What do you think?
Rate this book
Neural Networks
A critical examination of the figure of the neural network as it mediates neuroscientific and computational discourses and technical practices
Neural Networks proposes to reconstruct situated practices, social histories, mediating techniques, and ontological assumptions that inform the computational project of the same name. If so-called machine learning comprises a statistical approach to pattern extraction, then neural networks can be defined as a biologically inspired model that relies on probabilistically weighted neuron-like units to identify such patterns. Far from signaling the ultimate convergence of human and machine intelligence, however, neural networks highlight the technologization of neurophysiology that characterizes virtually all strands of neuroscientific and AI research of the past century. Taking this traffic as its starting point, this volume explores how cognition came to be constructed as essentially computational in nature, to the point of underwriting a technologized view of human biology, psychology, and sociability, and how countermovements provide resources for thinking otherwise.
Neural Networks proposes to reconstruct situated practices, social histories, mediating techniques, and ontological assumptions that inform the computational project of the same name. If so-called machine learning comprises a statistical approach to pattern extraction, then neural networks can be defined as a biologically inspired model that relies on probabilistically weighted neuron-like units to identify such patterns. Far from signaling the ultimate convergence of human and machine intelligence, however, neural networks highlight the technologization of neurophysiology that characterizes virtually all strands of neuroscientific and AI research of the past century. Taking this traffic as its starting point, this volume explores how cognition came to be constructed as essentially computational in nature, to the point of underwriting a technologized view of human biology, psychology, and sociability, and how countermovements provide resources for thinking otherwise.
- GenresArtificial Intelligence
124 pages, Kindle Edition
Published April 9, 2024
15 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 2 of 2 reviews
January 17, 2025
Suchman’s conclusion is “deep learning? What about we deeply learn about each-other instead?” That’s about the depth of it.
August 25, 2024
A very urgent book whose objective is a critical engagement with one of the most pervasive - and I think interesting - technoscientific concepts of our day: the neural network. I came to this book primarily because I was interested in what Lucy Suchman, an esteemed scholar within science and technology studies (STS), has to say about neural networks. Accordingly, I only read the book's introduction as well as Suchman's essay. There are two further essays which also looked interesting.
The introduction, co-written by Dhaliwal, Legape-Richter, and Suchman, acquaints us with the book's aim and methodological strategy. To speak in the book's words, "this book is an exploration of the conjuncture of nature and artifice enacted in the figure of the neural network" (p. 1). Obviously, the highly technological and scientific concept of the neural network remains somewhat rooted in "nature"; that is, in the human brain and its neurons. The authors explain their ambition to methodologically approach neural networks from a position of ambivalence (p. 2); more precisely, they seek not to decry neural networks as good or bad. Instead, this strategy is a "process of slowing down and learning to inhabit a compromsied environment with the discomfort, contradiction, and misalignment it entails" (p. 3). They liken it to Haraway's (2016) 'staying with the trouble'. They astoundingly point out the position of neural networks today: "neural networks have come to be recognized as some sort of universal machine capable of capturing the essential qualities of any system" (p. 3). They attempt to retrace how this became the way it is; so as to "locate and restore the various settings and configurations throguh biological invocations became [...] the structuring figures of contemporary conceptions of AI". (p. 5). The book, thus, seeks to study neural networks in specific sociotechnical contexts. While neurons were first theorized in the 1880s, networks were in the 1840s. How did these concepts come to be thought in conjunction? Neural networks, the authors suggest, are not created, discovered, found, or generated; instead, they are rendered (p. 13).
In Suchman's essay, which is the book's last, we find a diffractive reading of works by both Geoffrey Hinton (neural networks researcher) and Gillian Einstein (feminist neuroscientist). She examines "how the brain/computer analogy breaks differently" for these two scientists. While Hinton is committed to the analogy (but acknowledges its limits), Einstein takes a more critical approach. Essentially, Suchman argues in this chapter that the theory underlying computational neural networks is cognitivism - the idea that there is a correspondence ebtween mentral representations formed in brains and the 'external' world. By contrast, recent decades have seen the emergence of thorough critiques of cognitivism. In a nutshell, these critiques highlight the inseparability of cognition "from the lived experience of embodied persons-in-relation, with one another and with culturally and historically constituted social and material worlds." (p. 88). Then, Suchman traces how the "polysemy" of the term neuron enables Hinton to flexibly evoke the term as an analogy and identity; in other words, Hinton is somewhat ambiguous about the extent to which the brain's neurons serve as an inspiration as opposed to as a direct model. In the works of Gillian Einstein, on the other hand, we find a different approach to the figure of the neuron. Einstein takes an approach to the body that rejects the idea of clearly separable and atomized parts, instead emphasizing the fundamental interconnectivity among all parts of the body (p. 101). In Suchman's words: "For Einstein, in contrast, the biological brain is taken as inseparable from the body-in-the-world, and a neuroscience capable of fundamental insight requires methods that expand to incorporate their object's constitutive relations." (p. 104). It is the conclusion in which Suchman becomes here strongest: "The delineation of brains and computational systems, and the articulation of sameness and/or difference between them, is not an innocent matter of objective observation but rather a project of worlding in which all of us are engaged in the discussion are implicated" (p. 106). Towards the end she mobilizes Phil Agre's critical technical practice as a "continually unfolding awareness of its own workings as a historically specific practice." (p. 107).
