Algorithm, Thought, and the Humanities:A Review of Unthought: The Power of the Cognitive Nonconscious

N. Katherine Hayles's Unthought: The Power of the Cognitive Nonconscious is an important book that addresses the increasingly urgent issue of the relationship between human consciousness and all of the technical systems that act on our behalf. In defining the concept of the cognitive nonconscious and contrasting it both to consciousness and thinking more generally, Hayles provides a flexible and provocative framework that should inform debates on these matters for years to come.

This book is lively and wide ranging. Although many readers will be familiar with some of the topics addressed, few will have experience with all of them. Hayles's impressive ability to connect nonconscious cognition to a wide range of technical, financial, and literary issues today is invaluable. Hayles is after more than this, however: she has in mind a broader case about the nature and value of the humanities today.

1.

Hayles begins Unthought by severing the common link between consciousness and thinking. While thinking involves high-level reasoning, often using language, cognition, "by contrast, is a much broader faculty present to some degree in all biological life-forms and many technical systems" (14). She offers this definition: "Cognition is a process that interprets information within contexts that connect it with meaning" (22). In particular, she understands interpretation as connected to choice: "Choice here, of course, does not imply 'free will' but rather programmatic decisions among alternative courses of action, much as a tree moving its leaves to maximize sunlight does not imply free will but rather the implementation of behaviors programmed into the genetic code" (25).

This definition of cognition as related to choice allows Hayles to distinguish between what she calls cognizers and noncognizers (30): "On the one side are humans and all other biological life forms, as well as many technical systems; on the other, material processes and inanimate objects" (30). As she explains, "The crucial distinguishing characteristics of cognition that separate it from these underlying processes are choice and decision, and thus possibilities for interpretation and meaning. A glacier, for example, cannot choose whether to slide into a shady valley as opposed to a sunny plain" (28).

In defining cognition as a process much larger than thinking and consciousness, Hayles draws attention to the relatively small role that consciousness plays in our actions. She notes that recent research shows that "nonconscious cognition operates at the level of a neuronal processing inaccessible to the modes of awareness" for a simple reason: in order "to keep consciousness, with its slow uptake and limited processing ability, from being overwhelmed with the floods of interior and exterior information streaming into the brain every millisecond" (10). Consciousness and other modes of awareness represent, then, a very small portion of our cognitive activity, those that are the least time-sensitive.

Of course, Hayles's attempt to reconfigure and refine our language for speaking about cognition is most urgent because of how technical systems increasing take on the role of processing information for us and acting on that information – of cognizing for us. The most familiar and clear example of this fusion of human and technical cognition is the drone used in combat, which can be piloted by a human operator, but which performs a whole range of cognitive activities and acts on its own in a variety of ways – such as controlling its various fan to keep itself level. The possibility of increasingly autonomous drones that can fly missions on their own (133) represents merely an extreme version of the "cognitive assemblage" of the piloted drone. Such cognitive assemblages can include not just technical systems, but others like "tactical commanders, lawyers, and presidential staff, forming assemblages in which technological actors perform constitutive and transformative roles along with humans" (37).

Although the drone is a particularly obvious case of such nonconscious cognition, Hayles makes clear how commonplace these kinds of cognitive assemblages are in contemporary life. She dedicates an insightful chapter to high-frequency trading (HFT), the system by which algorithms trade financial instruments at speeds impossible for humans: "Within the space of this 'punctuated agency,' algorithms draw inferences, analyze contexts, and make decisions in milliseconds," creating a "temporal gap between human and technical cognition" (142). She notes the challenges of intervening into a "complex ecology that [...] can be understood as swarm behavior" (163), and dismisses the idea that the problems produced by HFT could be solved by regulation, given that "finance capital is so deeply enmeshed with the self-organizing ecology of ultrafast machine algorithms" (169). This strikes me as unnecessarily pessimistic – after all, both Bernie Sanders and Hillary Clinton proposed a tax on HFT in 2016 that would have had the effect of reducing such activities – but the chapter makes clear how pervasive these noncognitive systems have become.

