Effects of Forgetting on Performance on Various Intensity Scaling Protocols: Magnitude Estimation and Labeled Magnitude Scale (Green Scale)

Abstract

Judges rated the intensity of NaCl solutions using magnitude estimation and the labeled magnitude scale. They performed under four response conditions that varied in reliance on memory: (1) verbal response, (2) written response with no retasting and the response sheet removed, (3) written response with a single response sheet which allowed past scores to be reviewed and amended but with no retasting, (4) the same as ‘3’ but with retasting. Discrimination errors tended to decrease from conditions ‘1’ through ‘4’ yet the major and significant effect was allowing judges to retaste stimuli. The effects of how forgetting lowered discrimination were discussed in the context of experimental design and the absolute versus relative cognitive models of scaling.

‘Different-stimulus’ scaling errors; effects of scale length

“…Yet, both sets of comparisons must be approached with caution because the Rank-Rating and traditional protocols were performed by different judges. Interestingly, earlier studies Kim & O'Mahony, 1998;Koo et al, 2002;Lee et al, 2001) which included both same-and different-stimulus errors, demonstrated a lower overall error rate for RankRating.…”

“…These were determined as in previous studies Kim & O'Mahony, 1998;Koo et al, 2002;Lee et al, 2001;Park et al, 2004). This involved each stimulus being compared with every other stimulus, to determine whether it had a higher, lower or equal intensity score.…”

“…Lawless (Lawless, 1983;Lawless & Malone, 1986b) argued that scaling was relative because of context effects (Parducci, 1963, 1965, 1968Rankin & Marks, 1991;Riskey, 1986;Riskey, Parducci, & Beauchamp, 1979;Schifferstein, 1994Schifferstein, , 1995Schifferstein, , 1996Schifferstein & Frijters, 1992;Stillman, 1993;Vickers & Roberts, 1993) and Poulton's stimulus range equalizing bias (Poulton, 1979(Poulton, , 1989, whereby judges tend to use the same range of responses along a scale, for different intensity ranges of stimuli. Add to this the deleterious effects of forgetting, during scaling (Koo, Kim, & O'Mahony, 2002;Lee, Kim, & O'Mahony, 2001) and it is reasonable to accept that scaling is relative, calibrated judges excepted. Accordingly, judges' scaling data can be visualized as an expression of the rank order of the stimuli, with the scores used as a representation of the perceptual distances between the ranks.…”

“…It explores Jeon et al's (2004) hypothesis that a category scale with fewer categories would induce more 'different-stimulus' errors. Koo et al (2002) and Lee et al (2001) argued that giving a perceptually more intense stimulus, the same or a lower score than a perceptually weaker stimulus could only be regarded as an error, if it could be demonstrated in control experiments, that the appropriate stimuli were actually perceptually stronger and weaker under the conditions of the experiment. This meant that variables that might render the stimuli confusable had to be eliminated.…”

“…They went on to argue that having eliminated the possibility of stimulus confusion and adaptation effects, scaling errors noted during the scaling procedure would be most likely due to forgetting either the intensities of prior tasted stimuli or the scores given to them. They compared a protocol that has been called Rank-Rating Kim & O'Mahony, 1998;Koo et al, 2002;Lee et al, 2001;Metcalf & Vickers, 2002;Park et al, 2004) where judges were allowed to re-taste stimuli ad-lib and also review and adjust their scores, with what will be called more traditional protocols, where such re-tasting and reviewing was not allowed. Not unexpectedly, they found that Rank-Rating, which minimized the effects of forgetting, by allowing adlib re-tasting and reviewing, resulted in fewer scaling errors than the more traditional approach.…”