Chapter 2

Philosophy of the Web: Representation, Enaction, Collective Intelligence

Harry Halpin, Andy Clark, and Michael Wheeler

Introduction

There is an emerging vision of the human mind as essentially a social organ apt to make extensive and transformative use of whatever forms of local and global scaffolding other agents and technologies provide. In an increasingly wired and networked world, our very nature as cognitive beings is gradually changing.

The World Wide Web is a remarkable triumph of incremental com­putational engineering. In the wake of its technological success and the structural change it has effected on human social organization, Web designers and researchers are being forced to confront a range of foundational issues with clear philosophical dimensions. These include old philosophical issues in modern guises—issues concerning knowledge, identity, and trust, as well as new questions raised by the increasingly complex ways in which the Web is embedded in the larger world. Such new questions concern, for example, the character and status of Web objects, such as websites and mash-ups, the understanding of authorship within new collaborative and collective creative ensembles, and the relation between, on the one hand, the structure and functioning of the Web and, on the other, the strengths and weaknesses of basic biological cognition.

1. Is Philosophy Part of Web Science?

Sir Tim Berners-Lee, the inventor of the Web, has challenged philosophy to contribute to the future construction of the Web, calling the architects behind the Web “philosophical engineers” (Berners-Lee 2003). Our hypothesis is that profitable philosophical engagement with the Web will be achieved through the lens of contemporary debates in philosophy of mind and cognitive science over what is sometimes called 4E (embodied, embedded, enactive, extended) cognition. Indeed, it is this area of philosophy that has attracted the most interest from leading thinkers about the Web, although to date there are very few examples of philosophers of 4E cognition engaging directly with the Web. As we explain below, a number of critical and fundamental questions that Web researchers are beginning to confront are intertwined with issues at the forefront of recent work on 4E cognition. Importantly, however, the intellectual traffic here is not one-way. A carefully crafted philosophy of the Web will not merely draw on the aforementioned philosophical debates, it will make important contributions to them. The study of human–Web couplings provides a powerful way to pursue several unresolved and controversial contemporary philosophical issues. In this way, practical interests in the design and use of Web-based technologies dovetail with foundational questions concerning the nature of minds and persons.

This new philosophy of the Web should go beyond much “new media studies” work on the Web by engaging directly with certain pressing scientific and engineering concerns faced by Web architects. What was missing from this earlier work was the productive engagement with the relevant engineering community that the present project will foster. Much of the fault of this lies on the shoulders of philosophy, however (Dreyfus 2001). The debate over relevance sensitivity to be found in the 4E philosophical literature has seldom been connected to the performance of search engines or the Web more generally (Wheeler 2005). Similar comments might be made about enactivist thinking. By contrast, the extended mind hypothesis is often accepted as almost intuitively obvious in Web circles, even though in philosophical circles it is the subject of ongoing critical debate (Adams and Aizawa 2001). The established interest in the extended mind by the Web community is indicated by, for example, work of computer scientists and psychologists on applying the extended mind hypothesis to the Web (Smart et al. 2010). Moreover, the extended mind hypothesis was advanced as a possible foundation for a philosophy of the Web in a 2008 exchange between Halpin and Wheeler in the American Philosophical Association Newsletter on Philosophy and Computers (Halpin 2008; Wheeler 2008). Building upon this previous work, we can outline three linked themes that a philosophy of the Web based on empirical work and cognitive science should address: the problem of representations on the Web, enactive search, and collective intelligence.

2. Representations and the Web

Cognitive science and artificial intelligence (AI) have traditionally appealed to internal and neutrally realised representations to explain intelligence. But exactly how the notion of representation should be understood, and the extent to which neurally located representations are necessary for intelligence, have long been vexed issues, especially in the philosophical literature. The Web provides a new impetus for investigation into the notion of representation, because the advent of the Web has seen an explosion of novel external representations (e.g., hypertext web pages) that, through complex, iterated interactions with human users, enable intelligent information retrieval, complex commercial activity, and social communication and coordination. We must explore the relationship between these two seemingly different representational contributions to intelligent activity, asking whether there is a single account of representation that applies to both of them.

These are theoretical questions with practical implications. The Semantic Web (also known as Linked Data) is a project launched by Berners-Lee and deployed by august bodies, including the U.K. government (e.g., in the recent initiative to release government data) and the BBC (Berners-Lee, Hendler, and Lassila 2001). Its goal is to build large-scale structured knowledge representation systems using the Web. Accordingly, it has been identified by some computer scientists and philosophers as an attempt to restart the project of classical AI (Floridi 2009). If this is right, the Semantic Web will ultimately face the problems that, according to some critiques, plagued its intellectual predecessor (Clark 1997). One such problem is the Frame Problem, the recalcitrant difficulty of determining, in a wholly mechanistic manner, which items of information from a huge memory store are relevant (and which are irrelevant), and how retrieved items of information should be updated, within and across changing contexts of activity (McCarthy and Hayes 1969). Influenced by phenomenological philosophers such as Heidegger (who stressed embeddedness) and Merleau-Ponty (who in addition stressed embodiment), thinkers such as Dreyfus have argued that the root cause of such problems is the assumption that the mechanistic processes at work are representation guided (Dreyfus 1979). Perhaps anti-representationalist-embodied cognition has itself been taken too far? Alternative remedies depend on a reconceived, Web-friendly notion of representation, and identify the implications for how the Web may meet the challenge of relevance sensitivity.

3. Enactive Search

Currently, access to representations on the Web is mediated through search engines such as Google. One of the keys to the practical success of such search engines is that they use massive amounts of statistics gathered from user actions and user choices on the Web to dynamically adapt their algorithms to find appropriate content, and thus to grapple with the issues of relevance and context highlighted by the frame problem. The more information a search algorithm has about the conduct and interests of users, the better its adaptation, a trend accelerated by “Web 2.0” technologies, such as social networking and collaborative tagging.

We hypothesize that the complex adaptive dynamics of such statistics-driven user-action-based searches may be illuminated using the philosophical concept of enaction (roughly, the idea of “laying down a path in walking,” that is, of actions that change the world in ways that feed and structure those very actions, either now or at some future time [Varela, Thompson, and Rosch 1993]). Although much of the existing work on enaction has focused on the biological individual, an enactive paradigm can also be applied to the collective effects of our use and navigation of the Web. From this theoretical standpoint, we must investigate how search engines highlight a number of key issues in the way we think about minds, persons, and collective endeavours. These range from a kind of “quantification” of Wittgenstein's maxim “meaning is use” as a form of statistical language processing, an intellectual legacy dating back to the early search engines produced in Cambridge in the 1960s, to the empowering and disturbing (in about equal measure) vision of the not-too-distant future described by Google's CEO Eric Schmidt, in which Google is connected “straight to your brain” (Wilks 2005).1

4. Cognitive Extension and Collective Intelligence

According to the extended mind hypothesis as presented by Clark and Chalmers (1998), cognitive processes are not always confined within the boundaries of skin and skull. Under certain conditions in which biological brains and bio-external scaffoldings work together as integrated processing ensembles, cognition may extend into the world. The nature of the Web opens up this controversial philosophical issue in a distinctive and distinctively problematic way. Perhaps external representations on the Web, when integrated appropriately into the processes that govern an agent's behaviour, may count as parts of that agent's cognitive architecture. But now assume that multiple...