Concept Craft | Part 6: Fuzzy Logic & Family Resemblances

A Poetics of AI Art

Michael Filimowicz, PhD

Published in

Higher Neurons

13 min readFeb 2, 2024

This article series is the open access version of my eBook Concept Craft: a Poetics of AI Art.

Chapter 5

Eleanor Rosch is an influential psychologist known for her groundbreaking research in cognitive psychology, specifically in the areas of categorization, concept formation, and the study of human cognition. Her work has had a significant impact on the field, leading to paradigm shifts in our understanding of how humans perceive, categorize, and conceptualize the world around them.

Rosch’s contributions have been instrumental in challenging traditional views of categorization and introducing new perspectives that emphasize the role of prototypes, family resemblance, and the incorporation of cultural and contextual factors.

Rosch received her Ph.D. in Psychology from Harvard University in 1971. Her early research focused on the cross-cultural study of color perception, challenging the prevailing view that color categories are purely biologically determined. Her studies revealed that color categories can be influenced by cultural and linguistic factors, leading to the development of the influential theory of “basic color categories.”

However, Rosch’s most significant contributions came through her research on categorization and concept formation. Her experiments, conducted in collaboration with her colleague and husband, psychologist Roger Brown, examined how people form categories and make judgments about the membership of objects within those categories. The findings challenged the classical view of categorization based on necessary and sufficient conditions and instead proposed the concept of family resemblance and graded membership.

Rosch’s research demonstrated that people tend to categorize objects based on prototype representations, which are the most typical or representative examples of a category. These prototypes capture the central tendencies and common features that are most prototypical of a category, while other members exhibit varying degrees of resemblance to the prototype. Her work showed that humans are better at classifying objects that are more prototypical and that judgments about category membership are influenced by the typicality of an object.

This research had a profound impact on cognitive psychology and influenced a wide range of disciplines, including linguistics, philosophy, and artificial intelligence. Rosch’s work challenged the classical view of categorization and provided a more nuanced understanding of how humans form concepts and make sense of the world. Her concepts of prototypes and family resemblance revolutionized the field, leading to new theories and models of human cognition.

Rosch’s contributions have been recognized with numerous accolades and honors throughout her career. She has received prestigious awards such as the Distinguished Scientific Contribution Award from the American Psychological Association and the Rumelhart Prize in Cognitive Science. Her work continues to be influential, shaping our understanding of human perception, categorization, and the cognitive processes underlying concept formation.

Rosch’s theories of concept formation have made significant contributions to the field of cognitive psychology, particularly in understanding how humans categorize and form concepts. Two key aspects of her work are the concept of family resemblance and the incorporation of fuzzy logic. Let’s delve into these theories in depth:

Family Resemblance

Rosch proposed the concept of family resemblance as an alternative to the traditional view of categorization based on strict definitions or common defining features. According to Rosch, categories or concepts are not defined by a set of necessary and sufficient features but by a family resemblance structure.

This means that members of a category share certain characteristics or features, but there may not be a single feature common to all members. For example, in the category of “bird,” different birds such as sparrows, eagles, and penguins may not share a single defining feature, but they exhibit overlapping similarities.

Some birds may share the feature of having feathers, while others may share the ability to fly. The notion of family resemblance emphasizes the idea that concepts are formed based on a network of overlapping similarities among members rather than a rigid set of defining features.

Rosch’s research showed that family resemblance is a fundamental principle in human concept formation. People tend to categorize objects based on how well they fit into a prototype, which represents the typical or most representative member of a category.

The prototype captures the central tendencies and common features that are most prototypical of a category, while other members exhibit varying degrees of resemblance to the prototype. This theory of family resemblance has had implications for various fields, including linguistics, cognitive science, and philosophy of mind.

Fuzzy Logic

Rosch’s work also incorporated the concept of fuzzy logic into the study of concept formation. Fuzzy logic challenges the traditional binary view of categorization, which assumes that objects either fully belong to a category or do not belong at all. In contrast, fuzzy logic recognizes that membership in a category can be graded or partial, allowing for degrees of membership.

