Untitled Document

In her paper “Cognition Everywhere,” Katherine Hayles clarifies the differences between conscious, unconscious and nonconscious cognition by stating that “while all thinking is cognition, not all cognition is thinking. (...) the cognitive nonconscious operates at a lower level of neuronal organization not accessible to introspection” (Hayles, 2014). Conscious cognition is thus associated with self-reflexive awareness. On the other hand, while the unconscious may be reached through introspection, the nonconscious cannot.

According to Hayles, there are four criteria that make nonconscious systems cognitive. They

The crucial distinction between material and cognitive processes is choice. The action of choosing is context-dependent and context-driven, and it necessarily implies interpretation: if there is no choice, but only one option, there is nothing to interpret. A computer process, despite being deterministic, makes interpretative choices as it performs its tasks (yes or no, if then else, ...), which implies interpretation. As Hayles states, “[I]nterpretation is deeply linked with meaning (...). [F]or the cognitive nonconscious, however, meaning has no meaning.” This means that meaning is one thing for nonconscious cognitive systems and another for human subjects. To avoid this terminological problem, perhaps we should use the term “information” instead of “meaning.” I believe the difference between these two notions is central to the discussions on technology in the context of the humanities.

The Online Etymology Dictionary associates “interpretation” with “translation,” “explanation” or “exposition.” [1] The act of interpreting is linked with translating something into different terms, as a way to explain, expose or understand. In this sense, interpretation is prior to meaning production: we interpret by translating something, making it intelligible and, hence, meaningful. And, indeed, a computer interprets information precisely by translating one language into another, and commands into actions. The same dictionary associates “meaning” with “intend,” “have in mind,” “signify,” “make known,” “have an opinion,” or “to think.”[2] All these terms express a subjective position. Meaning is thus subjective and hence, ambiguous, it is relative and contextual. In contrast, information is quantifiable precisely by ignoring context. When Claude Shannon developed his information theory he saw how context adds noise to information, making the latter impossible to quantify. His solution was to discretize information, separating it from context. This digitization process inevitably implies a reduction, or compression. In this sense, perhaps we can say that information is what remains of the digitization of meaning. While information is quantitative, meaning is qualitative. Irreducible to the unambiguous discreteness of information, meaning is thus virtually incomputable. Moreover, while the regime of information is concerned with communication, or with efficiency in the transmission of messages, the regime of meaning is concerned with expression, or with the gradients of ambiguity that noise and context enact. Literature doesn’t belong to the field of information and communication, but to that of meaning and expression. But the regime of information that characterizes the cognitive abilities of our technical devices seems to be more and more culturally pervasive. In a cybernetic and biopolitical context, concerned with statistics and preemptive control, distant reading strategies and the search for patterns in the chaotic complexity of the world seem to be allegories of how human cultures are being influenced by technical nonconscious modes of cognition, associated with the efficiency of beehives or computers.

Digital devices are imbued with artificially generated cognitive abilities that are not neutral. Rather, their design reflects and reinforces the socio-political ground on which they operate. The relationships between digital technologies and the institutions of power leave an inevitable imprint on our digital devices’ operating modes, affecting their usage and perception as cultural objects and as actants. In his analysis of the relationships between contemporary cognitive capitalism and digital technologies, Matteo Pasquinelli has argued that economy and technology have co-evolved, mutually reinforcing each other:

Indeed, trading algorithms show us how the monetary and technological abstraction of value became intertwined: they embody the way capitalism accompanied the emergence of nonconscious cognition in technical systems, showing us how capital has become self-intelligent.

So how do these agents work? Nonconscious cognizers operate under the radar of human perception: they are too fast to be grasped, too small to be seen, too specialized and obscure to be understood.

Figure 1. Cognitive timeline presented by Katherine Hayles at the Rethinking the Mind of Architecture conference [3].

