Patterns before recognition: the historical ascendance of an extractive empiricism of forms

The concept of "pattern recognition" is ubiquitous today, encompassing diverse applications from data mining to psychological analysis and artistic design. However, its meaning remains ambiguous, encompassing natural and cultural patterns, human and machine processes. The relationship between human and machine pattern recognition is complex and historically contingent. Some, like Kurzweil, posit a uniquely human ability to recognize patterns contrasted with machine strengths in logic; others, like Goldstone, see it as a distinctive human capacity. Cognitive psychology, influenced by cybernetics, offers more inclusive definitions, applicable to both humans and machines. The widespread adoption of "pattern" as a central concept in various fields, including architecture, media theory, and social sciences, can be observed starting around the 1950s. Works like Kepes's *The New Landscape*, Hall's *The Silent Language*, Alexander's *Notes on the Synthesis of Form*, McLuhan's *Understanding Media*, and Bateson's *Steps to an Ecology of Mind* exemplify the influential role of the concept of informational pattern during that time. The article aims to unpack this complex history, tracing the tensions in pattern recognition's emergence as a general framework across different domains.

The author reviews existing literature on cybernetics, Gestalt theory, and their convergence in shaping the concept of pattern recognition. Key figures discussed include Norbert Wiener (cybernetics), Gestalt psychologists from the Berlin school, and Gregory Bateson, whose work bridges these two approaches. The author draws upon works by Orit Halpern, Reinhold Martin, Larry Busbea, and Seb Franklin, showcasing the ongoing resonance of 1950s-70s discussions around cybernetics, media, and architectural theory in contemporary debates. The author also delves into the critiques of Gestalt-inspired holism in social science by scholars such as Johannes Fabian and Theodor Adorno, highlighting concerns about ahistorical generalizations and the ideological implications of certain formalist approaches.

This article employs a historical and conceptual analysis approach. It traces the development of the concept of pattern recognition through key historical moments and influential figures. The author examines the theoretical underpinnings of cybernetics and Gestalt theory, analyzing their shared ground and points of divergence. The method involves close readings of seminal texts in cybernetics, Gestalt psychology, cognitive science, cultural anthropology, media theory, and science fiction. The analysis focuses on the conceptual transformations and shifts in meaning that the concept of "pattern" has undergone over time. The article also draws upon secondary literature that situates the development of pattern recognition in its historical context.

The paper identifies several key shifts and tensions in the history of pattern recognition. First, it details the initial attempt to translate human-centric concepts of form and pattern (rooted in Gestalt theory) into a cybernetic, machine-based framework. This involved reducing complex, holistic Gestalten into quantifiable and mechanically reproducible units. Norbert Wiener's “travel by telegraph” thought experiment exemplifies this informational approach. The study highlights the significance of the Macy conferences and the works of McCulloch, Pitts, and Bateson in bridging Gestalt and cybernetics. Bateson's tripartite classification of information coding—digital, analog, and Gestalt—is presented as a crucial intermediary stage. The research then delves into the application of Gestalt-influenced holism in social sciences, particularly within cultural anthropology, using examples like Ruth Benedict's *Patterns of Culture* and Edward Hall's work on cultural patterns. The author contrasts this qualitative approach with the emergence of computationally driven pattern recognition during the 1950s and 60s. The work of Oliver Selfridge and Ulric Neisser on "Pattern Recognition by Machine" exemplifies the shift toward a computationally defined notion of patterns. The article emphasizes the "mechanized significance" underlying this shift, showing the attempt to equate human and machine pattern recognition. The author examines how media theorist Marshall McLuhan incorporated pattern recognition into his understanding of media and technology, highlighting the impact of this conceptual extension. Finally, the research explores critiques of the oversimplification of human perception inherent in the computationally driven approach, emphasizing the limitations of data-centric and extractive empiricism and advocating for a more nuanced understanding of pattern recognition that incorporates human intuition and situated knowledge.

The findings address the central research question of how the concept of pattern recognition evolved from a Gestalt-based understanding of holistic forms to its computationally driven counterpart in machine learning. The article showcases the complex interplay between qualitative and quantitative approaches, demonstrating how the attempt to translate Gestalt-inspired understandings into a cybernetic framework led to an "extractive empiricism." This process highlights the limitations of reducing complex human cognitive processes into purely computational terms and highlights the ongoing tension between holistic and reductionist approaches to understanding pattern recognition. The persistence of non-quantitative intuition in computational contexts is highlighted, demonstrating the incomplete displacement of Gestalt theory's influence. The paper argues that overlooking these nuances within social and historical contexts can lead to an impoverished understanding of pattern recognition, reinforcing the need for a more holistic and contextualized approach. The ongoing challenge posed by the powerful and potentially alienating nature of platform-driven data extraction and computationally-driven pattern recognition is critically assessed, advocating for a conscious reconsideration and reappropriation of this technology.

The paper concludes that the concept of pattern recognition has a complex and contested history, shaped by the interplay between Gestalt theory and cybernetics. This historical trajectory has led to a situation where computationally driven approaches dominate, but with notable limitations. Future research should focus on developing a more nuanced model of pattern recognition that integrates qualitative and quantitative elements, embracing situated knowledge and resisting the tendency towards pure data extraction. The potential for reappropriating computationally-driven pattern recognition for critical purposes should be explored.

The paper’s focus primarily lies on the Western intellectual tradition, potentially overlooking important contributions and alternative perspectives from other cultures. While the author critiques extractive empiricism, there is limited engagement with alternative approaches to data analysis and machine learning which prioritize ethical considerations and human-centered design. Finally, the analysis relies primarily on textual sources, and further exploration using alternative methodologies could enrich the discussion.