Sensory influnces. Limits on Sensory Evaluation. Sensory Integration in Individuals’ Acceptances

A. Limits on Sensory Evaluation. The foregoing theoretical and methodological criteria for scientific understanding of influences on food acceptance have, until recently, appeared unachievably strict to those working on the sensory evaluation of foods, even when tackling relatively fundamental empirical or methodological issues. The lack of a psychologically scientific basis for sensory methods has not been noticed because the established atheoretical approaches to human data have helped food research and development personnel out of many problems that are beyond chemical or physical theory and measurement.

A technologically informed search for statistical differences or patterns in verbal responses to food samples has often sufficed to point to practical solutions to common problems, such as matching the current brand when the source of an ingredient changes, identifying and avoiding off-quality features (e.g., staling, spoiling, and taints), or working out the processes that yield the most popular version of a food for an acceptable level of expenditure by the manufacturer.

Moreover, the procedures for getting such pragmatic conclusions from descriptions of many characteristics of food samples by panels of expert, trained, or untrained assessors have been extended to more general issues. Careful use of quantitative descriptive analysis (i.e., “profiling”) has illuminated fundamental problems in food perception, at least in cases in which a monotonic relationship exists in most or all assessors between the levels of each relevant physicochemical factor in the food and the rated intensity of a readily described sensation.

However, sensory interactions are more difficult to elucidate. This is because additive combinations of intensity ratings will not suffice, even if every panelist uses qualitatively the same rule for integrating the two or more physicochemical influences on the complex sensation. Furthermore, without a realistic formula for the interaction, quantitative estimates of the sensed effects of different formulations become unreliable. This could be of practical significance, as well as blocking the development of theory.

Most crucial of all for the extension of sensory evaluation to elucidate influences on food acceptance, there is typically a nonmonotonic relationship between the acceptance response and levels of a physicochemical factor or the relevant intensity scores. This means that the group average response contains little, if any, more information than the distribution of most preferred levels across the panel, obtainable from a preference test without any sensory data. When more than one sensory factor is influencing acceptance, the interactions that enter into the response are lost by constructing a response space from panel data.

Advanced statistical models have been developed to tackle these problems. These have sometimes revealed possibilities for product development not readily identified by conventional market research (or by experts or sensory panels). However, the power of such approaches to distinguish between qualitatively alternative consensus models of sensory integration into acceptance is usually low. The prevalence of the consensus is hard to estimate, and nonconsensus patterns (as well as their prevalences) are not identified.

The reliability of the estimates of the consensus optimum is low for all of these reasons and is much lower than those of a psychological approach at the same data-gathering cost. Even more serious, the validity of the estimates of response in the actual market is quite dubious. The polynomials or ideal points extracted are distorted by forcing everyone through the same set of samples without regard to personal ranges of tolerance. This also degrades the estimate of the individual’s quantitative variant of the consensus preference structure. Worst of all from the practical point of view, as well as for growth in scientific understanding, the conclusions of a pure data-fitting operation apply only to the data set. In contrast, if the data had been collected and analyzed in a way that could identify ways in which the tested sensory variations were influencing each consumer’s acceptances in a normal situation of choice, then some generalizations to other stimulus sets could be made, as well as a rather precise aggregate response surface constructed for whatever subset of the panel was of interest.

B. Sensory Integration in Individuals’ Acceptances. Most appearances, flavors, and textures are themselves based on two or more physicochemical factors in the foodstuff. Insofar as these integrated percepts enter consciously or unconsciously into acceptance, each individual operates according to an algebraic formula for combining the inputs.

I. Intramodal Integration. a. Visual and Tactile Pattern. Psychological research on visual object recognition is, at the moment, grappling with highly abstract fundamental issues, often using verbal and pictorial materials. Visual integration phenomena directly relevant to food acceptance have yet to be examined.

