In lieu of an abstract, here is a brief excerpt of the content:

  • I, Robot
  • Brian O'Keeffe (bio)

The Freudian Robot forces a reckoning with what is arguably the last redoubt of the human being—the unconscious. Robots, cyborgs, machines and automata do not have an unconscious...do they? But what if they do, or what if our understanding of the mechanisms of the unconscious has always been entangled in a thinking of the mechanical? Would machines help us better understand the model of the unconscious we have inherited from Freud? For Lydia Liu, the answer is that we surely would. The unsettling, not to say uncanny consequence of such an undertaking, however, would be to turn human beings into robots: if the workings of the unconscious cannot be distinguished from the workings of a mechanism, an automaton, or indeed a cyborg, on what grounds can we presume to declare our difference from robots?

This question imposes itself every time we observe how close our embrace of technology is, how dependent we are on machines, and how much our language speaks in the idioms of Artificial Intelligence, information technology, and so on. Should we now investigate whether it is not just our bodies, our conscious existences and the language of our conscious self-expression that have been affected by such an embrace, but our unconscious too? Should we speak, today, of a "cybernetic" unconscious? Was Freud already gesturing at such an idea? Was E. T. A. Hoffmann (in Der Sandmann) before him, and after him, the surrealists with their automatic writing, or Joyce? Consider Finnegans Wake and that massive flood of language, a flood washed into print by a mind that ducks below the threshold of consciousness, as well as staying at that threshold, a mind at the limits of deliberate authorship. What data does such a work provide? Data for the psychoanalyst, or the information theorist? Do we have here an exercise in automatic writing, or is it a vast experiment in language processing, the like of which one would [End Page 313] only find at Bell Labs or MIT? (And in their case, can one really assume that their laboratories were not havens for psychoanalytic experimentation?)

Big questions, then, and a correspondingly big book to tackle them. Lydia Liu takes the reader on a historical account that ranges from Freud's reading of Hoffmann towards surrealism and high modernism. It examines critical theory and Derridean deconstruction, Saussurean linguistics and Lacan. But it also studies the development of AI, information theory and the innovations of digital media, both in parallel to developments in the humanities (loosely defined), and in terms of a very definite cross-fertilization, as scientists supplied ideas, discourses and experimental models that had a major impact on the way the unconscious was construed by psychoanalytic theorists. It is a compelling history, and one that she rightly considers needed to be written.

The early part of the book is concerned with approaches to the computerized processing of language. One of the heroes is Claude Shannon, the father of information theory. Highlighted here is his demonstration that the letter for the space between letters (the so-called twenty-seventh letter of the alphabet) makes possible the discrimination of letters, vital for any program that aims to cope with a particular flow of words—to "cope," here, means to set up reliable models of mathematical prediction in order to foresee what letter might come next. There would be no information theory if computers could not wrestle our language into a format in which it could be processed by a program. (And indeed, processed in order to be stored, compressed, archived, as well as sent and received with minimal loss of meaning.) Mathematics was needed, a mathematical theory of communication, so that language itself could be funneled down the wires and circuits of a communication network. Such innovations, Liu observes, are what make digital media possible.

A lot depends, however, on whether one takes the view that language processing of this sort is merely an enabling technology, and in which case meaning is produced by some other process: form, or formatting, not content. Or alternatively, if one regards the enabling process itself as all there is. The meaning is in the messaging...

pdf