Although in general I appreciate Hayles's wide range of examples, at times her tendency to look forward to the technical future (as in the case of fully autonomous drones) can make her analysis unnecessarily tentative. For example, she devotes a couple of pages to the topic of digital voice assistants. But instead of discussing the implementation of current voice technology like Amazon's Alexa or Apple's Siri, Hayles focuses on a new, undeployed voice system called VIV, which has received a lot of enthusiastic attention in the technology press. "The program," she explains, "combines GPS orientation with an open system that programs on the fly, parses sentences, and links to third party sources" (124). VIV promises to be smarter than current technology, but the activities that Hayles is describing are used by all such systems already. Indeed, the contrast between Amazon's extensible system that makes more precise syntactical demands on the user, and Apple's aspiration to have Siri respond to natural-language questions as much as possible (sometimes with jokes) strikes me as raising some interesting philosophical questions about how we engage with these kinds of assemblages that VIV's more hypothetical implementation does not.

Quibbles like these aside, Hayles's book will help to define the issues surrounding technical noncognitive systems for a wide range of readers, and almost everyone will find examples that are new to them.

2.

More controversial is Hayles's attempt to reconfigure the relationship between humanistic and scientific inquiry. Her chapter on "the new materialisms" finds some common cause with Deleuzian theories developed by Rosi Briadotti, Elizabeth Grosz, and Jussi Parikka. It is easy to see why these kinds of materialist theories are appealing as a topic to address in Unthought. It is clear Hayles has in mind a broad critique of this Deleuzian element of contemporary thought. She cites a passage from Grosz drawing on Deleuze's concept of "force," and concludes: "The eloquence of this passage notwithstanding, it remains extremely imprecise about the nature of 'force' and fails to distinguish between different kinds of forces, although these kinds of distinction have been extensively investigated in various scientific fields" (80).

This critique reveals one of Hayles's primary goals in this book: to make an argument for greater engagement in scientific facts even when doing philosophical work. It would have been revealing to see Hayles engage with Deleuze directly, rather than relying on the refraction of his work through these new materialist applications of his philosophy, because this indirect approach introduces some ambiguity. Is it Deleuze's theory itself that is imprecise, or does the fault lie in trying to apply his theory to contemporary scientific questions by these new materialists? This might seem like a trivial concern, but it goes to the question of how knowledge can move between science and the humanities. This relationship is probed more explicitly in another chapter when Hayles analyzes the debate about "how pervasive rationality is in human experience" between Hubert Dreyfus and John McDowell. Hayles zeroes in on what she conceives as the imprecision of their discussion of thought, in particular the invocation of "absorbed coping," such as when we open a door to enter a room without being conscious of some of our actions. She faults both philosophers for failing to recognize the distinction between conscious and nonconscious thought, an oversight she finds "remarkable," and then offers a broader argument that philosophers need to draw on experimental data:

Also remarkable is the confidence of both men that discourse and argument alone are sufficient to settle the dispute, a self-reinforcing circle of belief in McDowell's case that uses rational argument to elevate the importance of reason. Although both mention a wide range of other philosophers, from Aristotle to Heidegger along with many others, neither cites any experimental research that would bear on the questions they debate. (58).

Hayles isn't exactly wrong here. I suspect that most philosophers would feel that there is value in learning more about the variety and nuances of such "absorbed" activities that Hayles describes as nonconscious. But there also appears to be a bit of a disciplinary disconnect in her apparent surprise that philosophers are "entirely unaware" of some of these scientific ideas that can "settle" the debate. I suspect that most philosophers would respond to her surprise by saying that this is how the discipline of philosophy has been constructed: appealing to earlier formulations and exploring alternatives through critique and thought experiments.

