Taste intensity and the case of exponents greater than 1

Abstract

The magnitude estimation of taste intensity has often produced exponents greater than 1, suggesting that the perceived intensity of some taste stimuli increases at a faster rate than does stimulus concentration, i.e., the psychophysical functions are positively accelerating. However, a reanalysis of published data reveals that when these magnitude scales are replotted in linear coordinates, instead of the conventional log-log form, there is often no evidence of positive acceleration and the numerical value of the exponent bears no reliable relationship to the shape of its psychophysical function. Reasons for this finding are discussed and the implications for taste research are noted.The power function S = kZ n, where S is subjective magnitude, I stimulus intensity, n the exponent, and k a constant, has been widely used in psychophysics.An attractive feature of this formulation is the capacity of the exponent, a single numerical value, to reflect the shape of a psychophysical function. Thus, obtaining an exponent less than 1 implies that the psychophysical function would be negatively accelerating (compressed, concave downward) if plotted in linear coordinates; an exponent equal to 1 implies a straight line (linear) relationship; and an exponent greater than 1 implies a positively accelerating (expanded, concave upward) psychophysical function, i.e., implies that the subjective magnitude grows at a faster rate than the stimulus intensity (Bartoshuk, 1977, Figure 1; Stevens, 1961Stevens, , 1975.When taste stimuli are evaluated with the conventional sip procedure and the response technique of magnitude estimation, and when power functions are fitted to the data, the resulting exponents have often been found to exceed 1 . This is especially true of sucrose (Meiselman, 1971, Table 1; Moskowitz, 1970;Stevens, 1969) and many other sugars (Moskowitz, 1971a). The exponent for the salt stimulus sodium chloride has also been reported to be greater than 1 (Meiselman, 1971, Table 1; Stevens, 1969). Values for bitter stimuli (e.g., quinine sulfate and quinine hydrochloride) are around 1 or lower (Meiselman, 1971), and exponents for acids are usually below 1 (Moskowitz, 1971b).The results of magnitude estimation scaling are usually presented in the form of log-log plots, with the slope reflecting the value of the exponent. It follows, though, that if the psychophysical functions of those taste stimuli with exponents greater than 1 are replotted in linear (arithmetic) coordinates, then the functions should be positively accelerating. A preliminary analysis (McBride, Note 1) has revealed that this is not always the case. When magnitude scales of sucrose sweetness with exponents of around 1.3 are replotted in linear coordinates, the Requests for reprints should be sent to R.

Sweetness of binary mixtures of sucrose, fructose, and glucose.

“…Definitions aside, however, Figure 7 shows how supplemental action can be accounted for in terms of the addition (or near addition) of negatively accelerated psychophysical functions—that is, in terms of compression rather than expansion. If the psychophysical functions of taste stimuli generally tend toward negative acceleration, as has been suggested elsewhere (McBride, 1982b, 1983c), and if the present pattern also applies to other binary mixtures, supplemental action will commonly occur. In support of this reasoning, Kamen (1959) noted supplemental action with mixtures of sucrose and calcium cyclamate, and Homler (1984) with mixtures of aspartame and various other sweeteners (see also Yamaguchi, Yoshikawa, Ikeda, & Ninomiya, 1970).…”

“…Integration psychophysics, in contrast, does respect this possibility. One other major advantage of the present paradigm is that there is no fitting of theoretical curves (e.g., power functions) to the mixture and its components, a practice that can sometimes be problematical in itself (McBride, 1983c, McBride & Anderson, 1986). Instead, information on mixture perception is available from inspection of the factorial plot.…”

Evaluation of Bitterness of Caffeine in Hot Chocolate Drink by Category, Graphic, and Ratio Scaling

“…Subjects, who were not permitted to use zeros even for the sample with 0 g/L of added caffeine, might have had to use very small fractions thereby distorting the psychophysical function near the origin and yielding a poor fit of the ME data by a power function. Several other investigators have reported magnitude estimates which gave poor fits to a power function (O'Mahony and Heintz 1981; Giovanni and Pangborn 1981;Powers et al 1981;McBride 1983). Although opinions vary on the influence of judge training on ability to generate whole-number proportions, we concur with Teghtsoonian and Teghtsoonian (1971), Beck and Shaw (1975) and Robinson (1976) that many judges require practice in use of ratios to scale sensation in order to generate numbers which represent perceptual ratios.…”

Cautionary note on log-log plots

“…This note is directed at a problem in data analysis, and is not intended to address the issue of the correct exponent for electric shock. It simply demonstrates that the shape of a psychophysical function is not necessarily reflected by its exponent, and stands alongside other related questioning of power functions in psychophysics (Carterette & Anderson, 1979;Coleman, Graf, & AIf, 1981;McBride, 1983;Myers, 1982).…”

“…

The results of magnitude estimation studies are customarily presented in the form of log-log plots. However, a reanalysis of taste data (McBride, 1983) has revealed that such plots can be seriously misleading. This note reports a further example in a dimerent modality.Stevens.

…”