An AI analyzed wine reviews and found a surprising link to personality

Your choice of a heavy Cabernet Sauvignon over a light Pinot Grigio might reveal more about your psyche than your palate. New research suggests that specific personality traits, such as openness and extraversion, are reliable predictors of a consumer’s preference for alcohol strength in wine. These findings appeared in the Journal of Personality.

Psychologists utilize a framework known as the Big Five to categorize human personality. This model divides character into five distinct dimensions. These are openness, conscientiousness, extraversion, agreeableness, and neuroticism. Openness measures a person’s desire for new experiences and intellectual curiosity. Conscientiousness tracks discipline and organization. Extraversion involves sociability and enthusiasm. Agreeableness reflects a tendency toward cooperation and social harmony. Finally, neuroticism gauges emotional instability and sensitivity to stress.

Marketers and scientists have previously studied how these traits influence general shopping habits. However, few studies have looked at how personality dictates the specific chemical properties of the products we buy. In the world of wine, alcohol content is a primary characteristic. It is measured as Alcohol by Volume, or ABV. This percentage does more than determine how quickly a drinker becomes intoxicated. It also changes the texture, body, and intensity of the flavor profile.

Xi Wang, a researcher at the School of Culture and Creativity at Beijing Normal-Hong Kong Baptist University, led the investigation. Wang and colleagues sought to understand if the psychological makeup of a consumer drives them toward bolder, higher-alcohol wines or lighter, lower-alcohol options. They aimed to move beyond simple demographics. The team wanted to see if the words consumers use could unlock the secrets of their sensory preferences.

To achieve this, the researchers turned to the massive amount of data available on e-commerce platforms. They focused on textual reviews left by verified buyers. The team collected 9,917 reviews from a major online wine retailer. These reviews spanned nearly a decade of consumer activity. The dataset included the text of the review and the specific technical details of the wine purchased, including its ABV.

The researchers needed a way to translate these thousands of written reviews into psychological profiles. They employed a form of artificial intelligence known as Natural Language Processing. Specifically, they used a model called BERT. This stands for Bidirectional Encoder Representations from Transformers. This tool is designed to understand the nuances and context of human language.

Before analyzing the wine reviews, the team had to teach the AI how to recognize personality traits. They trained the model using a separate dataset called the “myPersonality” project. This project contains thousands of social media status updates linked to verified personality scores. By analyzing these updates, the AI learned which words and sentence structures correlate with specific traits. For example, it learned how an extravert writes compared to a neurotic individual.

Once the AI was trained, the researchers applied it to the wine reviews. The model read the consumers’ feedback and assigned scores for each of the Big Five traits. The team then used a statistical method called beta regression to look for patterns. They checked for connections between the inferred personality scores and the alcohol percentage of the wines those people reviewed. They controlled for factors like price, wine type, and flavor to ensure the results were specific to personality.

The analysis revealed distinct patterns in how different people select wine. Consumers who scored high in openness showed a clear preference for wines with higher alcohol content. High-alcohol wines often have a richer body and more intense viscosity. This creates a complex sensory experience. The researchers suggest that people with high openness seek out this complexity. They are naturally inclined toward novel and stimulating sensations.

A similar trend appeared for individuals high in agreeableness. These consumers also gravitated toward wines with higher ABV. The drivers here appear to be social rather than purely sensory. Agreeable individuals value social harmony and often adhere to group norms. High-alcohol wines are frequently perceived as being of higher quality or prestige. These consumers may select such wines to align with perceived social standards or to gain approval in group settings.

The results for extraversion were unexpected. One might assume that sociable, sensation-seeking extraverts would want the strongest drink. The data showed the opposite. Higher extraversion scores were linked to a preference for wines with lower alcohol content. The authors propose a functional explanation for this behavior. Extraverts thrive on social interaction. They often wish to extend their time socializing. Drinking lower-alcohol wine allows them to consume more over a longer period without becoming overly intoxicated. It is a strategy to maintain social stamina.

Neuroticism also showed a negative association with alcohol strength. Consumers who scored high on this trait tended to buy wines with lower ABV. Neuroticism is characterized by anxiety and emotional sensitivity. Stronger alcohol can amplify loss of control or lead to negative emotional spirals. These individuals likely choose lighter wines as a form of self-protection. They may be avoiding the physiological risks associated with heavy intoxication.

The trait of conscientiousness stood apart from the others. The researchers found no statistical connection between this trait and alcohol preference. Conscientious people are typically disciplined and health-conscious. This might lead them to choose lower alcohol for health reasons. However, they are also quality-oriented and goal-driven. This might lead them to choose high-alcohol wines for their perceived sophistication. These competing motivations likely cancel each other out.

The study does have limitations. The data relied on consumers who take the time to write online reviews. These individuals may not represent the average wine drinker perfectly. Their writing style might differ from the general population. Additionally, the personality scores were inferred by AI rather than measured by direct psychological testing. While the model was accurate, it is an estimation.

Future research could expand on these methods. Scientists could investigate if these personality patterns hold true for other beverages like coffee or craft beer. They could also explore how these preferences change across different cultures. The study focused on a Western e-commerce environment. Cultural norms regarding alcohol and personality expression vary globally.

The study, “From Personality to Pour: How Consumer Traits Shape Wine Preferences and Alcohol Choices,” was authored by Xi Wang, Jie Zheng, and Yiqi Wang.