In other words, while I applaud Hayles's advocacy for interdisciplinarity and support her call for humanities scholars to take fuller account of this kind of technical information and research, I think that she under-theorizes the complexity of this kind of cross-disciplinary movement. Hayles tends to treat the information produced by scientific research as a set of facts (or at least formulations) that humanists need to take into account, rather than the operation of one discipline with its own rules that might be imported into another with a different set of rules.1 I have in mind the model that Mieke Bal offers in her 2002 book, Traveling Concepts in the Humanities, which analyzes how concepts like performativity, intention, and tradition move between disciplines. She contrasts the cultural analysis that she is describing to anthropology:

At first sight, the object is simpler than anthropology's: a text, a piece of music, a film, a painting. But, after returning from your travels, the object constructed turns out to no longer be the "thing" that so fascinated you when you chose it. It has become a living creature, embedded in all the questions and considerations that the mud of your travel spattered onto it, and surround it like a "field." (4)

I think that this characterization is relevant to Hayles's argument that philosophers need to be more informed about scientific research on concepts like cognition or force. Bal emphasizes that "traveling" from one discipline to another is work that changes its object of study. A few pages later, Bal describes "[c]oncepts not so much as firmly established univocal terms but as dynamic in themselves. While groping to define, provisionally and partly, what a particular concept may mean, we gain insight into what it can do" (11). In other words, as we think about how a term like force can move between physics and philosophy, we will need to take into consideration (among other things) what the concept is designed to accomplish in each discipline.

3.

As her comments on the Dreyfus/McDowell debate suggests, Hayles's ultimate goal in this book is not merely to offer an account of nonconscious thought. Instead, she sees this emerging issue as an opportunity to rethink the vocation of the humanities today more generally. In her concluding chapter, she addresses the "scope of interventions that can make differences in real-world systems" by appealing to humanities scholars:

For these utopian possibilities to be realized, humanities scholars must recognize that they too are stakeholders in the evolution of cognitive assemblages, which implies an openness toward learning more about the computational media at the heart of cognitive technical systems. At present, the digital humanities are the contested sites where these issues are discussed and debated, sometimes in heated and angry exchanges. (205)

This certainly seems compelling. Who can argue that humanities scholars wouldn't benefit from greater technical knowledge, that would allow them to engage more directly in debates about drones and HFT?

But the story that Hayles tells about these "heated and angry exchanges" immediately after this passage reveals the implications of her attempt to fundamentally reorient the humanities. She describes a digital humanities project on a Danielewski novel. She addresses the objection raised by someone in the audience that "the computer algorithms we employed could deal only with 'dumb' questions, not interesting interpretive ones. I responded that the 'dumb' answers lead to interesting interpretive possibilities" to make sense of the patterns identified by this analysis. Hayles describes her interlocutor as rather obstinate (although well-intentioned) and draws from this the lesson that the humanities suffer from a disdain for "the emphasis on the sciences of finding answers to well-defined questions" (205). This passage reminds me of her surprise that Dreyfus and McDowell might not see the urgency of keeping up with research in cognitive science or that new materialists, influenced by Deleuze, don't address the varieties of force that scientists theorize. Hayles wants to advocate the results of scientific research, and attributes the resistance to a bias in the humanities.2

Given Hayles's opening story about how a "dumb" algorithmic analysis can produce results that demand interesting interpretation, it might initially seem surprising that she makes a point of critiquing the distinction between description and interpretation a few pages later, attributing it to "some in the humanities [who] may choose to ignore these questions and the possibilities they open" (208). She frames those who might resist the kind of algorithmic analysis as uninformed about the complexity of how science operates:

For those opposed to the digital humanities, the charge is often made that all algorithms can do is describe, a characterization that, in the prevailing value schema, automatically relegates them to a lower strata and therefore not the "real" or "important" humanities. The clear binary thus established between description and interpretation is open to objections on multiple counts. Science studies, for example, has long recognized that description is always theory laden, because every description assumes an interpretive framework determining what details are noticed, how they are arranged and narrated, and what interpretations account for them. (208-209)

Hayles's goal is to challenge the traditional association of the humanities with consciousness and interpretation focused on social relationships: this understanding "reinforces the idea that humans are special, that they are the source of almost all cognition on the planet, and that human viewpoints therefore count the most in determining what the world means" (213). If, instead, we embrace the idea that the vast majority of cognition is nonconscious, then algorithms need not be seen as some kind of "other" to the humanities.

