Virgin olive oil (VOO) quality is defined by both chemical and sensory parameters. While the chemical parameters are objective and measured using instrument-based methods, sensory quality evaluation is based upon human panels, which can be subjective, have less repeatability, suffer from fatigue, and require long and costly training. Tasting biases could be minimized by a trained panel, but using humans as a testing instrument is inevitably prone to various psychological biases, stimulus-related factors, and carry-over effects. The objectives of this study were to evaluate instrumental methodologies that will assist the existing human panel in assessing the sensory characteristics of VOO and to develop chemistry-based predicting models for sensory properties in the oil using VOO samples originating from the US and Israel. Our results indicated that oil rancidity highly correlated with the contents of chemical components contents; 1-penten-3-one, 3-hexen-1-ol, (E)-2-pentanal, and 1-octen-3-ol are the major volatiles associated with rancidity defects (low concentrations of these compounds). Positive sensory attributes, such as fruitiness, correlated with 1- acetoxypinoresinol and hexanal, while bitterness correlated with pinoresinol, the aldehydic form of oleuropein aglycones, and the dialdehydic form of oleuropein aglycone. The random forest model suggested that luteolin, (E)-2-hexenal, 1-penten-3-one, and C18:0 are the most useful measurements in predicting the occurrence of sensory defects in the olive oil samples included. In other words, when these compounds are below or above a certain threshold, a defect, such as rancidity, is more likely to be found by the sensory panel.
Read full abstract