Description
- Abstract:
- This dissertation takes as its focus three episodes when the notion of neural networks was not only deployed as a solution to the problem of intelligence, but also framed as such by being directly inscribed onto bodies that were marked as the bearers of various forms of human difference. While now known as a biologically inspired approach to artificial intelligence relying on probabilistically weighted neuron-like units to recognize patterns in data, neural networks were first introduced as a neuroanatomical approach to racial difference before being recast into a new way to both diagnose and treat mental illness based on neurophysiology alone. Taking on a media historical approach, this project explores how neural networks’ various iterations indexed not only changes in the articulation of intelligence, but also the development of new means to locate and attribute that faculty across humans and machines alike. By looking at the ways neural networks were materialized through practices as varied as silver staining and colonial hygiene policies, brain lacerations and electroshock therapy, and ultimately organizational reforms and the software implementation of simulated neurons, it argues that this model was directly inscribed onto certain bodies to produce a recognition of who—or what—qualified as an intelligent subject and who doesn’t. Far from signalling the gradual convergence of human and machine intelligence, neural networks’ various iterations exhibited a sustained experimental commitment that systematically involved intervening on, managing, and disciplining human bodies, while devaluing the practical, local, and contextual knowledge they held. By recovering the experimental and other situated practices underwriting neural networks’ history, this dissertation proposes to understand neurality itself as a mediating figure that, throughout the twentieth century, conflated communication with structure, meaning with medium, and cognition with neurophysiology. Building on this, it theorizes a broader class of media termed neural media defined by the way they communicate a highly technologized view of human biology, psychology, and sociability, while promoting the organizational principles embodied by neural networks—namely, atomization, interdependence, mediation—as biologically given.
- Notes:
- Thesis (Ph. D.)--Brown University, 2023
Citation
Lepage-Richer, Théo,
"Neural Media: Ascribing, Animating, and Artificializing Intelligence"
(2023).
Modern Culture and Media Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://repository.library.brown.edu/studio/item/bdr:fwnhyfqf/
Relations
Collection:
-
Modern Culture and Media Theses and Dissertations
Theses and Dissertations for the Modern Culture and Media department....