by PETER SKAGESTAD
Linguistic Systems, Inc. and the University of Massachusetts - Lowell. July 1993*
Published in The Journal of Social and Evolutionary Systems, vol. 16, no. 2 (1993), pp. 157-180.
Philosophical interest in the computer revolution has to date largely focused on the as-yet unfulfilled promise of artificial intelligence, and the debate of the past three decades has canvassed the two tightly intertwined questions of whether a thinking machine is possible, and whether the concept of such a machine provides an appropriate model of the human mind. While this debate has been going on, a different kind of computer revolution has taken place and is still taking place, to wit, the transformation of knowledge work through the ubiquity of the personal computer, and more specifically the networked personal computer. While philosophers, like other knowledge workers, have grown to depend on the personal computer and to take it for granted as a tool of their trade, the personal-computer revolution has attracted only negligible attention from philosophers. Part of the thesis of the present essay is that the (still unfolding) personal-computer revolution is philosophically as consequential and as interesting as the (actual or potential) AI revolution. Also, as computers increasingly become ubiquitous as internal components of automobiles, stereo systems, and dishwashers, discussion of how our lives are affected by "the computer", understood as any physical embodiment of a finite Turing machine, rather than by particular types of computers, understood in terms of their functionality, makes less and less sense. At any rate, as we shall see, the pioneers of the personal-computer revolution did not theorize about the essence of the computer, but focused rather on the essence of human thinking, and then sought ways to adapt computers to the goal of improving human thinking.
This goal was understood ambitiously: the automation of intellectual housekeeping tasks was intended to bring about qualitative changes in our thinking, not just to enable us to do more of the same kind of thinking we had been doing all along. Like the American, the French, and the Russian revolutions, the personal-computer revolution has been the work of people who had very definite ideas about what they were doing and who were not shy about committing those ideas to paper. Just as the Bolsheviks were dead wrong in their understanding of the project they were embarked on - to wit, preserving the Russian Empire from the disintegration that seemed a certain consequence of World War One - so this new breed of revolutionaries may prove quite mistaken about the meaning of their revolution; yet we can no more understand this revolution in isolation from the self-understanding of the computer revolutionaries than we could hope to understand the Bolshevik Revolution without at least some acquaintance with Lenin's ideas.
Specifically, this essay examines the seminal writings of some of those who pioneered the personal-computer revolution, and documents an explicit conceptual framework, comprising certain definite ideas about how people think, how people adopt artifacts as tools for thought and how those artifacts in turn affect our thinking through feedback loops, and finally, how computer systems can be developed to extend or "augment" our intellectual capabilities. I argue that this explicit conceptual framework - which, following Douglas C. Engelbart, I refer to as "augmentationism" - rests on certain implicit philosophical assumptions regarding language, mind, and knowledge, assumptions that have at various times and in various ways been made explicit in the philosophical literature. We trace the presence of these assumptions in the evolutionary epistemology which has been formulated by Donald Campbell and Karl Popper, and which itself restates some of the central doctrines of Charles Peirce's semiotic.
Although this paper does not directly concern itself with the philosophical critique of artificial intelligence, it documents the existence, in the sources mentioned, of a third philosophical paradigm, different from both the computational paradigm and the intentionalist paradigm, which is the chief source of critiques of computationalism.
The term "augmentation" was adopted as a slogan by Douglas C. Engelbart in a report written in 1962. Over the next fifteen years Engelbart was in charge of the Augmentation Research Center (ARC) at the Stanford Research Institute (SRI), a project funded by ARPA's Information Processing Techniques Office (IPTO), which was headed initially by J.C.R. Licklider of MIT and subsequently by Ivan Sutherland, the inventor of bit-mapped graphics. For a detailed account of the pioneering work done under the direction of these three men, I can do no better than refer the reader to Howard Rheingold's fascinating account Tools For Thought, which reads like a novel.
