The Intelligence Advanced Research Projects Activity (IARPA) – established in 2006 in the spirit of the Pentagon’s DARPA to sponsor research for groundbreaking technologies to support an “overwhelming intelligence advantage over future adversaries” – is a little-known US agency that social and behavioral scientists (especially sociocultural anthropologists) should pay more attention to. This is because IARPA is notably social scientific in orientation and has been developing concepts in specific ways for use by the intelligence community (IC) that US anthropology in particular is significantly historically responsible for introducing to the social sciences, if in different ways, most obviously: culture, its coherence and the extent of cultural consensus, its relationship to society and to human agency.
At its inception IARPA was tasked with developing better ways, in USA Today-speak, to “help analysts measure cultural habits of another society.” And its portfolio continues to sponsor research intended to develop big data-type tools to process the linguistic and cultural information of countries, societies and communities of interest to US espionage. While there are anthropologists who work along the frontier between their discipline and the rapidly emerging computational social sciences, it is unlikely that many anthropologists would approach cultural analysis in the terms currently pursued by IARPA. The agency’s formulations of cultural problems likely strike most social scientists as well outside of, or as at odds with, the standard or prevailing disciplinary usages of this concept, including the concept’s basic significance and what legitimately can be done with it. But, it is often the case – and unfortunately so – that there is scant traffic of any kind between academic anthropology and the IC, even when there are clearly things to talk about, like the culture concept.
A Public Anthropology of the IC?
In the era of Wikileaks and Edward Snowden, journalists have increasingly sought to shine a light on “top secret America,” to borrow Dana Priest’s phrase. And public debate has in large part focused on the new circumstances of privacy (or the lack thereof), clandestine data collection, and the ethics of new largely internet-based and social media-derived means used by intelligence agencies to amass colossal troves of information while mining people’s online signatures. Much less often considered, if at all, is whether the sociological or anthropological theory – the tissue of ideas and concepts underwriting these programs – actually makes any sense.
Instead, the vast majority of attention is given to extolling and further exploring the possibilities for data collection opened up by new computational and social media technologies. Too often, wide-ranging and critically grounded academic discussion and debate has played virtually no part in how these programs are conceived and implemented. A lack of more substantive dialogue about the social science informing IARPA’s programs, and the possibility of skewed or flawed results built upon misguided or unexamined assumptions, is a serious problem with the potential to negatively and mischievously – but perhaps not altogether obviously – influence intelligence priorities in the US, and if indirectly, the country’s foreign policy footprint.
IARPA’s several culture-focused programs point to the need for more critical discussion among social scientists about IARPA-style social science and related priorities of security and intelligence agencies, a discussion which could at once address and more trenchantly appraise the particular assumptions (and social scientific world view) underwriting such projects, their limits, and the ways that questionable or debatable concepts and practices with traction in the social science of the securityscape are potentially relevant, defensible, or ill-conceived. This is a conversation that should also include the IC itself. But this is not a conversation that social scientists outside the IC either are regularly aware of, want, or perhaps even know how to have, with a few exceptions.
When academic social scientists do address the social science of the securityscape, the prevailing approach is to take issue with the politics and ethics of social scientific involvement with the present version of the military industrial complex, advanced from a position well outside this work and often at a considerable distance from the specific details – and many of the implications – of it. But we also need more grounded and zoomed-in discussion about the epistemologies, research designs, data, analysis, and conclusions drawn by this work, and associated implications, which take account of the ways this realm of social scientific ideas and concepts also drives IC priorities and outcomes in ways sometimes constructive but perhaps at least as often, problematic. If such discussions sometimes do take place, they need to be broader, deeper, more inclusive, and sustained.
A program to measure cultural habits suggests a quantitative approach to a hermeneutical problem, which, at the very least, takes for granted a very different conception of culture as a source of insight than the various ways that anthropologists usually engage with this concept. These differences are not trivial. “Culture” is a concept from which US anthropology notably retreated in the 1990s and which the discipline has continued to qualify in multiple ways, while for the IC interest in “socio-cultural factors” – often as cultural intelligence and as enlisted in exercises of prediction – has been notable since the mid-2000s, if to various ends. The reasons why the IC and academic anthropology appear headed in opposite directions vis-à-vis the culture concept would certainly be a timely discussion.