This is a challenging idea, and one that I think is perhaps the most important and far-reaching in Unthought. As she remarks, this "shift in conceptual frameworks [is] so extensive that it might as well be called an epistemic break" (214). I have to admit that I find this concluding note both very intriguing and yet frustratingly abstract. She remarks, "If, on the contrary, interpretation is understood as pervasive in natural and built environments, the humanities can make important contributions to such fields as architecture, electrical and mechanical engineering, computer science, industrial design, and many others" (213-214). Hayles praises the methods that the humanities have developed "for analyzing different kinds of interpretations and their ecological relationships with each other" but doesn't provide an example of these methods at work in any of these fields. All of her examples in this concluding chapter are digital humanities projects focused on literary texts.

I balk at the other component of Hayles's challenge to the humanities as well: the idea that moving forward with her new vision of the humanities involves embracing the goal of finding answers. Those in the traditional humanities, she notes, "tend to believe that interesting questions do not have definitive answers at all, offering instead endless opportunities for exploring problematics" (205). I have to admit that this does, indeed, sound like the core identity of the humanities to me, and I can't quite envision the alternative that Hayles is offering. The examples that she provides of answers seem straightforward enough: a digital humanities analysis that shows that gothic literary texts have a lot of definite articles in their titles, which in turn leads to a next round of analysis of place names in titles, and so on (210). I suppose that these algorithmically-produced findings are facts and that they represent a certain kind of answer to a question. I can also see how to some literary critics this kind of algorithmic analysis might initially seem foreign. But there's a huge gap between this relatively small addition to the methodological toolbox of the literary critic and the "epistemic" change that would have humanists giving up their commitment to the idea that big questions can always be problematized and reinterpreted.

At the end of the day, I think there are two urges in Hayles's book regarding the role of scientific knowledge in humanities research that aren't quite resolved. One the one hand, Hayles wants to advocate for the sciences in a more traditional way as offering empirical facts and ways to "settle" debates that philosophers might pursue endlessly. As she remarks in an early chapter, "in my view a position that can claim empirical support is preferable to one that cannot; otherwise, [...] it is impossible to distinguish between what is actually the case and what is ideologically driven fantasy" (79). She sees the resistance to these scientific findings as simply a matter of disciplinary siloing or a suspicion towards science in general, and wants humanities scholars to engage with a wider range of disciplinary knowledge. Although I think that she needs a more nuanced understanding of the way that scientific knowledge can "travel" to philosophy (and vice versa), encouraging humanists to be more interdisciplinary in their reading is a laudable goal.

On the other hand, Hayles recognizes that algorithms can only produce interesting results as long as we understand the need to investigate the terms and structures that produce that analysis recursively. Here she seems to be suggesting that research is open to potentially unlimited reinterpretation since every study is "theory laden, because every description assumes an interpretive framework." I find this advocacy for algorithms as an open-ended tool for interpretation to be compelling, but it seems at odds with her respect for scientific consensus on other topics like the nature of cognition and varieties of force, and her desire to settle questions.

4.

Unthought is an important book. It introduces an intriguing account of consciousness and cognition that has wide application to many fields in technology, finance, and culture. More importantly, Hayles challenges her readers to reflect on their assumptions about the relationship between science and the humanities. Although I think that there are some tensions in her formulation of this relation, this is a profoundly important question today, and I am sure that Unthought will help to prompt further debate and analysis.

Works Cited

Bal, Mieke. Traveling Concepts in the Humanities: A Rough Guide. Toronto: University of Toronto Press, 2002.

Hayles, N. Katherine. Unthought: The Power of the Cognitive Nonconscious. Chicago: University of Chicago Press, 2017.

Latour, Bruno. Science in Action: How to Follow Scientists and Engineers Through Society. Cambridge, MA: Harvard University Press, 1987.

Cite this Essay: