文摘
A novel, smart, chemical taste sensor that realisticallymimics the behavior of the human gustatory system isdescribed. The taste sensor consists of an array ofelectrochemical sensors that represent the gustatoryreceptors on the human tongue, and a two-phase optimized radial basis function network (RBFN) to representthe human brain, which comprehensively analyzes thegustatory stimulation and judges the overall taste. In theillustrated model, eight electrodes were fabricated todetermine the eight major taste-causing substances, Na+,K+, Cl-, H+, sucrose, glucose, glutamate, and caffeine.The detected signals were fed to a two-phase RBFNoptimized by the implementation of a basis optimizationalgorithm and weight decay term for appropriate dataprocessing. The first phase of the two-phase RBFNquantifies the amount of taste-causing substances in foodsamples from the responses of the electrodes. Theseresults are then fed to the second phase, which correlatesthe amount of substances with the overall taste. The finaloutput is scored on a scale of 1-5 for each of the fivebasic tastes sensed by the human gustatory system, whichare saltiness, sourness, sweetness, bitterness, and umami.The constructed network estimated the intensity of thebasic tastes of 30 drink varieties with an average relativeerror of 7.0% compared to the human scores. The networkcould also estimate the variance in the human sensoryperception. Moreover, the sensor successfully predictedthe interactions of tastes such as suppression of bitternessby sweetness and enhancement of umami by saltiness,which are illusions sensed by the human gustatorysystem. With these abilities, the novel taste sensor canbe considered as a quantitative yet humanlike sensor witha great potential for practical applications.