With the digital evolution, privacy and perception are becoming more prominent concerns for consumers. Our images are everywhere, thanks to the internet and social media. As a result, consumers are now facing the “face dilemma”—a trade-off between the informative and beneficial use of facial images and the harmful effect of these images being misused or used for discriminatory purposes. Previous research has shown that face perceptions significantly impact many aspects of business, including dating, social networking, employee allocation, and services. As such, using facial images can lead to concerns about privacy and discrimination. To address these concerns, Yinghui Zhou, Shasha Lu, and Min Ding (2020) have developed the contour-as-face (CaF) framework, and they highlight its effectiveness in their recent article in the Journal of Marketing Research.
Can contours help combat the face dilemma?
Yes! The authors find that using the novel CaF framework to show images of contours allows researchers and practitioners to predict how people react when they see someone’s face, without revealing the face itself (i.e., when they see a contour). When we asked the authors about the greatest strength of their research, they emphasized that their research is the first to apply the Fourier transformation technique to resolve the face dilemma. This new method enables privacy protection, combats discrimination, and allows firms to quantify the effects of holistic contour information of facial parts on various face perceptions.
Highlights from the Article
- Facial information is used by consumers and marketers to make inferences in many contexts; meanwhile, consumers are increasingly concerned about privacy and discrimination, resulting in a privacy-perception trade-off.
- The authors provide a comprehensive literature review on face perceptions in business-related contexts.
- They compare facial contour imaging with visual face images for the first time in the academic or practitioner domain.
- The authors introduce the contour-as-face (CaF) framework using novel methodology.
- Three robust empirical studies test the effectiveness of the CaF framework and demonstrate its effectiveness in a real-world context (i.e., online dating).
Generating the Research Question
The idea for this research initially came from the authors’ observation of the face dilemma in the online dating industry: users’ dating decisions are largely determined by profile photos; however, some users are reluctant to post any facial photos due to privacy and discrimination concerns. This observation led the authors to formulate a research question to resolve the trade-off between the value of face perceptions and facial privacy protection. To identify a valid solution to the face dilemma, the authors used the principle of reduction. They sought to determine how much information could be reduced from a photographed face to inhibit recognizability while still preserving enough information for perception. After considering different options, the authors decided to remove all facial information except for the facial contours, which allowed for a rigorous modeling method using the Fourier transformation.
The authors first developed the CaF representation, which presents the contour of a facial image beside the actual face image as the visual stimulus. Then, they performed modeling on the contour-as-face image in two steps: (1) extraction of the features for the contour and (2) quantifying the effects of those facial features through machine learning techniques. To extract the features, the authors use the CD-FD (centroid distance–Fourier descriptor) method, which has been widely used in computer vision applications like image retrieval and object recognition. This technique uses the discrete Fourier transformation, which converts the contours to mathematically precise Fourier coefficients.
Next, the authors tested the effectiveness of this framework in three empirical studies. Study 1 assessed whether humans would have similar perceptions of contours versus facial images. The authors found that the CaF framework was indeed effective as a stimulus and as a modeling method. Specifically, their findings show that humans make inferences on contours consistent with those made with the corresponding face images. They also found that using the contours masked the ability to identify gender or age, thereby enhancing privacy and reducing discrimination.
Study 2 further investigated the effectiveness of the CaF framework in protecting privacy. They found that when participants were asked to pick out the faces of the corresponding contours, they were mostly unsuccessful (i.e., a success rate of 2.68%). Additionally, their findings show that it is quite difficult for humans to guess the age or gender of the contour image. Study 3 then applied the CaF framework to a real-world context using a field study methodology. The authors created an online dating site implementing their framework and found that consumers were willing to use contour images to make decisions in real life, and they appreciated the privacy that contours provided.
The CaF framework provides important implications for business practices. Specifically, it can help practitioners overcome the face dilemma by (1) protecting customers’ facial identity privacy while preserving the ability to infer useful information from the face and improving marketing effectiveness; (2) accommodating customers’ face-based preferences while masking sensitive information (e.g., age, gender) and avoiding discrimination; and (3) developing face-related recommendation systems.
The managerial implications of this research extend beyond the facial data related to areas such as dating, social networking, employee allocation, and service. The CD-FD method can also be used as a feature extraction method—a method widely used in computer vision applications such as image retrieval and object recognition. In addition to faces, it can also be used to analyze other types of contour information in business contexts, such as contours of brand logos, product designs, and package designs.
The authors’ research has several substantial contributions to academic literature. First, the authors introduced a rigorous method to model face data that has never been employed in the marketing literature. Second, the authors used face contours to elicit perceptions from humans on 15 fundamental dimensions, providing a comprehensive understanding of the trade-offs between desirable (e.g., emotional state) and undesirable (e.g., age, gender) facial inferences. Finally, the authors’ CaF framework was shown to effectively address the privacy–perception trade-off associated with using facial data.
Future Research Opportunities
The authors lay the foundation for a fruitful future research stream, and in an effort to springboard upcoming work in the optimal direction, they note several promising extensions of their research:
- Exploring the use of more realistic-looking CaF representation, such as 3-D CaF representation or a representation that includes other facial parts (e.g., ears).
- Testing the CaF framework in larger-scale studies with more diverse faces/contours, which would explore the heterogeneity in how people respond to specific features in contour images of different faces using larger data sets.
- Extending the use of the CaF method to model other types of visual information in related marketing contexts, such as perceptions of brands’ visual representations (e.g., BMW’s iconic grill).
The authors hope their research stimulates further discussion regarding privacy-related problems associated with visual data in various marketing contexts and that their framework becomes a valuable tool for improving consumer welfare in business marketing. It will be exciting to see what insights will be discovered using the authors’ novel CaF framework.
Zhou, Yinghui, Shasha Lu, and Min Ding (2020), “Contour-as-Face Framework: A Method to Preserve Privacy and Perception,” Journal of Marketing Research, 57 (4), 1–23.