Sweetness and intensity of artificial sweeteners

HOWARD R. MOSKOWITZ2 BAR VARD UNIVERSITY

In four experiments, groups of Os saccharin) and from commercial brands magnitude estimate of each O. Thus each 0 judged either the or the entire (Domino cane ; "Sucaryl," "Ann Page was given a different positive number intensity of solutions of , Sweetener"). Solutions prepared with corresponding to his estimate of "0," with cyclamate , cyclamate-saccharin Cambridge tap water were made up 3 to 4 the property that it was the lowest of his mixtures, and saccharin. The days prior to use. When not in use they numbers. sensory functions obtained by magnitude were stored under refrigeration at 5°C. The geometric mean has been employed estimation suggest that over the middle Different groups of Os, chosen from a as the measure of central tendency for range of concentration the sweetness and pool of 60 Harvard students, participated magnitude estimates because the intensity of the foregoing substances grow in the experiment. Each 0 sat in front of a logarithms of the judgments tend to be as power functions of concentration. As a table on which were arranged the stimulus distributed symmetrically. In the present first approximation, the exponents for solutions in a linear array. Solutions were case the presence of "0" required that the sweetness and intensity are, respectively, given in paper cups (%-oz Sweetheart median judgments be used instead, because 1.6 to 1.4 for sucrose, 1.0 to 0.8 for souffle cups), each containing 4 to 6 ml of the geometric mean of a set of numbers cyclamate salts, 0.6 to 0.85 for solution. The order of presentation was containing a zero becomes zero. cyclamate-saccharin mixtures, and 0.3 to irregular and was changed for each O. In In all the experiments reported here 0.6 for sodium saccharin. most of the experiments the 0 sampled each 0 was left free to choose the modulus between 40 and 50 solutions, usually for his scale. As a result some individuals grow in sweetness according to a taking about 30 min to complete the used numbers in the hundreds and power function of concentration, with a session. thousands, whereas others used fractions. probable exponent of 1.3 (Moskowitz, Prior to each session the 0 was given the In order to reduce the magnitude estimates 1968, 1970; Stevens, 1969). In log-log following written instructions: "In front of of all Os to a common modulus, without, coordinates this relation between sweetness you is a series of solutions in paper cups. however, changing the ratios of the (S) and concentration (C) is expressed by Your task is to tell how sweet/intense they numbers, a normalization procedure called the equation log S = 1.3 log C + log k. The seem by assigning numbers to them. If the modulus equalization was used (Lane, intercept, k, differs for different sugars and second is nineteen times as sweet/intense as Catania, & Stevens, 1961). The is a measure of relative sweetness at the the first, then assign it a number nineteen normalization required two steps. First, the concentration given by the intercept times as large. Ifthe second is one-eleventh responses of each 0 to a core set of six (Moskowitz, 1970). There are, however, as sweet/intense, then assign it a number sucrose stimuli (2%, 4%, 8%, 16%, 32%, many other sweeteners besides sugars. The one-eleventh as large, and so forth. Use 64%) were normalized by a single present study concerns the sweetness fractions, decimals, and whole numbers, multiplier so that their geometric mean functions for the artificial sweeteners, but make each assignment proportional to became to.O. The ratios of magnitude cyclamate and saccharin, whose quality of the sweetness/intensity as you perceive it. estimates were left unchanged, but the size sweetness differs markedly from the Rinse after each stimulus solution. If you of the numbers was equalized for all Os. sweetness produced by sugars. Usually the cannot taste anything in the solution then Second, the responses of each 0 to the sweet taste of cyclamate and saccharin is rinse again and try the sample once more. other stimulus concentrations in the same accompanied by a bitter off-taste (Helgren I[ you still cannot taste anything, then use experiment were also multiplied by the et al , 1955) that increases with "0" for your magnitude estimate." same constant that had been used concentration. At the higher There was no time limit on the session. previously to normalize his response to concentrations the bitter predominates and Each 0 was free to proceed at his own sucrose. Again the ratios of magnitude the artificial sweeteners no longer evoke pace. During the course of the sessions estimates were left unchanged. The only the sweet taste. several Os complained about a bitter taste change was in the absolute level of the In view of the presence of at least two that remained after rinsing. In these cases magnitude estimates, and, for each 0, that distinct in artificial sweeteners, it is they were instructed to rinse thoroughly, change was identical for all his numerical important to separate the sweetness gargle, and wait a few seconds before estimates. In each experiment every 0 had component from the rest of the taste. In proceeding. a unique normalization multiplier, and the present study, functions both for when Os participated in two or more sweetness and for total taste intensity were ANALYSIS experiments they usually required two or obtained for each of four sweetening The numerical matches (magnitude more different multipliers. By this substances: sucrose, sodium and estimates) were analyzed with a local normalization two sweeteners that had cyclamate, sodium and calcium Suearyl (a computer program called PSYCHOFIT been scaled in entirely different mixture of cyclamate and saccharin), and (panek & Stevens, 1965). PSYCHOFIT experiments, e.g., sodium saccharin and sodium saccharin. provided geometric means, medians, and calcium cyclamate, could be compared to measures of variability (standard deviation each other. PROCEDURE and interquartile range) for the magnitude Stimulus solutions were prepared on a estimates of each stimulus concentration. RESULTS percent-by-weight basis (grams In those instances in which Os reported In Fig. 1 the logarithms of the median solute/grams solution) from reagent-grade "0" as their magnitude estimate the "0" magnitude estimates of "sweetness" and chemicals (cyclamate salts and sodium was replaced by 0.1 X the lowest "intensity" are plotted as a function of the

