TABLE 1.
Summary of current findings on EEG-based affective computing in marketing.
| Reference | Journal | Marketing substance | Affective states | EEG features | Method (classification accuracy) |
| Reeves et al., 1989 | Human Communication Research | TV commercials | Valence | Hemispheric differences (alpha) | ANOVA |
| Ma et al., 2007 | Neuroreport | Brand | Conflict | ERPs (N270) | ANOVA |
| Ohme et al., 2009 | Journal of Neuroscience, Psychology, and Economics | TV commercials | Valence | Hemispheric differences (alpha) | t tests and Pearson’s linear correlation |
| Chen et al., 2010 | Biological Psychology | E-commerce products | Valence | ERPs (N500) | ANOVA |
| Handy et al., 2010 | Journal of Cognitive Neuroscience | Commercial logos | Liking | ERPs (P1, N2) | ANOVA |
| Ohme et al., 2010 | Journal of Economic Psychology | TV Commercials | Valence | Hemispheric differences (alpha) | ANOVA and post hoc tests |
| Vecchiato et al., 2010 | Brain Topography | TV commercials | Pleasantness | GFP (theta, beta) | ANOVA |
| Vecchiato et al., 2011b | Medical,Biological Engineering and Computing | TV commercials | Pleasantness | PSD, ERD (alpha, theta) | t-test |
| Jones et al., 2012 | Biological Psychology | Pricing | Anxiety | ERPs (FN400, P3, LPC) | ANOVA |
| Guo and Elgendi, 2013 | Journal of Advanced Management Science | Recommender system for e-commerce | Valence | Spectral power (alpha, beta) | Pearson’s linear correlation |
| Khushaba et al., 2013 | Expert Systems with Applications | Food property | Liking | PSD, ERS (delta, theta, alpha, beta, gamma) | Phase locking value |
| Lin et al., 2014 | Frontiers in Neuroscience | Music | Valence, arousal | PSD, DLAT, DCAU, MESH (delta, theta, alpha, beta, gamma) | SVM (valence: 82.5%; arousal: 79.1%) |
| Kuan et al., 2014 | Journal of Management Information Systems | Group-buying information | Valence, liking | Hemispheric differences (alpha) | ANOVA |
| Vecchiato et al., 2014 | Cognitive Computation | TV commercials | Valence, arousal | PSD (alpha), IAF (alpha) | t test |
| Friedman et al., 2015 | International Conference on Affective Computing and Intelligent Interaction | TV commercials | Valence | Spectral power hemispheric differences (delta, theta, alpha, low beta, high beta) | MANOVA, SVM (77.3%), LMT (81.2%) |
| Pozharliev et al., 2015 | Journal of Marketing Research | Luxury goods | Arousal | ERPs (P2, P3, LPP) | ANOVA |
| Telpaz et al., 2015 | Journal of Marketing Research | Consumer goods | Liking | ERPs (N200), spectral power (theta) | t tests and spearman correlation |
| Venkatraman et al., 2015 | Journal of Marketing Research | TV commercials | Valence, arousal | Occipital activity and frontal asymmetry (alpha) | SUR regression |
| Yang et al., 2015 | Journal of Physiological Anthropology | TV commercials | Happiness, surprise | PSD (delta, theta, alpha, low beta, high beta, gamma) | ANOVA, FLDA (happiness: 88.6%; surprise: 87.5%) |
| Berčík et al., 2016a | Periodica Polytechnica Social and Management Sciences | Music preferences | Pleasantness | Spectral power (alpha, beta) | Descriptive statistics |
| Berčík et al., 2016b | Appetite | Store illumination | Valence, arousal, dominance | Spectral power (alpha, beta) | Non-parametric Wilcoxon signed rank test |
| Chew et al., 2016 | Cognitive Neurodynamics | Industrial design | Liking | ERS/ERD (alpha, theta, delta) | SVM (79%), KNN (80%) |
| Gupta and Falk, 2016 | Neurocomputing | Music videos | Valence, arousal, dominance, liking | EEG graph-theoretic features | SVM (valence: 64%; arousal: 64%; dominance: 59%; liking: 64%), RVM (valence: 65%; arousal: 68%; dominance: 63%; liking: 67%) |
| Horska et al., 2016 | Agricultural Economics | Consumer preferences | Valence | Wave fluctuating tendency | Kruskal–Wallis test |
| Lee, 2016 | Journal of Business Research | Willingness to pay | Valence | Spectral power (theta) | sLORETA |
| Gauba et al., 2017 | Neural Networks | TV commercials | Valence | Statistical mean of band oscillations of each electrode | RF (68%) |
| Guixeres et al., 2017 | Frontiers in Psychology | Online commercials | Liking | GFP (delta, theta, alpha, beta, Gamma) | ANN (82.9%) |
| Lobato and Garza, 2017 | IEEE Latin America Transactions | Purchasing behaviors | Liking | Hemispheric differences (alpha) | ANN (76%) |
| Yadava et al., 2017 | Multimedia Tools and Applications | E-commerce products | Liking | Band oscillations (delta, theta, alpha, beta, Gamma) | HMM (70.3%) |
| Avinash et al., 2018 | Procedia Computer Science | Advertisement jingles | Valence | Frontal asymmetry (theta) | KNN (100%), FLDA (90%) |
| Jin et al., 2018 | Frontiers in Human Neuroscience | Eco-labeled products | Valence | ERPs (P2, N2) | ANOVA |
| Wei et al., 2018 | Frontiers in Neuroscience | Commercials | Valence | Wavelength, signal quality (delta, theta, low alpha, high alpha, low beta, high beta, low gamma, high gamma) | SVM (77.3%) |
| Kumar et al., 2019 | Information Fusion | E-commerce products | Valence | Spectral power (delta, theta, alpha, beta, Gamma) | RF (48%), ABC + RF (72%) |
| Aldayel et al., 2020 | Applied Sciences | Purchasing behaviors | Pleasantness | PSD (theta, alpha, beta, gamma) | DNN (94%), RF (92%), SVM (62%), KNN (88%) |
| Shang et al., 2020 | Psychology Research and Behavior Management | Webpage layout | Valence | ERPs (P2, LPP) | ANOVA |
ANOVA, analysis of variance; ERPs, event-related potentials; GFP, global field power; PSD, power spectral density; SVM, support vector machine; IAF, individual alpha frequency; MANOVA, a multivariate analysis of variance; LMT, logistic model tree; FLDA, Fisher linear discriminant analysis; ERS/ERD, event-related synchronization/desynchronization; KNN, K-nearest neighbors; RVM, relevance vector machine; sLORETA, standardized low-resolution electromagnetic tomography; RF, random forest; ANN, artificial neural network; HMM, hidden Markov model; ABC, artificial bee colony; DNN, deep neural network.