Suffice it to list their chief practical results in (roughly) chronological order: interactive computing, time sharing, word processing, networking, and graphical user interfaces. ARC was dissolved in 1977; by then, dozens of Engelbart's collaborators were already at Xerox's Palo Alto Research Center (PARC), developing the first personal computer, the Alto, many of whose most revolutionary features - e.g. windows, pulldown menus, and mice - were later incorporated into the Apple Macintosh, under the direct management of Engelbart's erstwhile colleague Bob Belleville. Although regarding current technology as a literal embodiment of Engelbart's ideas would be an error, there is nonetheless an unmistakable causal chain from his ideas to today's technology.
In a 1960 paper Licklider inaugurated the augmentationist program in all but in name, by proposing the development of a temporary "man- computer symbiosis". Instead of directly challenging the artificial intelligence paradigm, Licklider simply granted that, ultimately, by 1980 at the earliest, the computer might outdo the human brain in most respects; meanwhile, however, most intellectual advances would still have to be made by people, who could benefit from computer systems specifically designed to extend the intellectual capabilities of humans. To demonstrate the need for such a symbiotic relationship, Licklider noted that he had at one point kept track of exactly how he spent his working hours, and reached the depressing conclusion that about 85 per cent of his time was spent getting ready to think, e.g, finding information, plotting graphs, or converting units of measurements to make different data sets comparable:
"Throughout the period I examined, in short, my "thinking" time was devoted mainly to activities that were essentially clerical or mechanical: searching, calculating, plotting, transforming, determining the logical or dynamic consequences of a set of assumptions or hypotheses, preparing the way for a decision or an insight. Moreover, my choices of what to attempt or not to attempt were determined to an embarrassingly great extent by considerations of clerical feasibility, not intellectual capability. "
Licklider went on to list the requirements of a human-computer symbiosis: greater computer speed, more memory, greater tolerance of human natural language, and new, graphically based user interfaces, such as were soon to be made possible through the efforts of Licklider's proteges Engelbart and Sutherland.
What a human-computer symbiosis might look like was fleshed out in great detail by Engelbart in his 1962 report, which in part draws on, and includes lengthy quotations from, Vannevar Bush's 1945 description of the "Memex", a notional workstation using dry photography rather than electronic storage media, and incorporating the earliest description of what later has become known as "hypertext", a term introduced by Theodore Nelson in 1965. In his article Bush addressed the problem - an aggravated one in the age of rapidly increasing information - of getting at data organized in counterintuitive ways, e.g. alphabetically, numerically, or in descending hierarchies of subclasses. The human mind, Bush observes, functions rather by association; hence information retrieval via trails of association is something that ought to be mechanized:
"When the user is building a trail, he names it, inserts the name in his code book, and taps it out on his keyboard. Before him are two items to be joined, projected onto adjacent viewing positions. ... The user taps a single key, and the items are permanently joined. ...Thereafter, at any time, when one of these items is in view, the other can be instantly recalled by tapping a button below the corresponding code space. Moreover, when numerous items have been thus joined together to form a trail, they can be reviewed in turn, rapidly or slowly, by deflecting a lever like that used for turning the pages of a book. It is exactly as though the physical items had been gathered together from widely separated sources and bound together to form a new book. It is more than this, for any item can be joined into numerous trails. "
Nonlinear thinking is popularly considered a discovery of the 1980's; amazingly, the above was written in the 1940's. Bush's essay is also striking in that, though he mentions that the Memex might be connected to a computer to provide calculations, the storage-and- retrieval system of the Memex is not itself computerized. In other words, both the concept of the personal workstation and that of hypertext were formulated quite independently of the concept of the digital computer, which was later to prove the enabling technology for their realization. But perhaps this ought not to surprise us. In the nineteen-sixties computers were generally thought of as data- processing machines. However, punched card data processing was invented by Herman Hollerith in 1890 - decades before the advent of the digital computer. The uses to which computers are put have always been independently conceived, never dictated by the potentialities specifically inherent in the digital computer. We note the irony, however, that Bush, himself a pioneer of analog computers, initially had no use for digital computers and actually opposed their early development.