40 Copyright 1970, Psychonomic Journals, Inc., Austin, Texas Perception & Psychophysics, 1970, Vol. 8 (1) 1000 .. 500 • INTENSITY • I' .OC·6 INTENSITY I' 3.8C·8~ INTENSITY I' Z& C·8 S. EETNE" S·8.5C.3 o / K / 1/ -- • o 1 [ s-r: SWEETNESS • S'9C •.0 d/~_t.~- • o o \ 0 o SWEETNESS o S. 5.4C ,E; --- \ swEETNESS 0 • S·20C·3 o

o •. CALCIUM SUCARYL o O' EXP. I SODIUM SUCARYl.. .6 A• •• EXP. • CALCIUM CYCLAMATE .Z o •. xr• • ANN PAGE SwEETENER 0.,

2.0 4.0 8.0 \6.0 32D 640 ,00625 .025 .I 4 1.6 6.4 .062~ .25 1.0 4.0 16-0 4.0 5 .02~ 0.1 0.4 1.6 PERCENT SUCROSE 8"" WEIGHT PERCErIi't CYCLAMATE SALTS 8Y WEIGHT PERCENT CYCLAMATE- SACCHARIN MIXTURE 8Y WEIGHT PERCENT SODIUM SACCHARIN BY WlIG"l

Fig. 1. Sweetness and intensity functions for four groups of sweet substances: sucrose, sodium and calcium cyclamate, cyclamate­ saccharin ("Sucaryl") mixtures (sodium and calcium forms), and sodium saccharin. The coordinates are log-log in each case, so that straight lines suggest power functions.

logarithm of concentration. When sensory the difference between the functions is between the slopes of c-yclamate and functions are linear in these logarithmic marked, and the sweetness exponent is saccharin. coordinates, they conform to power approximately half of the intensity On the basis of the power-function functions. In most cases the sweetness exponent. exponents the sweeteners may be placed functions are linear only over a limited The three simple sweetness functions into three classes: sucrose, cyclamate salts, range of concentrations. In particular, the (sucrose, cyclamate salts, sodium and saccharin. Sucaryl mixtures appear to high concentrations of sucrose and the saccharin) differ from each other in slope be intermediate between the cyclamate extreme concentrations (at both ends) of (exponent). Sucrose grows most rapidly in salts and saccharin but resemble more the artificial sweeteners systematically sweetness with concentration, followed by closely the cyclamate salts, which depart from a straight line. The straight the cyclamate salts, and finally by sodium predominate in Sucaryl by a factor of lines in Fig. 1 were fitted only to the most saccharin. In the case of cyclamates the 10:1. linear segment of the function. Both the nature of the cation appears to make little No single exponent accounts for either slope and the intercept were calculated by difference in either the sweetness or the the sweetness or the intensity of the a least-squaresprocedure. intensity function. There are two different On the other hand, the functions that sweetness functions for sodium saccharin, resulted when the Os were instructed to both with an exponent of approximately o judge the overall intensity of the taste 0.3, but with intercepts that differ by a conform more nearly to power functions. factor of 2.5. Thus there are two separate The differences between the sweetness and estimates of the relative sweetness of the intensity functions become more sodium saccharin, differing by a factor of pronounced as the exponents decrease in 2.5. On the other hand, the growth in magnitude. In the case of sodium saccharin sweetness for sodium saccharin is approximately the same in each