This cognitive timeline, presented by Katherine Hayles at a conference called “Rethinking the Mind of Architecture,” shows the temporal scale within which trading algorithms operate. While human consciousness takes half a second to process information, trading algorithms take one to five milliseconds. There is thus a “missing half-second” in human consciousness, and a gap between the latter and electronic interactions in non-organic materials (such as those that take place through optical fiber networks), which are physically faster than neurotransmission in brain circuits.

In a world increasingly inhabited by interconnected algorithmic agents whose cognitive modes belong to a temporal and perceptual scale that is incompatible with that of human consciousness, it becomes important to reflect on what it means to articulate human life with cognitive technical systems. We are already taking advantage of our digital media cognitive abilities, for example in the digital humanities, using computation to macro-analyze vast quantities of cultural data. This quantitative aspect of information management has proven its importance in diverse fields of human culture, and it has fueled the development of our digital technologies through, precisely, their economical embodiments. So although our digital tools have much to offer us in what concerns information and knowledge, they must also be understood as artifacts that are deeply and already embedded in the economic and technological infrastructure.

Rather than polarizing the debate on technics between techno-fobia and techno-filia, one must come to terms with the contradictory nature of technology. Bernard Stiegler recovered the Greek notion of the pharmakon to consider how technology is always a poison and a remedy at the same time. Artists and poets have also been critically exploring programmable tools in order to better understand their affordances and constraints. The case study I am bringing to our discussion is one of such experiments. HIIICT is a generative literary work by John Cayley and Daniel C. Howe, based on the programming of an algorithm to produce a text. But unlike most generative literary works, which are based on combinatory procedures that reshuffle a predefined textual database, this program operates with Google’s search engine, taking the whole of the Internet as a database and making searches of combinations of words that replicate Samuel Beckett's How It Is. HIIICT is thus a reconstruction of Beckett’s novel.

Beckett is one of the most protected authors in terms of copyright laws. What Cayley and Howe did was to transform a proprietary text into a non-proprietary text, erasing the figure of the author. As stated in the work’s last page,

Figure 2. John Cayley and Daniel C. Howe, How It Is In Common Tongues (2012).

Discretized in blocks of words, Beckett’s text was entirely cited from the Internet and all the links for the sources of each group of words are available as footnotes. We might say that as we read HIIICT we are reading Beckett’s How It Is but, as John Cayley argues,

HIIICT explores the practice of writing on and with the web, and understands the web as a tissue (or textum) composed of many automatic reading and writing processes. When a given piece of language (or even a simple utterance expressed in the gesture of clicking a button) is indexed by Google, it becomes data, meta-data, information, and value. And all this work of subjection to the informational regime of cybernetic administration, where technique and knowledge are monitored and monetized, all this work is made by algorithms operating in a scale and frame that is alien to that of human conscious perception. More and more aspects of human culture are being mediated by a complex network of algorithmic cognizers designed to operate on and to maintain a certain grid of economic positions and practices. The machines of the industrial age evolved to become abstract machines operating on hardware, on logistics and infrastructures. Human knowledge generates nonconscious cognizers that extract value from human knowledge, in a cyclic movement connecting separate cognitive systems, combining them in a complex network of human and algorithmic agents performing under a set of unilaterally and non-explicit established rules. Given such a context, HIIICT engages in exposing and resisting the ways in which “cognitive” capitalism captures, regulates and extracts value from the shared field of cultural production, including our common uses of language. HIIICT thus resembles the practices of the Luddites, textile workers who, in the nineteenth century, and faced with the threat of the effects of mechanization and the consequent threat of obsolescence of their work, sabotaged industrial machines in an attitude that, far from the specter of techno-phobia, aimed at resisting the naturalization of the exploitation of the human as a variable (in Flusser’s terms) in the human-machine binomial. As Cayley and Howe explain,

As a re-writing of Beckett’s text, HIIICT subverts the questions associated with authorship, copyright and property that characterize print culture and that have also been adopted by monopolistic tech corporations, ending the illusions of the peer-to-peer global network imagined during the first years of the so-called digital revolution. And it does so in two different and combined ways: by rewriting a book that is heavily protected by copyright laws, and by programming a nonconscious cognizer that uses one of the biggest tools of one of the largest Big Data corporations, ignoring its unilaterally imposed terms of use.