Finally, in his 1962 report Engelbart formulated a detailed conceptual framework for developing new technologies to "augment" the human intellect. He also accepted the cyberneticist William Ross Ashby's term "intelligence amplification", but with a caveat. In his Design for a Brain (1952), Ashby had introduced the term "amplification" - apparently in the sense of leverage - specifically with the example of knowing how to use a dictionary as opposed to knowing the meanings of a number of words. But the word is used differently in his classic 1956 paper 'Design for an Intelligence-Amplifier'. Ashby here presented a possibility proof of a machine that would solve problems its creators were incapable of solving. This machine would amplify human intelligence the way human physical power is amplified by the steam engine, and unlike the way it was earlier amplified by the lever or cog and pulley. This analogy, along with its companion contrast, blurs the distinction between intelligence amplification and artificial intelligence. The steam engine, while human-made, is an independent source of power, unlike the lever which augments the human power exerted on it. (Of course human energy may be expended on shovelling coal into the steam engine, but that is not the only, or most important, energy which the steam engine transforms.) Without going into the details of Ashby's usage, Engelbart stresses (with Ashby) that his objective is not to increase our native intelligence, but to augment the human being with means for organizing experience and solving problems so that an intelligent system results in which the human being is the central component (possibly in contrast with Ashby, whose intelligence-amplifier was conceived of as a free-standing system).
The other components are called "augmentation means", of which Engelbart defines the four basic classes - artifacts, language, methodology, and training - and accordingly defines the intelligent system as a "Human using Language, Artifacts, Methodology, in which he is Trained", or, for short, the H-LAM/T system. Humans living in our contemporary society have already been significantly augmented by the language, artifacts, and methodologies provided by our culture, and Engelbart identifies three stages in the evolution of our intellectual capabilities: first, concept manipulation, by which he means the formation of mental abstractions "in their raw, unverbalized form"; second, symbol manipulation, by which Engelbart does not intend the (admittedly extremely important) ability to communicate through symbols, but simply the ability of the individual to represent objects to him or herself through symbols (e.g. the ability of a shepherd to keep track of his flock by remembering the cardinal number 27, instead of having to visualize 27 different sheep); and third, manual, external symbol manipulation, such as that made possible by "a stick and sand, pencil and paper and eraser, straight edge or compass, and so on". Again, the evolutionary significance Engelbart is concerned with is not the "value derived from human cooperation made possible by speech and writing," but the effect that supplementation of memory and powers of visualization with graphical representations of symbols has on an individual's thinking.
Taking note of Benjamin Lee Whorf's hypothesis that a culture's world view is constrained by the expressive repertoire of its language, Engelbart proceeds to formulate an analogous hypothesis (termed "the Neo-Whorfian Hypothesis") regarding the other augmentation means: "Both the language used by a culture, and the capability for effective intellectual activity, are directly affected during their evolution by the means by which individuals control the external manipulation of symbols." As a corollary of this hypothesis, Engelbart deduces that significant modifications in our language and our way of thinking could result from "integrating the capabilities of a digital computer into the intellectual activity of individual humans...", and which such modifications would be useful ones remains to be determined. This integration would constitute a fourth stage in the evolution of our intellectual capabilities, the stage of automated external symbol manipulation:
In this stage, symbols with which the human represents the concepts he is manipulating can be arranged before his eyes, moved, stored, recalled, operated upon according to extremely complex rules - all in very rapid response to a minimum amount of information supplied by the human, by means of special cooperative technological devices.