Table I experiment. The discrepancy may have '0. Conditions of Experiments resulted from different groups of Os, since ~ :: z 7. Exp. No. Os Judgrnents/Os Substance the two functions were obtained in o .. separate experiments. In general, the !il!: .. 10 2 Sweet Sucrose artificial sweeteners grow in sweetness less z .. SLOPE. 1.7 10 2 Intensity Na Cyclamate ..... rapidly than sucrose, typically at a 8 .. DATA fROM MAGIDSON AND Ca Cyclamate QOIItAIlACHOW. 1923 2 10 2 Sweet Sucrose decelerating rate with concentration. At ..z .. Tiu'EL AND KL[..... 192~ 10 2 Intensity Na Sucaryl* high concentrations the sweetness ...i Ca Sucaryl* functions for the artificial sweeteners begin :a AnnPage* = to decrease with increasesin concentration, I.+-.L..-,---,r--r-r-r-r-rrlr--...... ,.­ 3 15 1 Sweet Sucrose 1.0 1.0 loG 4.0.0 7.0 10.0 15 1 Intensity Na Saccharin presumably because a bitter side-taste "0 &0 4 30 1 Sweet Sucrose becomes predominant. '''0 Na Saccharin The functions for intensity are linear PERCENT suelltoSI£ BY WEIGHT ..Na Sucaryl is a commercial brand ofsweetner, (i.e., power functions) throughout most of containing a 10:1 ratio ofsodium cyclamate to the range, except for the two highest Fig. 2. Equal_eetness functions traced sodium saccharin. Ca Sucaryl is a similar com­ concentrations of sucrose and the lowest out by direct matches between sucrose and mercial brand containing a 10:1 ratio ofcalcium saccharin sweetness (Magidson & cyclamate to calcium saccharin. Both are sold concentration of saccharin. The slopes by Abbott Laboratories, Inc., North Chicago, (exponents) for the intensities of Gorabachow, 1923; Tilufel & Klemm, Ill. Ann Page is a 10:1 mixture of sodium cyclamate salts and Sucaryl mixtures are 1925). The coordinates are log-log. The cyclamate and sodium saccharin marketed by virtually identical, whereas the slopes for slope (exponent) of the power function is the A & P Company. the sweetness of Sucaryl mixtures lie approximately 1.7.

