Matthias Wannhoff

Matthias Wannhoff has studied communications, German studies and media studies at the Universität Bonn and the Humboldt Universität Berlin. In 2014, he received his M.A. degree with a thesis on rivaling concepts of "life" and "death" in modern physics, cybernetics and information theory. Between 2011 and 2014, he worked as a research assistant at the chair of media theories at Humboldt Universität. Among his research interests are theory and history of digital coding, practice and theory of "digital humanities", and the media history of modern psychometrics. Recent publications include Unmögliche Lektüren. Zur Rolle der Medientechnik in den Filmen Michael Hanekes, 2013 Berlin; "Aufheizen/Abkühlen. Bausteine zu einer diagrammatischen Epistemologie des Kalten Krieges." In: S. Höltgen and I. Gradinari (ed.): Heiße Drähte. Medien im Kalten Krieg, 2014 Bochum/Freiburg, pp. 101-122.



Standards of the mind: A media history of intelligence testing

Intelligence tests are experiments that intrinsically rely on mediation. As they strive to measure what cannot be immediately observed and claim exactitude where apparatuses fail, they had to originate their own methods of collecting data. Founded by French scholar Alfred Binet (1857-1911), modern psychometrics could at first sight seem vintage with regards to such gadgets like questionnaires, picture puzzles or fill-in-the-blanks, especially when compared to the psychophysical approach conducted in Germany by Binet’s contemporaries Wilhelm Wundt or Gustav Theodor Fechner.

However, in my project I attempt to reverse the perspective. My interest lies in the question to what extent the experimental setup of modern psychometrics in particular laid the ground for the "mechanization of the mind" (Jean-Pierre Dupuy) determining conceptions of "artificial intelligence" since the 1950s. I claim that there is a mechanistic notion of image and language recognition already embedded in the first intelligence test by Binet and his assistant Théodore Simon in 1905, which has since yielded curious varieties of text and image: Being conceived as semiotic forms maximally rid of ambiguity, uncertainty and the contingency of interpretation, they circumvent any belated opposition between "human" and "artifical" intelligence.

In order to underpin this argument, my approach is threefold: My first aim is to reconstruct the theoretical foundation on which Binet unfolded his very idea of how to measure intelligence, rendering both physiological measuring and psychological introspection as insufficient. A particular focus shall be devoted to Binet’s early work on the subconscious and virtually automatic nature of cognition that has so far been neglected by media studies and also largely by the history of science.

Secondly and technically speaking, I intend to analyze from a semiotic and media-theoretical point of view the interfaces modern psychometrics employ to attack the black box of the human mind. Taking into account that the ambiguous and connotative nature of natural language and image is commonplace, my principal question is to what extent modern psychology‘s methodological endeavour can be fostered in such a semiotic framework. How can one standardize language and sensation?

Thirdly and finally, confronting these psychometric strategies with some key principles in later artificial intelligence research may render fragile the common distinction between "human" and "artificial" intelligence. Both the measurement and simulation of thinking not only requires a systematic, objective modeling of cognitive ability. In their maximal degree of standardization, intelligence tests can also be charged with a criticism notoriously brought up against the work in AI laboratories; that is, an ignorance of the supposedly creative factor in human cognition. On this assumption, intelligence as conceived in modern psychometrics would appear as being "artificial" in the first place.