The application of EEG power for the prediction and interpretation of consumer decision-making: A neuromarketing study

The application of biometric data has been the center of attention for neuromarketing researches. Understanding the underlying mechanisms behind consumer shopping behaviors and the way that advertising affects such behavior are the most important issues that need more investigations. In this study, two purposes were focused including (1)the potential of EEG spectral power for prediction of consumers' preferences and (2)interpretation of the alteration of consumers' decision-making in shopping behavior when the content of an advertisement including background color and promotions was changed. For this purpose, advertisements related to different mobile phone brands which were different according to the content were shown to the participants followed by EEG (electroencephalography) recording. The power of the EEG data was used for finding the most important brain regions for distinguishing between preferences and predicting the incidence of decision-making. Furthermore, the results were used for interpretation of the observed participant behavior. The obtained results showed that the extracted features from EEG power could predict consumer's decision-making incidence with relatively high accuracy (>87%) and distinguished between “Like” and “Dislike” preferences with accuracy higher than 63%. Also, the most discriminative channels for predicting the incidence of decision-making about liking/disliking or buying a product were found to be frontal and Centro-parietal locations (Fp1, Cp3, Cpz) while the difference between “Like” and “Dislike” decisions was observed mostly in the frontal electrodes (F4 and Ft8). Furthermore, the results showed that adding the background color to the designed advertisement had a negative impact on the degree of liking a product. In conclusion, EEG data analysis can be used as a useful tool for predicting costumer decision-making, while in order to obtain higher accuracies, other features should be tested for distinguishing between different preferences.