes these differences by comparing the reflected light from one area against another area. Because this ratio stays constant under every luminance level, it is easily recognisable by the visual system.
So the most important factor in this process is the relative luminance. A phenomenon characterised by this principle is simultaneous contrast.
2.3.2. Psychophysical Methods
As previously mentioned, psychophysical methods have been developed in order to assign numerical values to visual perception. There are several psychophysical methods designed for visual experiments relating to the investigation of colour appearance and image quality. Terms for the most common psychophysical methods are given in Appendix A.3.
Psychophysical experiments can be divided into four basic categories: (1) thresholds and matching; (2) measuring differences; (3) direct ratio scaling;
and (4) multidimensional scaling (Bartleson and Grum, 1984; ASTM, 2003b).
Each category consists of methods for investigating different types of attributes. Examples of psychophysical methods are: paired comparison, rank ordering, categorical sort, magnitude estimation, etc. Some methods can even be combined so as to produce more specific results. “The choice of the best method for a particular application may be difficult to make, and interpretation of the rating scales produced by the numerical analyses is frequently ambiguous” (ISO, 2005).
There are also two viewing techniques; having the observer’s eyes under the same viewing conditions or under different ones. For instance, binocular or else haploscopic memory matching where each eye is adapted to different test areas. “The two eyes are situated in slightly different positions, thus receiving slightly different images of the same external objects” (Baird, 1970 p. 209). Each eye is adapted under different viewing conditions and the objective is to determine threshold of stimuli (Bartleson and Grum, 1984).
2.3.2.1. Ratio Scaling
In general, there are four types of one-dimensional scales: nominal, ordinal, interval and ratio (Engeldrum, 2000; Stevens, 1946). In nominal scales, the numbers are used as a distinguishing mean rather than as a mathematical property; thus they express equality among the numbers. Ordinal scales use descriptions to define the scaling. Thus, observers monotonically evaluate
an attribute in terms of greater than or less than the specified attribute.
Interval scales are like ordinal scales with the difference that equal distances within anywhere in the scale have the same significance. Finally, ratio scales are similar to interval scales but with a zero point or origin. The Table 2.3-1 summarises the transformations that can be applied in the described scales.
Table 2.3-1 Stevens’ classification of scale types and possible transformations
Permissible Transformations
Nominal 𝑦𝑦 = 𝑓𝑓(𝑥𝑥), any one-to-one transformation Ordinal 𝑦𝑦 = 𝑔𝑔(𝑥𝑥), any monotonic transformation Interval 𝑦𝑦 = 𝑎𝑎 𝑥𝑥 + 𝑏𝑏, any linear transformation Ratio 𝑦𝑦 = 𝑎𝑎 𝑥𝑥, any constant scale factor
“Psychometric scaling is the generation of rulers” used to measure the human response (Engeldrum, 2000 p. 43). In colour psychophysical experiments, there are various techniques that produce ratio scales; i.e. ratio estimation, ratio production, magnitude estimation and magnitude production (Wyszecki, 1982). In ratio estimation, the observer evaluates a test stimulus by referring to a standard stimulus. While in ratio production method, the observer has to adjust the test stimulus so as to produce a predefined ratio between a given standard stimulus and the test stimulus. Similarly, in magnitude estimation method, the observer evaluates the perceived magnitude of a test stimulus for perceived attributes such as colourfulness.
Whereas, in magnitude production, the observer produces a magnitude in order to match the magnitude of a given attribute.
In this study, ratio estimation is used. The observer is given two pairs of stimuli. The one pair is a standard colour pair with predefined colour difference of one unit and the other one is a testing pair whose colour difference is to be evaluated in terms of a ratio against the standard pair (Elamin, 1983). This has been used by other studies for colour discrimination research (Cheung and Rigg, 1986; Elamin, 1983).