In the context of concept formation, fuzzy logic acknowledges that objects may possess varying degrees of resemblance or similarity to a category. Rather than categorizing objects as either “in” or “out” of a category, fuzzy logic allows for a spectrum of membership degrees. For instance, a chair may be more prototypical of the category “chair” than a stool, but a stool may still possess some degree of membership in the category.

Fuzzy logic provides a more flexible and nuanced framework for understanding concept formation, as it accommodates the inherent variability and context-dependence of human categorization. It acknowledges that concepts are not fixed and rigid, but rather exhibit degrees of membership and fuzzy boundaries.

Rosch’s incorporation of fuzzy logic into her theories of concept formation has had implications for fields such as artificial intelligence, cognitive science, and linguistics. It recognizes the inherent ambiguity and variability in human categorization, providing a more realistic and adaptable framework for modeling cognitive processes.

Rosch’s theories of concept formation emphasize the concept of family resemblance, where categories are formed based on overlapping similarities among members rather than strictly defining features. Her work also incorporates fuzzy logic, recognizing that membership in a category can be graded and that concepts exhibit degrees of resemblance. These theories have had far-reaching implications for understanding human cognition, language, and artificial intelligence, providing a more nuanced and flexible approach to the formation and representation of concepts.

Influences

Rosch’s theories of concept formation have had a significant impact on the field of art and aesthetics. Artists, art theorists, and critics have drawn upon her work to explore and challenge traditional notions of categorization, perception, and the nature of artistic practice. Here are several examples of how Rosch’s theories have been utilized in the realm of art:

Rosalind Krauss, an influential art critic and theorist, incorporates Rosch’s theories in her examination of sculpture and its relationship to categorization. In her essay “Sculpture in the Expanded Field,” Krauss argues that traditional sculptural categories are limiting and proposes a more expansive understanding of sculpture based on Rosch’s concept of prototypes. She suggests that sculptures can be seen as prototypes that possess central tendencies and share family resemblances with other artworks, challenging fixed definitions and encouraging a more fluid approach to categorization.

George Lakoff and Mark Johnson, cognitive linguists, have applied Rosch’s theories to the study of metaphor in art and language. In their book “Metaphors We Live By,” they explore how conceptual metaphors, rooted in bodily experience and grounded in prototypes, shape our understanding of art. They argue that metaphors play a crucial role in our perception and interpretation of artworks, highlighting the influence of Rosch’s work on the study of metaphorical thinking and its implications for aesthetics.

Arthur Danto, a prominent art critic and philosopher, incorporates Rosch’s ideas into his exploration of the artworld and the nature of art. In his book “The Transfiguration of the Commonplace,” Danto draws on Rosch’s concept of family resemblance to challenge traditional definitions of art. He argues that artworks share family resemblances with non-art objects, blurring the boundaries between art and everyday objects. Danto’s ideas highlight the influence of Rosch’s theories on the redefinition of art and the exploration of conceptual art practices.

Suzanne Langer, a philosopher and art theorist, engages with Rosch’s work in her examination of aesthetics and the process of symbolization. In her book “Feeling and Form,” Langer incorporates Rosch’s theories to explain how symbols evoke emotional responses and shape aesthetic experiences. She argues that symbols operate through prototypes and family resemblances, allowing for fluid and dynamic interpretations. Langer’s incorporation of Rosch’s theories highlights their relevance to the study of aesthetics and the understanding of artistic symbolism.

Installation and Conceptual Artists

Many installation and conceptual artists have embraced Rosch’s theories in their artistic practices. These artists challenge fixed categorizations and explore the fluidity of meaning. For instance, artists like Joseph Kosuth, who is associated with conceptual art, often incorporate Rosch’s ideas to blur the boundaries between art and language, emphasizing the role of context and interpretation in artistic communication.

Artists, art theorists, and critics have drawn upon Eleanor Rosch’s theories of concept formation to challenge traditional categorizations, explore the fluidity of meaning, and reshape our understanding of art and aesthetics. By incorporating Rosch’s concepts of prototypes, family resemblance, and graded membership, they have opened up new possibilities for artistic practice, interpretation, and the redefinition of art forms.

Roschian Insights in Prompt Writing

Speculating on how Eleanor Rosch’s theories of concept formation can be applied to prompt writing for generating AI art opens up interesting possibilities for enhancing the creative potential of AI systems.

