Bourdieu Vectors: A Framework for Quantifying Taste in Social Space
[1]S. Hecht, “Bourdieu Vectors: A Framework for Quantifying Taste in Social Space,” 2025, http://dx.doi.org/10.2139/ssrn.5585570
Listen to a NotebookLLM podcast discussing the method:
Bourdieu Vectors is a method for describing social practices, preferences, or activities using attributes inspired by the social fields of Pierre Bourdieu. Each social practice is represented as a multidimensional vector that quantifies its alignment across different types of capital—economic, cultural, social, and symbolic. These vectors allow us to systematically quantify taste, revealing which cultural, social, or economic characteristics are typical of people liking or engaging in a given practice. Each value in the vector represents the profile of person who likes or engages in that social practice, activity, or preference.
Applications are: social research and taste analysis or recommendation
Example: Vector for "american football"
Let's look at what the vector for "american football" looks like. Attributes that describe the social and demographic profiles of individuals associated with "american football"are evaluated and quantified by a large language model (LLM). Try out different events, activities, or preferences to see how their vectors differ! This could be "jazz concerts", "hiking", "taylor swift", "art exhibitions", "vegan cooking", or anything else you can think of. Also t to set the context to "watching it" or "playing it" when you try out a sport.
| Attribute | Value |
|---|---|
| cultural capital cultural art | 0.00 |
| cultural capital cultural culinary | 0.00 |
| cultural capital cultural design | 0.00 |
| cultural capital cultural entertainment | 0.70 |
| cultural capital cultural fashion | 0.30 |
| cultural capital cultural literature | 0.00 |
| cultural capital cultural music | 0.20 |
| cultural capital cultural theater | 0.00 |
| cultural capital educational certifications | 0.10 |
| cultural capital educational formal | 0.10 |
| cultural capital educational informal | 0.30 |
| cultural capital educational teaching | 0.10 |
| cultural capital educational training | 0.10 |
| cultural capital environmental conservation | 0.00 |
| cultural capital environmental sustainability | 0.00 |
| cultural capital innovation creativity | 0.10 |
| cultural capital innovation openness | 0.10 |
| cultural capital lifestyle gaming | 0.30 |
| cultural capital lifestyle outdoor | 0.50 |
| cultural capital lifestyle travel | 0.40 |
| cultural capital media digital | 0.70 |
| cultural capital media entertainment | 0.80 |
| cultural capital media journalism | 0.50 |
| cultural capital media social | 0.70 |
| cultural capital media traditional | 0.80 |
| cultural capital medical fitness | 0.40 |
| cultural capital medical healthcare | 0.10 |
| cultural capital medical prevention | 0.10 |
| cultural capital medical research | 0.00 |
| cultural capital religious affiliation | 0.00 |
| cultural capital religious community | 0.00 |
| cultural capital religious practice | 0.00 |
| cultural capital religious rituals | 0.00 |
| cultural capital sport competition | 0.90 |
| cultural capital sport fitness | 0.60 |
| cultural capital sport individual | 0.00 |
| cultural capital sport outdoor | 0.80 |
| cultural capital sport team | 0.90 |
| economic capital economic assets | 0.30 |
| economic capital economic entrepreneurship | 0.10 |
| economic capital economic income | 0.30 |
| economic capital economic investments | 0.10 |
| economic capital economic luxury | 0.20 |
| economic capital economic wealth | 0.20 |
| economic capital power economic | 0.10 |
| habitus subjective emotional intelligence | 0.20 |
| habitus subjective lifestyle alignment | 0.30 |
| habitus subjective motivation | 0.20 |
| habitus subjective resilience | 0.20 |
| habitus subjective satisfaction | 0.30 |
| habitus subjective selfesteem | 0.20 |
| habitus subjective values | 0.30 |
| overall cultural capital | 0.30 |
| overall economic capital | 0.18 |
| overall habitus alignment | 0.24 |
| overall social capital | 0.13 |
| overall symbolic capital | 0.32 |
| social capital community local | 0.50 |
| social capital community volunteering | 0.20 |
| social capital legal judiciary | 0.00 |
| social capital legal law enforcement | 0.00 |
| social capital legal legislation | 0.00 |
| social capital legal profession | 0.00 |
| social capital network professional | 0.20 |
| social capital network social | 0.60 |
| social capital political activism | 0.10 |
| social capital political engagement | 0.10 |
| social capital political government | 0.00 |
| social capital political parties | 0.00 |
| social capital power political | 0.00 |
| symbolic capital power cultural | 0.60 |
| symbolic capital power educational | 0.10 |
| symbolic capital power professional | 0.10 |
| symbolic capital power public | 0.10 |
| symbolic capital power social | 0.50 |
| symbolic capital power symbolic | 0.50 |
Example: Taste Map
Scale: 1.00
Example: Visualization of vectors between upper, middle and low class associated terms
The following example shows a Visualization of vectors between upper (=high), middle and low class associated terms. As you can see the upper and low class associated terms are distinctive. While upper and low class are well separated the middle class acts as a transitional zone.

