Can an odour-induced saltiness effect be expressed in grams of salt? A conversion audit of sixteen studies and a minimal reporting standard

Abstract

Correction, version 2.1 (2 September 2026): the first version of this record read the umami column of the source table as saltiness. The saltiness scores are 7.03 to 7.45, not 2.83 to 3.04, and the standard error is 0.14, not 0.04. The sixteen-study audit does not use these numbers and is unaffected. The central claim is strengthened: the spread in derived effect across defensible exponents widens from a factor of 1.6 to a factor of 3.3. One claim is withdrawn. The first version stated that a product carrying 3.6 times the sodium of the reference scored 37% below it; it scored 56% above. The abstract below is corrected throughout. Odour-induced saltiness enhancement (OISE) is reported across many foods and proposed as a route to sodium reduction, but it is unclear whether magnitudes from independent studies can be placed on a common scale. Sixteen studies were read against an extraction schema fixed before reading, and a conversion of one study's reported effect into equivalent sodium chloride concentration was attempted using panel-level data supplied by its corresponding author. That labelled intensity scales anchored in one context cannot be compared across groups is not a new observation; it is the argument of Bartoshuk et al. (2003). The question asked here is narrower and distinct: whether the anchors printed alongside a scale are sufficient to recover a physical concentration from a rating within a single study. The conversion could not be completed. The published record carries a single calibration point, which does not identify a scale: four defensible psychophysical exponents place the same significant 0.42-point effect anywhere between 9.3 and 13.2 g/L NaCl, a spread of a factor of 3.3. The anchor is aqueous while the product is an emulsified meat formulated at 25.0 g/L NaCl, so back-converting the control score through the study's own anchor puts the product well below the concentration at which it was made. In that study's own ballot, every aroma and texture attribute is referenced to a commercial frankfurter or to the study's own control sample, while saltiness and umami are referenced to aqueous solutions. The published record does not give the reason for the split; the point is that in-matrix referencing was demonstrably available to that panel, so the condition proposed here asks for nothing the laboratory had not already shown it could do. The same effect expressed against the product's measured sodium is 1.39-4.58%; expressed against an inferred equivalent concentration it is 2.6-12.3%. The two differ by a factor of 1.9 to 2.7. Both are arithmetically correct, but only the first has a denominator that was measured. Two studies using sodium-anchored intensity scales agree at exactly one concentration, 3.00 g/L, and diverge by up to 1.2 scale points elsewhere. Each group works away from the point of agreement: one runs its odorant testing at 3.00 g/L, where the scales coincide, and the other evaluates at 5.00 g/L, which is where they diverge most. Across the sixteen studies, saltiness was measured on at least six mutually incompatible instruments, and no study reported an effect size, a confidence interval, or a salt-equivalence value for its saltiness result. That claim is tabulated per study and its limits are stated: two of the sixteen are reviews, one is a preprint that measures no saltiness, and four report one of the three statistics for something other than saltiness - confidence intervals on purchase intent, odds ratios on purchase intent, and an equivalence figure quoted from outside the corpus. Saltiness in the converted study was measured to a standard error of 0.14 on a 15-point scale and cannot be expressed in grams of sodium chloride to better than a factor of 3.3. The precision is in the instrument, not in the unit. The conversion assumes a psychophysical model that the descriptive method used does not supply, which is itself the argument: anyone converting such a rating into grams must assume one. Two reporting conditions would make such results convertible: at least two anchor concentrations rather than one, and anchors measured in the test matrix rather than in aqueous solution. Neither is a new psychophysical method, and salt-equivalence methodology already exists. The proposal is a reporting convention, so that the equivalence methods already in use can be applied to published odour-induced results rather than only inside the laboratory that produced them. A seventeenth study, requested during the acquisition and received after the analysis was complete, is reported separately rather than folded into the census. It satisfies neither condition and reports none of the three statistics.