Metaphor for the IC
Several of IARPA’s programs have attracted at least some journalistic attention of late as well, such as 2011’s Open Source Indicators program. But here I consider instead IARPA’s Metaphor Program, also launched in 2011, because it is a particularly revealing example of the recent technologically-enhanced version of the cultural turn by the US intelligence community. Most simply, a “metaphor” is a linguistic relationship of similarity, where one experiential domain (the target) is understood by way of reference to another (the source). Astronomer Fred Hoyle coining the term “big bang” to refer to one theory for the origin of the universe is a case in point. IARPA’s program aspires to provide decision-makers with a more systematic understanding of the “shared concepts and worldviews of members of other cultures” by compiling a given culture’s metaphors and making these available to intelligence analysts.
My questions about this objective are several, if connected: As part of a larger IC project for culture, what does “metaphor” currently mean to the US intelligence community? And, given what “metaphor” means for the IC, how does this understanding influence the ways the IC might conceptualize specific cultures or foreign publics of interest? And what, in turn, might this mean for the footprint of the US intelligence community, as it offers policy decision-makers a particular account of global geopolitics, at least in part informed – if indirectly and in ways most likely invisible to any given decision-maker – by programs like this one?
IARPA’s solicitation for its Metaphor Program promotes the goal of a better understanding of “the tacit backdrop against which members of a culture interact and behave,” or the patterned “cultural norms” which compose the “worldviews of particular groups or individuals.” And metaphors are the program’s choice because they are both “pervasive in everyday language” and, IARPA assumes, metaphors “shape how people think about complex topics.” More importantly, IARPA understands metaphors to “reduce the complexity of meaning” because their usage is patterned.
As the program’s manager, Heather McCallum-Bayliss, observed, “Culture is a set of values, attitudes, knowledge and patterned behaviors shared by a group.” IARPA’s conception of metaphor is assumed to be a key to understand cultures, in large part because cultures, in turn, are understood – channeling the ghost of Ruth Benedict – as patterned and shared group behavior. Such a preference for disciplinarily obsolete but hyper-coherent conceptions of culture like Benedict’s in the broader military and security environment is far from unique. And a consistent preference for such starting points is telling about IARPA’s objectives and the computational steps it intends to take to achieve them.
IARPA is investing in research on metaphors because it is convinced such research has the potential to uncover the “inferred meanings,” “conventional understandings” and “underlying concepts that people share,” thus allowing the intelligence community to gain better analytic purchase on identified “cultures of interest,” but more importantly for the agency, on the “decision-making and perception of foreign actors.” To this end, IARPA’s approach to metaphor is largely derived from one influential story about metaphor most closely related to the species of cognitive linguistics associated with George Lakoff and colleagues. And Lakoff, it turns out, is not coincidentally, a member of a research team now working to develop a multilingual metaphor repository with IARPA funding from its Metaphor Program.
Lakoff’s Tristes Tropes
So, first we need to know a few things about the Lakovian approach to metaphor, since there are other contenders in the scholarly field of trope theory. If beginning with his influential Metaphors We Live By, co-authored with Mark Johnson in 1980, in recent years Lakoff has also established a reputation as a public intellectual of sorts, applying his metaphor-heavy analytic hand to the US political landscape. In his most recent incarnation, Lakoff has used his approach to metaphor to support the progressive cause, and has often presented his work in the form of guides, handbooks and toolkits instead of as research. But, while Lakoff’s heart might lie with progressives, his conception of metaphor is deeply conservative, as I make the case below. And this has direct consequences for IARPA’s program, taking for granted as it does the Lakovian world view on metaphor.
Here’s Lakoff’s take on metaphor, in a nutshell. As he explains, conceptual metaphors – which typically employ a more abstract concept (e. g. politics) as a target and a more concrete topic (e. g. family) as a source – shape the ways we think and act, and underwrite a system of related metaphorical expressions that appear more directly on the surface of our language use, which Lakoff calls linguistic metaphors. If much more can be said about this, here’s the rub so far as IARPA is concerned: conceptual metaphors are the key for understanding how speakers – typically, members of the same “culture” – systematically map relationships between conceptual domains. Mapping in the Lakovian mode refers to the patterned set of correspondences that exist between source and target domains.