These devices, Engelbart goes on, could conceivably be provided by a highly interactive digital computer hooked up to a three-dimensional color display, as in the technology we today call "virtual reality" . Imagining all the new types of symbol manipulation that this technology would make possible was impossible in the pre-computer age (1962), just as, at the earlier stage of mental symbol manipulation - prior to the evolution of external symbol manipulation - explicit imagining of the bar graph or the process of long division would have been impossible.
The scope and magnitude of the intellect augmentation that is involved could nonetheless be illustrated by an exercise in de-augmentation. For this purpose Engelbart performed the simple experiment of tying a pencil to a brick and measuring the effect on writing speed; what he found was that a sentence which took 7 seconds to type and 20 seconds to write in cursive script, took 65 seconds to write in "de-augmented" cursive script - although the writing time could be reduced to 42 seconds by enlarging the script to the point where the sentence took up a whole page.
Now, if Engelbart had wanted to prove experimentally that a brick is not an ideal writing implement, then that of course was hardly a startling result. What was intended to provide a mental jolt here is rather the very idea of the extent to which our intellectual, social, and cultural life depends on the physical characteristics - literal weight and otherwise - of such tools as pencils. Indeed, that graphite has such a weight relatively to human muscular strength as to produce the vectors resulting in "normal" human writing speed is only a fortuitous fact - this could have been different and with profoundly different consequences:
"Brains of power equal to ours could have evolved in an environment where the combination of artifact materials and muscle strengths were so scaled that the neatest scribing tool (equivalent to a pencil) possible had a shape and mass as manageable as a brick would be to us - assuming that our muscles were not specially conditioned to deal with it. ... How would our civilization have matured if this had been the only manual means for us to use in graphical manipulation of symbols?"
In response to his own question Engelbart makes the two observations that (a) the record-keeping that underlies our organization of commerce and government would undoubtedly have taken a different form, and (b) far fewer people would have made the physical effort required to engage in reasoned argumentation or experiment with new concepts. This same point has recently been made by Henry Petroski in his fascinating history of the pencil; referring to the ancient use of lumps of charcoal as writing implements, Petroski writes: "Writing or drawing with a lump of anything for an extended period of time can cramp the fingers, thus cramping one's style and perhaps even cramping one's mind." Think about trying to write the Principia Mathematica with a lump - not a stick - of charcoal, and the dependence of abstract thought on the specific physical characteristics of the implements of intellectual work positively leaps out at you.
Three further comments are in order here. First, as has been noted by Michael Heim, Engelbart's observations strikingly parallel Eric Havelock's almost simultaneously appearing interpretation of Plato's philosophy as, in part, a cultural by-product of the emergence of literacy - Havelock's observation being that Plato's preference for abstractions over sense experience reflects a commitment to modes of thought that could not have been sustained within a predominantly oral culture. And second, although the brick-pencil has a certain dramatic force, there is actually no need to take recourse to make- believe to make Engelbart's point. We do, after all, know of cultures like that of the ancient Norse, whose only writing implements were chisels and stone tablets which must have made the writing of even the simplest sentence an extremely laborious and time-consuming task. (This point is also illustrated by Petroski's description of the ancient use of charcoal.) And we know that, among the ancient Norse, literacy was far from widespread, and that the runes were regarded as sacred symbols understood only by a few initiates, and not as tools for developing reasoned arguments or experimenting with abstract concepts. [Editor's Note: And see also Innis and McLuhan in the works cited in Note 23 for discussion of similar effects in the slow-writing culture of ancient Egypt. See also my Mind at Large: Knowing in the Technological Age, Greenwich, CT: JAI Press, 1988, pp. 131-132, "The Book in History" for consideration of writing systems and their impact on religions.] But third, Engelbart was not writing for the same audience as Havelock. He was writing specifically for the technocrats within the computer science community and, as he has later made clear, his objective was to distinguish his research project, on the one hand, from the speculative approach of artificial intelligence and, on the other hand, from the nuts-and-bolts issues of information retrieval technology. Borrowing C.P. Snow's dichotomy of the "two cultures", Havelock's book on Plato may be described as an event in a different culture from Engelbart's report, making the parallel all the more striking.
