Correlation between sensory data, instrumental data (gas sensors, physico-chemical analysis) and palatability measurements of twelve moist foods for cats

Twelve moist foods for cats were characterised using different analytical m ethods (sensory analysis, gas chromatography (GC) coupled with a mass spect rometer, texture and gas sensor measurements) and the links between the dif ferent data sets and cat preferences were analysed. Gas sensor measurements were reliable, discriminating and non-correlated with the water activity o f the products. Multiple Factor Analysis described the relations between th e different data sets while GC and sensory data helped to give a better und erstanding of the reasons for preferences. Although gas sensor data gave a representation of the products quite similar to that obtained with palatabi lity results, these data gave less rich information on product sensory char acteristics than the other types of data. Partial Least Squares regression produced predictive models and revealed quantitative links between data set s. The various data sets give complementary information which facilitate th e prediction of palatability. In particular, some palatability parameters w ere predicted through the responses of certain gas sensors and texture meas urements. Even if sensory data give better predictions, gas sensor and text ure measurements have the advantage of being rapid.