In the current issue

Conference Report
Yoel Bergman, "Thirty-Eighth Symposium of the International Committee for the History of Technology"

Essay Reviews
Simon A. Cole, "The Faces of Biometrics: Lisa S. Nelson, America Identified; Kelly A. Gates, Our Biometric Future"

Orit Halpern, "Affective Machines: Elizabeth A. Wilson: Affect and Artificial Intelligence"

Philip Mirowski, "Minding the Cybernetic Gap: Andrew Pickering, The Cybernetic Brain"

Megan Mullen, "Demystifying Some Momentous Changes: Richard R. John, Network Nation"

Casey O'Donnell, "On the Back of a Flying Gryphon: Soaring
Over/Through the Global Game Industry: Nick Dyer-Witheford and Greig de Peuter, Games of Empire; Bonnie A. Nardi, My Life as a Night Elf Priest; William Sims Bainbridge, The Warcraft Civilization"

Technology and Culture
January 2012

Affective Machines

Elizabeth A. Wilson,
Affect and Artificial Intelligence

Orit Halpern
The histories of both computing and psychoanalysis are full of performative machines that inspire the imagination, incite desire, and sometimes terrify the mind. We might, for example, recall Claude Shannon's "rat" machine, which skillfully learned to navigate mazes at the Macy conferences in 1951 and amazed a bevy of behavioral and physical scientists with its intelligence and liveliness, only to have a few parameters changed, its memory readjusted and turned into a psychotic beast running in endless and futile circles. We might also think of Walter Grey's small turtles wandering around in their lifelike efforts to find their way, feed themselves, and learn about their environment. Jacques Lacan labeled these small machine-animals "courageous," for they revealed more than most analysts ever could about the nature of the psyche and the operations of drives. Like cybernetics, psychoanalysis is also full of machines and drives—the uncanny, the
influencing machine, the death drive—that operate with sometimes seemingly mechanical repetitiveness and consistency and other times with autonomous unpredictability, causing love or pain, affiliation or hatred. What joins these machines, whether mythic or "real," is their power in allowing us to envision and construct both ourselves and our technologies.

Elizabeth A. Wilson's new book Affect and Artificial Intelligence (Seattle: University of Washington Press, 2010. Pp. xiv+182. $25) is another experiment in imagining machines. Like these aforementioned experiments, she brings the computational and psychoanalytic together to produce a stage upon which to make visible the many different possibilities lying latent in our relationship to, imagination of, and desire for intelligent machines. Wilson demonstrates with elegance in argument and prose that our understanding of what constitutes an artificial intelligence and a machine is deeply inflected by fantasy, performance, and emotion. The stories of cybernetic theatrics and psychoanalytic automata are also apt, because Wilson makes her argument by parading a series of performances on the part of programmers, logicians, robots, and programs that, like Shannon's rat, demonstrate how things can get either very lively and neurotic or very automatic, repetitive, and paranoid with but a few adjustments in parameters.

The arc of the book mirrors the human with the machine, each a doppelgänger haunting and troubling the stability of the other. The book is roughly partitioned into two parts. The first is an examination of the effort to model machine intelligence. Here, Alan Turing's innovations in mathematics, logic, and computing are juxtaposed with the effort to produce intelligence through emotion in the robot Kismet. The second part is the inverse image of the first; it interrogates efforts to model psychology, cognition, and neural nets in humans by way of comparing the computer programs for artificial psychotherapy (ELIZA and PARRY) with the biography of Walter Pitts's personal life and work on neural net logic.

Organizing this mirror narrative is the thesis "that humans have, from the beginning, direct affiliative inclinations for artificial-objects" (p. 95). Wilson argues that we seek affiliation with these artificial objects. It is not merely that we respond, react, or differentiate ourselves from these machines, but that we positively desire them; our relationship to technology is, Wilson imagines, not one of lies, deception, and antagonism, but one of hope, adoration, and aspiration.

Wilson grounds her discussion of subject and mechanism in Ferenczi's famous statement on paranoia and neurosis: "Whereas the paranoiac expels from his ego the impulses that have become unpleasant, the neurotic helps himself by taking into the ego as large as possible a part of the outer world, making it the object of unconscious phantasies" (p. 26). Thus paranoiacs create enemies by expelling painful or repressed desires from within their own drives onto an external world—they "project." In contrast, neurotics redirect their psychic pain toward affiliation, bringing objects and subjects from "outside" in and incorporating them into the subject in a positively constitutive act—they "introject."

While this may all appear distant from concerns of technology and culture, we ignore the implications of these two psychological moves at our peril. In the paranoid model (one that Wilson argues dominates artificial intelligence research and media studies), the object always "provokes our response" (p. 95) and appears to seduce, guile, or mislead us into loving or desiring it. For Wilson, the idea of the machine always acting as a deceitful and misleading mechanism is part of a narrative, and an engineering practice, that makes computing antagonistic to human subjects. When framed within the broader relationship between computing and militarism, one recognizes the importance and value of Wilson's insistence on the affective, emotional, and material nature of human–machine interactions. Our exclusion of affect in the modeling of either our own or machine intelligence ensures that alternative forms of love, care, and object relations will be denied.

While recent texts like Andrew Pickering's The Cybernetic Brain have demonstrated that cybernetics and computing have an embodied and progressive potential, Wilson extends these arguments to interrogate gendered and sexualized difference. She examines the subjective impact on individuals, offering us Turing's ostracism, persecution, and suicide for the repression of his homosexuality and Pitts's self-destruction and ultimately death by alcoholism as examples of what happens when the affective element of computational logic is denied. Wilson also demonstrates how, for example, the exclusion of shame in the Kismet robot's coterie of emotion was linked to its designer's assumptions of human behavior as cognitive, behavioral, and modelable, rather than filiative, responsive, and responsible to those with which the machine must interact.

If there is a limit to this book it lies in two places: attention to historical detail, and interest in the agency of technology. For readers seeking a close examination of experimental practice, a thorough account of historical links between practices and ideas, or the place that technologies, equations, and logics have as autonomous actors, this book will not provide it. Nor does Wilson seek to interrogate the limits and constraints that come from within engineering and programming that also produce affect and effect. This subject-orientation and lack of technology as an actor are refracted in a thin treatment of the structural social, institutional, and historical contexts within which particular decisions are made.

Affect and Artificial Intelligence's real contribution lies in its revisiting of older questions of mind and body, abstraction and materiality, and organism and mechanism—questions that haunt almost all the literature on cybernetics and computing. It takes a new perspective on these questions, asking not whether we are or are not post-human, but rather what kind of humans we would like to become. Like the fantastical games of cyberneticians, psychoanalysts, and communication theorists, Wilson offers a mirror in which we can gain an image of the present and envision possible futures.

Orit Halpern is an assistant professor of history at The New School for Social Research in New York City.

©2012 by the Society for the History of Technology.