Having distinguished the above four stages of evolution, Engelbart fills in his conceptual framework with a detailed description of a hypertext system - one which builds on and significantly extends Bush's seminal ideas. But I shall not on this occasion pursue these ideas further. Let us instead conclude our overview of the augmentationist framework. In this framework, thinking is not a process that takes place inside the brain. Thinking is an activity in which the brain participates along with the eyes and the hands and a multitude of external devices, ranging in complexity from the pencil or the straight edge and compass to today's interactive computer systems. The computer, in turn, is not a neutral tool for the more efficient execution of intellectual projects antecedently conceived and adopted; rather, the computer is a medium which informs our choices of which intellectual projects to adopt, and which thereby transforms our conception of what we are doing and perhaps of who we are.
We have noted Engelbart's acknowledged indebtedness to Ashby, Bush, and Whorf. I can find no trace of any explicitly philosophical influences on Engelbart, nor, as far as I know, have Engelbart's ideas as yet had any influence on contemporary philosophy. In what follows I examine a philosophical framework which, I shall argue, stands in a logical, not an historical, relationship to Engelbart's conceptual framework. No influences in either direction will be imputed. In other words, in Popper's and Peirce's doctrines we find a more general framework which elucidates the augmentationist framework by relating it to other phenomena, and which itself is validated by whatever success the personal computer may eventually turn out to have in augmenting the human intellect. The question of such success will not be raised here, but I cannot leave the subject without concurring with both Engelbart and his critic Heim that the personal computer must be evaluated not simply on the basis of the greater efficiency with which it enables us to carry out familiar tasks, but on the basis of the value of the new modes of thinking this machine enables, inspires - or perhaps enforces.
Karl Popper's evolutionary epistemology was originally sketched in his 1965 Arthur Holly Compton Memorial Lecture 'Of Clouds and Clocks', and further elaborated in the two papers 'Epistemology Without a Knowing Subject' (1967) and 'On the Theory of the Objective Mind' (1968). The term 'evolutionary epistemology' is not actually found in these articles, but was introduced in 1974 in the landmark essay of that title by the psychologist Donald Campbell who, incidentally, has been profoundly influenced by Ashby as well as by Popper. Campbell's essay explores the blind-variation-selective-retention paradigm exemplified by both Darwinian evolution and the Popperian scientific method of conjectures and refutations, but Campbell himself was first introduced to this paradigm by Ashby in the nineteen-fifties. The past twenty years have seen a veritable explosion of publications on evolutionary epistemology, but I shall here focus simply on Popper's original formulations.
In his 1965 lecture Popper raised and discussed two problems: "Compton's problem", or the problem of how to reconcile rationality with physical determinism, and "Descartes' problem", or the problem of mind-body interaction. Drawing on both Charles Peirce's indeterministic cosmology and the evolutionary language theory propounded by Popper's erstwhile teacher Karl Buehler, Popper proceeded to sketch an account of human evolution as being largely exosomatic: "Human evolution proceeds, largely, by developing new organs outside our bodies or persons: 'exosomatically', as biologists call it, or 'extra-personally'. These new organs are tools, or weapons, or machines, or houses." Noting that other animals build lairs, nests, and dams, Popper points to the greater role played by exosomatic organs in human evolution, stressing the higher functions of language already postulated by Bühler:
"Yet the kind of exosomatic evolution which interests me here is this: instead of growing better memories and brains, we grow paper, pens, pencils, typewriters, dictaphones, the printing press, and libraries. These add to our language - and especially to its descriptive and argumentative functions - what may be described as new dimensions. The latest development (used mainly in support of our argumentative abilities) is the growth of computers."