To conclude, it might be interesting to link Suchman's contributions to this book to earlier works of her, where she critically approaches the ways in which humans and machines are delineated from one another. Similarly, this analysis of neural networks can be seen in terms of an analysis of boundaries: how has the neural network become isolated and figured as an universal model for modelling human intelligence? What sociomaterial practices have shaped this isolation of the neural network? What are the boundaries of the neural network? Suchman does not offer a definite judgment of whether the model of the neural network is a good or bad one; the juxtaposition of Hinton and Einstein instead allows a situated analysis of how the extent to which the neural network is seen as inspired by "nature" is deeply contingent. At most, her essay reads as a critique of cognitivism and an embrace of Einstein's feminist proclivities. All in all, a recommendable read.
The introduction, co-written by Dhaliwal, Legape-Richter, and Suchman, acquaints us with the book's aim and methodological strategy. To speak in the book's words, "this book is an exploration of the conjuncture of nature and artifice enacted in the figure of the neural network" (p. 1). Obviously, the highly technological and scientific concept of the neural network remains somewhat rooted in "nature"; that is, in the human brain and its neurons. The authors explain their ambition to methodologically approach neural networks from a position of ambivalence (p. 2); more precisely, they seek not to decry neural networks as good or bad. Instead, this strategy is a "process of slowing down and learning to inhabit a compromsied environment with the discomfort, contradiction, and misalignment it entails" (p. 3). They liken it to Haraway's (2016) 'staying with the trouble'. They astoundingly point out the position of neural networks today: "neural networks have come to be recognized as some sort of universal machine capable of capturing the essential qualities of any system" (p. 3). They attempt to retrace how this became the way it is; so as to "locate and restore the various settings and configurations throguh biological invocations became [...] the structuring figures of contemporary conceptions of AI". (p. 5). The book, thus, seeks to study neural networks in specific sociotechnical contexts. While neurons were first theorized in the 1880s, networks were in the 1840s. How did these concepts come to be thought in conjunction? Neural networks, the authors suggest, are not created, discovered, found, or generated; instead, they are rendered (p. 13).
In Suchman's essay, which is the book's last, we find a diffractive reading of works by both Geoffrey Hinton (neural networks researcher) and Gillian Einstein (feminist neuroscientist). She examines "how the brain/computer analogy breaks differently" for these two scientists. While Hinton is committed to the analogy (but acknowledges its limits), Einstein takes a more critical approach. Essentially, Suchman argues in this chapter that the theory underlying computational neural networks is cognitivism - the idea that there is a correspondence ebtween mentral representations formed in brains and the 'external' world. By contrast, recent decades have seen the emergence of thorough critiques of cognitivism. In a nutshell, these critiques highlight the inseparability of cognition "from the lived experience of embodied persons-in-relation, with one another and with culturally and historically constituted social and material worlds." (p. 88). Then, Suchman traces how the "polysemy" of the term neuron enables Hinton to flexibly evoke the term as an analogy and identity; in other words, Hinton is somewhat ambiguous about the extent to which the brain's neurons serve as an inspiration as opposed to as a direct model. In the works of Gillian Einstein, on the other hand, we find a different approach to the figure of the neuron. Einstein takes an approach to the body that rejects the idea of clearly separable and atomized parts, instead emphasizing the fundamental interconnectivity among all parts of the body (p. 101). In Suchman's words: "For Einstein, in contrast, the biological brain is taken as inseparable from the body-in-the-world, and a neuroscience capable of fundamental insight requires methods that expand to incorporate their object's constitutive relations." (p. 104). It is the conclusion in which Suchman becomes here strongest: "The delineation of brains and computational systems, and the articulation of sameness and/or difference between them, is not an innocent matter of objective observation but rather a project of worlding in which all of us are engaged in the discussion are implicated" (p. 106). Towards the end she mobilizes Phil Agre's critical technical practice as a "continually unfolding awareness of its own workings as a historically specific practice." (p. 107).
To conclude, it might be interesting to link Suchman's contributions to this book to earlier works of her, where she critically approaches the ways in which humans and machines are delineated from one another. Similarly, this analysis of neural networks can be seen in terms of an analysis of boundaries: how has the neural network become isolated and figured as an universal model for modelling human intelligence? What sociomaterial practices have shaped this isolation of the neural network? What are the boundaries of the neural network? Suchman does not offer a definite judgment of whether the model of the neural network is a good or bad one; the juxtaposition of Hinton and Einstein instead allows a situated analysis of how the extent to which the neural network is seen as inspired by "nature" is deeply contingent. At most, her essay reads as a critique of cognitivism and an embrace of Einstein's feminist proclivities. All in all, a recommendable read.
Displaying 1 - 2 of 2 reviews