The perception of mechanoreceptor patterns (i.e., tactile textures) has not been extensively studied. Some work on roughness and smoothness of touch to the skin of the arm should be extended to the feel of food in the mouth. Many of the textural sensations from homogeneous fluid foods are empirically related to various measures of viscosity, but the precise uses that the tongue and the palate make of viscous forces, and how they are combined with other physical actions of the fluid, remain to be elucidated. The geometry of the globules in cream is important to its smoothness, for example, over and above the effect of viscosity or thickness. Solid and inhomogeneous foods are not understood psychophysically at all, and progress is slow because of the physical transformations of the stimulus during mastication.

b. Aroma. Olfactory pattern recognition will probably be elucidated first by working “top-down” from the perceptual relationships among food aroma concepts, to complement the currently dominant “bottom-up” approach to odorant receptors and olfactory bulb neurophysiology. This would replace the traditional structure-activity approach of building statistical spaces out of supposed descriptions of subjective features corresponding to receptor types. Instead, personal summations and—more likely—interactions of molecular moieties in the recognition of familiar namable odor sources would be characterized by the discrimination analysis outlined earlier. [See Olfactory Information Processing.]

c. Taste. Gustatory integration is now gaining attention, although effective links to food acceptance have yet to be published. Group analysis of raw ratings have been applied to mixtures of different sweeteners. The results have supported qualitative rules of combination; however, it has been pointed out that physical interactions could account for the observations and so these are not necessarily cognitive integration rules. Moreover, these designs are not strong enough to yield quantitative cognitive integration formulas; in any case this might require individual analysis.

There are many obvious hypotheses as to how tastant perceptions interact to contribute to well-known food flavors. Psychophysical data support the intuitive impression of generally subtractive relationships between sweetness and either sourness or bitterness. Individual combination of discriminable differences from ideal levels of sweetener and coffee solids in instant coffee provides further support.

The individualized discrimination approach to the tastants’ contribution to likings for real foods could elucidate issues such as whether umami (the taste of, e.g., monosodium glutamate in tomato or chicken) is a fifth “basic” taste or only a mixture of the four classic tastes, as it certainly can be described subjectively and classified neurophysiologically. [See Tongue and Taste.]

2. Cross-modal Integration. An advantage of taking the cross-modal integration approach to sensory influences on acceptance is that work can progress in parallel at different levels of interaction, or “chunking.” The natural stimuli for food aromas, tastes, textures, and visual appearances can be used to study the higher level of integration into the entire food percept and on into its most preferred form for a certain purpose.

C. Sensory Influences on Later Food Acceptance. If the only perceptible differences between the foods in the test situation are inherent features (rather than verbal or pictorial labels or packaging, e.g., or sight of the choices that others are making), then sensory influences must at least have contributed to the choice or relative acceptance of one food over the other(s) and indeed to any aftereffects of ingestion.

However, the perceived physical or chemical characteristics of a food can be heavily interpreted, and these attributions affect immediate or subsequent acceptance of foods as much as the “pure” sensations: for example, sweetness or creaminess might be sensually attractive, but feared for its usual health implications. The latter would be a clearly “cognitively” or conceptually mediated sensory influence.

Such effects are immediate, on the acceptance of the same food as is sensed, while it is being sensed. That is what is usually meant by palatability or sensory influences on acceptability. Also, however, the high acceptability of a food, or even the innately liked taste of sweetness itself, can elicit physiological responses that could affect subsequent acceptance of any foods. Proposed mediators of possible delayed effects of sweetness, for example, include neurally triggered changes in liver metabolism or the secretion of insulin from the pancreas or of endogenous opioids in the brain; there is, though, no satisfactory evidence for such effects in adults.

It is arbitrary to assume that a delayed sensory effect is mediated either conceptually or physiologically unless the alternative has been excluded by sensitive measurements. This cannot be done by a single verbal rating or just one biochemical assay, because a wide variety of mechanisms of conceptual and physiological mediation are possible, some quite complex.

On the cognitive front, for example, sweetness from an artificial sweetener in a test food might be interpreted by someone who is trying to limit eating as indicating the consumption of a substantial amount of sugar calories, an idea that might so worry some people that they despair and lapse to unrestrained eating, instead of cutting back in compensation. We have evidence for this disparity between dieters and nondieters after eating soup with a “rich” aroma and texture. This might also happen with a low-calorie creamer.