While there is good reason to assume that the map is not the same as the territory, one imagines that IARPA sees potential in such a mapping exercise because maps promise empirical predictability. Another probable attraction is that Lakoff’s work on metaphor has, in recent years, become increasingly slanted toward neuroscience. He now describes “neural metaphorical mappings,” where metaphors are “fixed in the brain” along “pathways ready for metaphor circuitry.” Lakoff’s marrying of cognitive linguistics to neuroscience has transformed a woolly term from the humanities – metaphor – into a building block for a new “neural theory of metaphor,” now presented as a scientific tropology, in ways conversant with a growing obsession across US military and security agencies with the potential of neuroscience.
Machines Learning Metaphors
IARPA’s Metaphor Program is, essentially, about combining emerging techniques and technologies in computational modeling with cognitive linguistic theories of metaphor like Lakoff’s. At the Proposer’s Day brief explaining its new metaphor program, IARPA described linguistic metaphors as “realizations of the underlying pattern or systematic association of abstract concepts” – a set of relationships IARPA assumes to be “defined by mapping principles.” IARPA would like to be able to data-mine online textual data on a large scale, as a “rich source for identifying cultural beliefs” about key societies of interest, and to develop new automated techniques to identify, map and then analyze the metaphorical language of entirely online native-language text. (I won’t take up here why online text – as a particular technological platform, set of expressive conventions, and kind of performance – is unlikely to be unproblematically representative of peoples’ cultural beliefs.)
What is critical for evaluating this project is making sense of the conviction that the relationship, for example, between a given metaphoric target and source (e. g. understanding “government corruption” as a “disease”) is conventional and predictably mappable; or that the development of unsupervised machine learning of such metaphor mappings is possible; and that this will then enable computational metaphor identification and categorization, as part of a “metaphor repository,” a database IARPA would build and maintain for a given language; against which analysts will eventually and ideally be able to compare “real-life statements” to predict intentions of people who may represent a threat to the US. (The agency has identified American English, Farsi, Russian and Mexican Spanish as initial languages of interest.)
For IARPA’s program to be successful, a basically Lakovian approach to metaphor has to be uncritically accepted as correct: linguistic metaphors, assumed to be representative and available in large numbers at the surface of online native-language texts, will be massively mined; their relationships of source to target, it is further taken for granted, will be able to be systematically reliably mapped; these analogical maps, goes the reasoning, will enable identification of more fundamental conceptual metaphors among cultures of interest; and this will allow analysts to infer relevant cultural patterns informing the behavior of foreign nationals; and perhaps even to help predict their likely decision-making on complex topics.
Lakoff on metaphor, in other words, has to be coded into the computational tools to be used to build the repositories before any such metaphors are even collected. And Lakovian-type metaphorical maps seem to be the extent of IARPA’s data-mining game. This is to say, the theoretical starting point and technological requirements of IARPA’s metaphor program are largely determinative of what “metaphor” can mean in this case. But, since a scholarly consensus about metaphor eludes us, and since one could choose to emphasize other features of the diverse work of metaphor, IARPA’s choices tell us perhaps more about its own world view than about anyone else.
Metaphor through the Looking Glass
Each metaphorical mapping in a given repository, we are told, will be validated using metrics designed to confirm “native-speaker knowledge of the metaphorical relations.” Such an idea works only if each language were a reliably monoglot standard, underwritten by conventional metaphoric associations recognized as such and in the same ways by any typical and competent native speaker. And so, each metaphor is at once culturally-specific – let’s set aside that languages and cultures are not the same – but also culturally entirely conventional. Yet the idea of native competence is an increasingly suspect one among linguists.
IARPA’s choices have consequences. As with its consistently topographical conception of culture, where patterned cultures can be organically decomposed into constituent and mappable relations of figure to ground, IARPA seemingly relies almost entirely upon the conventionality of metaphor. A consequence of its peculiar approach to metaphor and to culture as a limiting condition upon how people think, is that IARPA’s working conception of its notional publics – the people it is trying computational to figure out – is seriously limiting. IARPA is all in with a conception of metaphor, we might say, as stuck in the mode of mechanical solidarity, giving its attention to what are otherwise called “dead metaphors,” which, it can be argued, are in fact no longer really metaphors at all.