Fluid Categorization

Rosch’s notion of family resemblance and the fluidity of categorization can be integrated into AI prompt writing for generating art. Instead of rigidly predefined categories, AI systems can be designed to consider overlapping similarities and prototype-based associations among concepts. This allows for a more flexible and dynamic generation of art that transcends conventional boundaries and explores novel connections between diverse elements.

Prototype-based Generation

AI systems can be trained to recognize and incorporate prototypes within the prompt writing process. By capturing the central tendencies and common features of artistic concepts, the system can generate responses that align with the prototypical characteristics of different artistic styles, genres, or themes. This approach enables the AI to produce art that resonates with the essence of specific artistic concepts while allowing for variations and unique interpretations.

Graded Membership and Variation

Rosch’s concept of graded membership can be applied to AI prompt writing to introduce variations and explore different degrees of resemblance to artistic concepts. By allowing for fuzzy boundaries and degrees of association, the AI system can generate diverse outputs that embody different shades of artistic styles or incorporate elements from multiple genres. This promotes creative exploration and encourages the production of art that defies strict categorization.

Contextual and Cultural Sensitivity

Rosch’s research highlights the influence of cultural and contextual factors on categorization. AI prompt writing can be enriched by incorporating cultural references, historical context, and artistic traditions into the generation process. By considering the broader cultural landscape, the AI system can produce art that reflects a deeper understanding of the cultural and historical context, resulting in more meaningful and resonant creations.

User Interaction and Feedback

Rosch’s theories emphasize the role of user feedback in shaping categorization and concept formation. In the context of AI prompt writing for generating art, user interaction becomes crucial. AI systems can be designed to learn from user feedback and adapt their output accordingly, improving their ability to capture the nuances of artistic concepts and refine the generation process over time.

Applications of Rosch’s theories in AI prompt writing for generating AI art require careful implementation, ongoing refinement, and consideration of ethical implications. However, by integrating Rosch’s insights into the design of AI systems, we can envision a more dynamic, context-aware, and culturally sensitive approach to generating AI art that embraces the fluidity and creative potential of concept formation.

Family Resemblances at the Image Output

Rosch’s theories of family resemblance in concept formation provide a valuable lens for understanding and analyzing the image output of generative AI art systems. When AI systems generate images based on prompts, there is often a perceptible family resemblance among the resulting images. Let’s explore how Rosch’s theories can be applied to this aspect in detail:

Prototypical Features

Rosch’s concept of family resemblance suggests that members within a category share certain prototypical features while not necessarily possessing identical characteristics. In the context of generative AI art, the prompts set the stage for the creation of images that align with certain artistic concepts or styles. The resulting images can be seen as sharing prototypical features that correspond to the given prompt.

For example, if the prompt is “generate abstract landscapes,” the AI system would produce a set of images that exhibit features typically associated with abstract landscapes, such as vibrant colors, fluid forms, and geometric abstractions. Although the individual images may differ in specific details, they would display family resemblance by sharing these prototypical features.

Variation within Resemblance

While there is a family resemblance among the generated images, Rosch’s theories also account for variation within the resemblance. This variation allows for the exploration of different interpretations and manifestations of a concept. Generative AI art systems often incorporate randomness and variability, resulting in images that possess unique qualities while still adhering to the overarching family resemblance.

The variation within family resemblance can be observed when examining different outputs from the same prompt. While the images may share common elements, such as color schemes or compositional characteristics, there can be diversity in the specific arrangements, textures, or details. This variation adds richness and interest to the generated art while maintaining the overall coherence and resemblance.

Cultural and Contextual Influences

Rosch’s theories also acknowledge the influence of cultural and contextual factors on concept formation. Similarly, in generative AI art, the resulting images can be influenced by cultural and contextual factors embedded within the training data and the prompt itself.

For instance, if the AI system has been trained on a dataset primarily consisting of artworks from a particular art movement, such as Impressionism or Cubism, the generated images will likely exhibit family resemblance to the visual language and stylistic elements of that movement. The cultural and contextual influences shape the generated images and contribute to the overall family resemblance within specific artistic traditions.