Visualization of separation between upper (=high), middle and low class associated terms
Dataset: bourdieuvectors-ds-1000-high-mid-low-eyebrow-1-0-1
What can I do with these vectors?
Bourdieu vectors provide a powerful way to analyze social practices and quantify cultural tastes. Here are some key applications:
- Measure similarity: You can quantify how similar two social practices are using metrics such as cosine similarity. For example, the similarity between "american football" and "basketball" might be high, while the similarity between "american football" and "banana" would be low.
- Generate recommendations: By identifying practices with the highest or lowest similarity to a user’s interests, you can suggest new cultural events, activities, or experiences tailored to their taste.
- Create interest profiles: Averaging multiple vectors allows you to build a profile of a person’s or group’s overall taste, capturing preferences across different types of capital and social practices.
Ideas for Applications of Bourdieu Vectors
- Compare social practices across different groups, regions, or demographic segments.
- Identify emerging trends in cultural consumption or participation.
- Recommend cultural events, products, or activities based on similarity to a user’s profile.
- Suggest complementary interests by finding vectors that are semantically or socially related.
- Cluster users or audiences based on their taste profiles to tailor marketing or content strategies.
- Quantify the appeal of products or experiences to different segments.
- Understand access to different forms of capital (cultural, social, economic) in populations.
- Guide policy-making or educational interventions based on the distribution of cultural and social resources.
- Map media, arts, or entertainment content along social and cultural dimensions.
- Evaluate how well content aligns with different social or cultural capital profiles.
- Track changes in tastes or cultural practices over time.
- Compare past and present practices quantitatively.
Ethical Considerations
The framework avoids profiling based on political ideology or other sensitive attributes. Scores are strictly limited to estimated capital levels, and no personally identifiable information is used. Researchers and users are encouraged to apply the method responsibly, with transparency about the assumptions and limitations.
Ethical Considerations:
⚠️ Practice-Level Analysis Only: Bourdieu Vectors represent associations between cultural practices and capital dimensions—not individual attributes. The framework should be used for analyzing cultural practices, fields, and aggregate patterns, not for profiling or making decisions about specific individuals.
🚫 Prohibited Uses: Do not use Bourdieu Vectors for employment screening, creditworthiness assessment, educational admissions, or other high-stakes individual decisions without informed consent, independent validation, and regulatory oversight.
🌍 Cultural Context Matters: This framework was developed and validated using English-language data and Western cultural contexts. Capital dimensions and taste hierarchies vary substantially across societies. Users should specify cultural and geographic contexts when generating vectors and exercise caution when applying the framework to non-Western cultural practices.
📊 Bias & Limitations: LLM-generated estimates reflect societal biases, stereotypes, and hierarchical assumptions present in training data. Vectors capture collective cultural schemas—the "common sense" understandings of social positioning—rather than objective measurements. They complement but do not replace empirical measurement.
🔍 Transparency: All prompts, dimensions, and methodology are publicly documented. Scores are based on practice-level associations, not individual-level data. The framework has been validated across multiple LLMs (0.84-0.95 intra-model stability, 0.86-0.96 inter-model agreement) and against empirical GSS data (69.8% significant correlations, 65.8% classification accuracy).
✅ Responsible Use: This framework enables research in social sciences, digital humanities, and cultural analytics. Users bear responsibility for ensuring ethical use, validating outputs in their specific contexts, and critically interrogating assumptions embedded in LLM-generated estimates.