IARPA’s metaphor repositories would be cross-cultural collections of metaphoricized commonsense, that is, composed of already recognized and accepted metaphoric relations, informing the predictable parameters – maybe more accurately, limiting frames – of analogic reasoning of members of a given culture. This has the potential to be perversely conservative, since IARPA would understand decision-makers as drawing upon an identifiable cultural aggregate of figurative relationships which are always already assumed to exist. Such a situation makes of prediction, paraphrasing Yogi Berra, an exercise in déjà vu all over again.
Given Lakoff’s fashionable redressing of his approach to metaphor in the terms of neuroscience, and the ways a technologically-enhanced culture concept is being engineered by IARPA’s Metaphor Program as a difference engine keyed to cultural consensus, the conventional, and metaphoric persistence, it would not be hard to imagine analysts, as beneficiaries of this data and when considering how the people they study make decisions, adopting an analytic shorthand to refer to the “Russian brain” or “Farsi brain,” in ways reminiscent of a Cold War era fascination with American, Russian or German “modal personality types.” For many anthropologists, research scenarios like these are troubling because they raise a Levy-Bruhl-like specter of “how natives think,” troubling because also aggressively “othering.” A cynic might go even farther to suggest programs such as this one are developing technologies for “enemy-making.”
Metaphor’s Multiple Futures?
Ignored or sidelined in IARPA’s efforts are competing conceptions of metaphor. Ricoeur, to pick one, emphasized the ways that metaphors creatively transform language by revealing new ways to conceive of a referent. Metaphors generate and regenerate meaning. Black explored the open-endedness of metaphors, which he understood as too unstable to function referentially, but as introducing previously unavailable meanings in the dynamic interplay of figure and ground. Davidson remained unconvinced that metaphors could function as propositional at all, insisting instead that it was a mistake to assume metaphors possess any particular or stable “meaning.” These several conceptions of metaphor point to the limits of consensus around the conventionality of metaphor and the ways that backward-looking exercises in mapping and archiving metaphoric relations can fail to anticipate the future.
To take a case in point: Genetics historically has been a field shot through with metaphors. Metaphors describing the work of genes are particularly ubiquitous, including: map, code, blueprint, and recipe, where DNA is understood to “write” the hereditary possibilities for our biological future. The biologist Richard Dawkins’s influential concept of the “selfish gene,” for example, promotes a gene-centric theory of evolution, where human beings are mere vehicles for successfully self-propagating individual genes, as the architects of natural selection. But the success of Dawkins’s selfish gene metaphor is beginning to obscure the changing meaning of “gene,” including a growing variety of technical usages.
Researchers now emphasize the idea of a “post-genomic” biology, where combinations of networks of less selfish and more managerial genes are also influential, where “writing” can be less important than “reading,” and the relation of heredity to the environment appears increasingly complex and dynamic. But there are as yet no convincing off-the-shelf metaphors to describe what we continue to learn about the behaviors of genes. In other words, even given the technical and highly shared vocabulary among evolutionary biologists, the shape-shifting of genes under scientific inspection eludes easy description. And whatever might follow the selfish gene story is still emergent as a set of metaphors that cannot be mapped without significant distortion.
If sharply divergent from IARPA’s starting point, what these several conceptions of metaphor share is an attention to the arguments at the center of culture, to the work of metaphor for social shape-shifting, and where identity is always in motion in relationship to – paraphrasing William James – the blooming buzz of experience. They attend to the translational and problem-solving work of metaphor, and to the ways metaphor might animate new inquiry. Conceived in such ways, metaphors do not so much express similarity but create new relations among “unlike things.”
Accounts like these foreground the properties of metaphor as extensive rather than conventional, and as emergent rather than underlying. Concerned as they are with the ways that metaphors, in the words of anthropologist James Fernandez, are strategic predications upon the inchoate – that is, predications upon frontiers of life and experience that elude our ready classification – these offer alternatives to the conception of metaphor currently being reinforced in the social science of national security. And these alternatives run devastatingly counter to any possibility for a predictive tropology of the near future.
This post originally appeared here: http://www.ethnography.com/2014/01/troping-the-enemy-culture-metaphor-programs-and-notional-publics-of-national-security/