This passage calls to mind Engelbart's use of the de-augmented pencil as an inverted metaphor for computer augmentation. In a footnote to an earlier passage in Popper's paper there is an even more striking echo of Ashby's characterization of an "intelligence-amplifier" as a machine that will solve problems people are incapable of solving. Against the view, attributed to Alan Turing, that people are - or are indistinguishable from - computers, Popper points out that it is precisely the difference between people and computers that makes computers useful to people:
"Moreover, we use, and build, computers because they can do many things which we cannot do; just as I use a pen and pencil when I wish to tot up a sum I cannot do in my head. 'My pencil is more intelligent than I', Einstein used to say. But this does not establish that he is indistinguishable from his pencil."
In Popper's and Einstein's view, then, as in the view propounded by Engelbart and Ashby, the specialized tools devised for symbol manipulation are valuable to the extent that they extend, augment, or amplify human intelligence, rather than replicating it. As far as I understand, no one denies that trying to make computers reason "like people" - whatever exactly that may mean - may be heuristically useful. What is denied, however, is that success in making computers perform certain specified tasks as well as people will prove that people, in some meaningful sense, are computers.
Central to Popper's evolutionary epistemology, as set forth in 'Of Clouds and Clocks', is the notion of hierarchical systems of plastic (i.e. nondeterministic) control. This epistemology implies a "view of organisms as incorporating - or, in the case of man, evolving exosomatically - this growing hierarchical system of plastic controls." This system, of course, is in all essential respects the same as Engelbart's H-LAM/T system. Thus far, I am pointing to parallels between what Popper is saying in the context of trying to solve two age-old philosophical problems and what Engelbart had already said in the context of formulating an R & D (research and development) strategy for computer technology. The existence of these parallels is intriguing - especially as there is no evidence that Popper had ever heard of Engelbart or vice versa , although, as we shall see, they share an indebtedness to Whorf - but what Popper the philosopher has to add to what Engelbart the engineer had already said is a more comprehensive framework, unifying the above ideas with a framework for understanding the nature of knowledge in general and of scientific knowledge in particular.
To examine that framework we turn to Popper's 1967 paper, 'Epistemology Without a Knowing Subject', published a decade before the appearance of personal computers. This is where Popper introduces his distinction among the first, second, and third worlds, later renamed world 1, world 2, and world 3. World 1 is the physical world, and world 2 is the mental, or inner world, both familiar from Descartes's dualism. World 3 is less familiar; this is "the world of objective contents of thought, especially of scientific and poetic thoughts and works of art." Popper admits that world 3 bears resemblances to Plato's world of ideas and to Hegel's objective spirit, while also differing significantly from these; it most closely resembles Gottlob Frege's realm of objective contents of thought - actually, Frege himself referred to this as "a third realm," a nomenclature apparently unnoticed by Popper, most of whose references to Frege are to his classic paper 'On Sense and Reference', where this expression does not occur. The ontological status of Frege's "thoughts" or "thought contents" is of course debatable; since Frege states that they cannot be perceived by the senses, it would seem that they are intensional objects. Popper does not, however, engage in an exegesis of Frege, but proceeds to list the principal denizens of world 3; they are: "theoretical systems, ... problems, ... problem situations, ... critical arguments, ... the state of a discussion or the state of a critical argument; and, of course, the contents of journals, books, and libraries."