At the same time, the maneuver of appropriation (and liberation) of a copyrighted text is parallel (and opposite) to the maneuver of appropriation adopted by Google in what concerns the “commons of language.” The artistic value of this work resides in this double-fold operation of regeneration and reclaiming: taking a digital tool that embodies the contradictions of digital culture, and turning it upside down, to make it work as a tool for liberating a proprietary text.

The regime of quantification and datafication of language and life is socially situated and serves old dynamics of privatization of the commons, today by means of the circumscription of human activity in a cybernetic system. Language reacts when poets speak, and HIIICT is a speech act that creates meaning in the algorithmic landscape of the network by creating, within it, a literary context. HIIICT is engaged in reflecting on the tensions between human and machinic cognition, between human and posthuman language, between language as an instrument for meaning production and expression, and the algorithmic language that works behind the scenes of our writing, categorizing, indexing and monetizing it. Finally, and to return to the question of meaning production in nonconscious cognizers, we may argue that the literary value and the meaning of both How It Is and HIIICT resides only in human reading and interpretation. To the nonconscious cognizers that generated this work none of the two texts exist, but only sets of patterns and links corresponding to binary sets of information. The text, understood as a tissue of meaningful linguistic utterances, is thus incomputable.

This article is part of my PhD research in the Program in Materialities of Literature. Funded by the FCT — Foundation for Science and Technology (PhD Fellowship reference: PD/BD/52247/2013).

CAYLEY, John (2011). “Writing to be Found and Writing Readers.” Digital Humanities Quarterly 5.3. 20 May 2017. http://www.digitalhumanities.org/dhq/vol/5/3/000104/000104.html
CAYLEY, John, and Daniel C. Howe (2012). How It Is In Common Tongues. 30 May 2017. http://thereadersproject.org/?hiiict2012
–––––––––– (2013). “Reading, Writing, Resisting: Literary Appropriation in the Readers Project.” Proceedings of the 19th International Symposium on Electronic Art, ISEA2013, Sydney. 22 May 2017. http://ses.library.usyd.edu.au/handle/2123/9708
HAYLES, N. Katherine (2014). “Cognition Everywhere, The Rise of the Cognitive Nonconscious.” New Literary History, 45.2: 199-220. 20 May 2017.
PASQUINELLI, Matteo (2014). “The Labour of Abstraction: Seven Transitional Theses on Marxism and Accelerationism.” Fillip magazine #19. 27 May 2017. http://matteopasquinelli.com/labour-of-abstraction-theses/

[1] interpretation (n.) mid-14c. “a translated text, a translation” (late 13c. in Anglo-French), from Old French interpretacion, entrepretatiun “explanation, translation” (12c.) and directly from Latin interpretationem (nominative interpretatio) “explanation, exposition,” noun of action from past participle stem of interpretari “explain, expound; understand” (see interpret). From late 14c. as “act or process of explaining or interpreting; an explanation; construction placed upon an action.” Meaning “dramatic or musical representation” is from 1880. https://www.etymonline.com/word/interpretation

[2] mean (v.1) “intend, have in mind,” Old English mænan “to mean, intend, signify; tell, say; complain, lament,” from West Germanic *mainijan (source also of Old Frisian mena “to signify,” Old Saxon menian “to intend, signify, make known,” Dutch menen, German meinen “think, suppose, be of the opinion”), from PIE *meino- “opinion, intent” (source also of Old Church Slavonic meniti “to think, have an opinion,” Old Irish mian “wish, desire,” Welsh mwyn “enjoyment”), perhaps from root *men- (1) “to think.” Conversational question you know what I mean? attested by 1834. https://www.etymonline.com/word/mean

[3] See minute 23 of the video available at: https://www.youtube.com/watch?v=4p0bXPdZoAA.