User Perception

Rosch’s theories highlight that family resemblance is a perceptual phenomenon. In the context of generative AI art, the perception of family resemblance among the generated images relies on the subjective interpretation of viewers. Different individuals may perceive and interpret the family resemblance differently, depending on their background, cultural experiences, and artistic preferences.

The viewer’s perception plays a significant role in recognizing and attributing family resemblance among the generated images. Some viewers might focus on color palettes, while others may pay attention to compositional elements or subject matter. This subjective aspect of perception adds a layer of complexity to the analysis of family resemblance in generative AI art.

Rosch’s theories of family resemblance provide a valuable framework for understanding the image output of generative AI art systems. The concepts of prototypical features, variation within resemblance, cultural and contextual influences, and user perception all contribute to the notion of family resemblance in the generated images.

By applying Rosch’s theories, we can deepen our understanding of how generative AI systems create images that exhibit both shared characteristics and individual variations, offering insights into the process of concept formation and artistic creation in the context of AI-generated art.

Creation as Curation: the Artist as a Second Pass on Machine Concept Formation

Rosch’s theories of family resemblance in concept formation offer valuable insights into the curatorial aspect of AI art making, where the artist’s role involves selecting and shaping a series of artworks based on their own sense of family resemblance. Let’s explore in depth how Rosch’s theories inform this process:

AI-generated Family Resemblance

Generative AI art systems produce images that exhibit a family resemblance within the given prompt or artistic concept. The AI system generates a range of images that share prototypical features associated with the prompt. These images may possess variations within the family resemblance, reflecting the inherent diversity and flexibility of the generative process.

The AI-generated family resemblance provides a starting point for the artist’s curatorial role. The artist considers this output as a collection of possibilities, each displaying some level of conformity to the prototypical features associated with the prompt. However, the artist’s engagement goes beyond this initial output to shape the final selection according to their own aesthetic intent.

Artist’s Subjective Evaluation

The artist’s role in the curatorial process involves a subjective evaluation of the AI-generated images. Drawing on their artistic sensibility, the artist filters and assesses the images based on their own understanding of family resemblance. This evaluation considers various factors, including visual coherence, thematic consistency, emotional impact, and alignment with the artist’s artistic vision.

The artist imposes their own sense of family resemblance onto the images, aligning them with their aesthetic intent and conceptual framework. They select the images that best fit their artistic vision, capturing the essence of the prompt while aligning with their personal artistic style, preferences, and thematic concerns. In this sense, the artist becomes the curator who shapes the final collection of artworks through their individual interpretation of family resemblance.

Artistic Intent and Narrative

The artist’s curatorial decisions are driven by their artistic intent and the narrative they seek to convey through the selected artworks. The artist examines the AI-generated images, considering their individual and collective potential in communicating their intended message, exploring a particular theme, or evoking a specific emotional response.

The artist’s curatorial choices may involve selecting images that share common visual elements, such as color palettes, compositions, or visual motifs, thereby reinforcing the family resemblance within the curated collection. The selection may also involve intentionally introducing variations within the family resemblance to create contrasts, juxtapositions, or progressions that enhance the overall aesthetic and conceptual impact.

Coherence and Expression

The curatorial process guided by Rosch’s theories aims to achieve coherence and expression within the curated series of artworks. The artist ensures that the selected images possess an internal harmony and resonant relationship with each other, forming a cohesive body of work. The family resemblance, as interpreted and imposed by the artist, contributes to the aesthetic unity and thematic cohesion of the final collection.

Through their curatorial choices, the artist shapes the narrative and context within which the selected artworks are presented. The artist’s imposition of their own sense of family resemblance allows them to convey a distinct artistic voice and create a cohesive visual dialogue that reflects their unique artistic vision and intentions.

Rosch’s theories of family resemblance inform the curatorial aspect of AI art making by providing a framework for the artist to evaluate, select, and shape the AI-generated images. While the AI system produces its own family resemblance at the image output, the artist’s subjective evaluation and imposition of their own sense of family resemblance play a crucial role in the final selection and curation of a series of artworks.

The artist’s curatorial decisions align the AI-generated output with their aesthetic intent, narrative, and artistic vision, resulting in a coherent and expressive collection of artworks that resonates with their artistic sensibility.