Perception & Psychophysics, 1970, Vol. 8 (1) 41 Fig. 3. Variability of magnitude estimates (interquartile range) expressed as 12 decilogs (10ths of a logarithmic unit). In 10 the region where the function is flat the • ~ variability is proportional to the mean • ~ 0 ~ ~ judgment. • 0 .. I e 0 artificial sweetners. The functions suggest ..ii that there are probably two and possibly •a 12 three classes of substances. One class 10 0 0 l'i 0 a (sugar) grows in sweetness at an .. ·..• • 0 • ·:; • ~ 0 , accelerating rate (exponent greater than ...... • ~~ 1.0), whereas the other class grows at a .. • I i! decelerating rate (exponent less than 1.0), · 0 CYClAUATE-SACCHARIN .. SUCIIIOSE CYCLAMATE SALTS SODIUM SACCHARIN .. MIXTURES either mildly (cyclamates) or severely .." I , II , I II i r . 1.6 (sodium saccharin). .. ·0:.·1 , I ,'. I .J, I ,. I ... I ,obs" I .1 .025 .,4 .ze 4, , .025 Direct matches between the sweetness of =:!: ...... sucrose and saccharin suggest that their PERCENT CONCENTRATION BY WEIGHT exponents differ considerably. Two such matches were done in the 1920s and A simple graphical procedure may be REFERENCES reported in the German literature used to calculate relative sweetness and HELGREN, F. J~ LYNCH, M. L, & (Magidson & Gorabachow, 1923; Taufel & KIRCHMEYER, F. J. A taste panel study of equal-sweetness contours. For the the saccharin "off-taste." Journal of the Klemm, 1925). The equal-sweetness sweetness of different substances to be American Pharmaceutical Association, 1955, function (Fig. 2) traced out by their results compared at a single concentration, one 44, 353-355. is linear in log-log coordinates, with a slope need only find the median magnitude LANE, H. C., CATANIA, A. C., & STEVENS, S. of approximately 1.7. If the exponent for estimate given for those concentrations and S. Voice level: Autophonic scale, perceived sucrose is 1.3, then the results of these loudness, and effects of sidetone. Journal of then calculate the ratios. These ratios are the Acoustical Society of America, 1961, 33, direct matches suggest a probable saccharin estimates of relative sweetness at that 160-167. exponent of 0.8, rather than the value 0.3 concentration. MAGIDSON, O. J., & GORABACHOW, S. W. found here. Stevens (1969) also suggests A graphic algorithm may also be used to Zur Frage der Susigkeit des Saccharins. Das that the saccharin exponent is 0.8, rather o-Benzoylsulfimid und seine elektrolytische obtain equal-sweetness functions. A Dissoziation. Chemische Berichte, 1923, 56B, than 0.3. The discrepancy in exponents straight line extended horizontally, at a 1810-1817. may be due to differences in particular sweetness level, will intersect MOSKOWITZ, H. R. Scales of intensity for single concentrations used. In both cases each of the four sweetness functions. The and compound tastes. Unpublished doctoral mentioned the concentrations are lower physical concentrations corresponding to dissertation, Harvard University, Cambridge, Mass., 1968. than those used here. Figure 1 suggests that the points ofintersection are those that are MOSKOWITZ, H. R. Ratio scales of sugar there is a steep drop in the saccharin approximately equally' sweet. sweetness. Perception & Psychophysics, 1970" function at low concentrations, so that the Unfortunately, the graphic procedure must 7, 315-320. function may actually be two-legged. The be used with extreme caution in the PANEK, D. W., & STEVENS, J. C. Psychofit, a flatness of the function at high regions of stimulus concentrations beyond computer program for the treatment of psychophysical data. Laboratory of concentrations may result from the those used here. The behavior of the Psychophysics, Harvard University, 1965, increase in the bitter side-taste with sweeteners at high concentrations suggests PPR 315. increases in concentration, an effect that a decrease in sweetness, but the form of STEVENS, S. S. Sensory scales of taste intensity. would hinder the growth of sweetness. On the decrease is as yet undetermined. Perception & Psychophysics, 1969, 6, 302-308. the other hand, at low concentrations the TAUFEL, K., & KLEMM, B. Untersuchungen bitter taste is faint, if present at all, and Variability uber naturliche und kiinstliche Susstoffe, I. increases in concentration could produce The variability of the magnitude Studien tiber den Siissungsgrad von Saccharin concomitant increases in sweetness. The estimates, both for the sweetness and for und , Zeitschrift fur Untersuchung der actual exponent for saccharin remains open the intensity functions, behaves in a Lebensmittel, 1925,50,264-273. for inquiry. fashion consistent with Weber's law. Figure 3 presents the interquartile range of Relative Sweetness of Sweeteners the magnitude estimates after the The relative sweetness of 15 sugars variability due to the scale sizes of NOTES remained constant across concentrations, a individual Os was removed by modulus 1. Research supported in part by a National convenient outcome of the finding that for equalization. The inter quartile range was Science Foundation predoctoral fellowship and the 15 sugars the sweetness function had computed from the logarithms of the in part by Grant NB-02974 from the National an exponent of 1.3 (Moskowitz, 1970). Institutes of Health. (Laboratory of judgments, and it is plotted against the Psychophysics Report PPR·356). One number, the intercept, sufficed as a logarithm of concentration. The 2. Current address: U.S. Army Natick measure of relative sweetness. In the interquartile range decreases with Laboratories, Pioneering Research Laboratory, present study the large differences found in concentration and usually becomes stable. Natick, Massachusetts 01760. the slopes of the four sweeteners preclude 3. The cyclamate salts and the sodium Flat functions relating variability in saccharin were provided by Abbott Laboratories, a .single measure of relative sweetness. logarithmic units to concentration suggest North Chicago, Illinois. Instead, relative sweetness changes 'that variability is directly proportional to continuously with concentration. concentration. (Accepted for publication October 20.1969.)

42 Perception & Psychophysics, 1970, Vol. 8 (1)