Although these items, so characterized, could be regarded as purely intensional objects, Popper goes on to list not only the contents of books, etc. as members of his World 3, but the books-as-media themselves, and he makes clear that he regards these not as expressions of mental states, or as tools of communication, but as autonomous carriers of knowledge. Since World 3 contains both physical objects, such as journals and books, and intensional objects that can be physically embodied, such as theories and arguments, one is led to conclude that the inmates of World 3 are physical objects regarded as carriers of meaning. This corresponds closely to Charles Peirce's definition of a 'sign', to be explored below; also, of course, this makes World 3 a category of 'augmentation means', in Engelbart's sense. To continue, the inmates of Popper's World 3 are all human-made; they are artifacts like spider's webs and bird's nests; but, like other artifacts, they also have an autonomous existence, and Popper's chief thesis is that, to the extent that the theory of knowledge seeks to understand scientific knowledge, it should focus on World 3 objects, rather than on the beliefs and other subjective states of mind emphasized by mainstream epistemology from Descartes to Russell. Popper views as a fundamental mistake the consideration of books and articles as only the outwards expressions of knowledge that is 'really' residing inside the human mind. A book must obviously be readable or it would not be a book, but it need never actually be read and it would still be a book. In a passage eerily reminiscent of Peirce's definition of a 'sign' as something capable of being interpreted, Popper goes on:
"It is its possibility or potentiality of being understood, its dispositional character of being understood or interpreted, or misunderstood or misinterpreted, which makes a thing a book. And this potentiality or disposition may exist without ever being actualized or realized."
In an extreme case, proposed by Popper, a book of logarithms could be computer generated and printed, and it need never be read by anybody, but it would still constitute knowledge, in the scientific sense. Similarly, though human languages are products of the human mind, the externality of language is far more important than its mind- dependency. Referring to Karl Buehler's distinction of three functions of language - the expressive, communicative, and descriptive functions - Popper notes that the externality of language is an enabling condition of the critical method that is at the heart of science in particular and of rational thought in general:
"Without the development of an exosomatic descriptive language - a language which, like a tool, develops outside the body - there can be no object for our critical discussion. But with the development of a descriptive language (and further, of a written language), a linguistic third world can emerge; and it is only in this way, and only in this third world, that the problems and standards of rational criticism can develop."
Popper reiterates the evolutionary theme from 'Of Clouds and Clocks', to wit, that human cultural evolution proceeds largely through the invention and improvement, via selection, of exosomatic organs that embody ideas that are generated by the human mind, but that could not play the role they do if they remained within the mind. False ideas in the mind (e.g. about what happens if you jump out a tenth-floor window) may kill the person who has them and acts on them; false ideas outside the mind can be bloodlessly killed through rational criticism: "Scientists try to eliminate their false theories, they try to let them die in their stead. The believer - whether animal or man - perishes with his false beliefs." This analogy between natural selection and the scientific selection among hypotheses was stated already in the earlier lecture along with a warning against concluding that only the "utilitarian" theories which help us survive, will themselves survive. Artifacts are indeed selected from the vantage point of their fitness to some human aim - say, discovering truth or understanding nature - and obviously that aim will survive only if we survive; nonetheless, Popper holds the indirectness of the selection process as essential to its functioning.
The most important thesis Popper adds is that we can learn more about the human mind by studying its externalized products without regard to the specifics of their human origins than vice versa: "Contrary to first impressions, we can learn more about production behaviour by studying the products themselves than we can learn about the products by studying production behavior." There are two reasons for this. First, Popper notes - with an acknowledgement to the economist and Nobel laureate Friedrich Hayek, who has in turn repeatedly credited Descartes, David Hume, and Adam Ferguson - the most important products of the human mind are the unintended by-products of purposive behavior directed towards some other end. Language is paradigmatic in this respect; it is the result of intelligent, purposive behavior, but it is not the result of any attempt to create language. Second, the mind itself is affected and shaped through feedback from the evolution of its own externalized creations. As an example Popper instances Whorf's celebrated interpretation of the difference between English and Hopi time perceptions as products of the respective languages - an interpretation rooted in Whorf's relativistic hypothesis later embraced and extended to the realm of non-linguistic symbols by Engelbart, as we saw above. In Popper's words:
"I personally find Benjamin Lee Whorf's report on the Hopi Indians and their utterly different intuition of time convincing. ... Should Whorf be right, then our intuitive grasp of time - the way in which we 'see' temporal relations - would partly depend on our language and the theories and myths incorporated in it: our own European intuition of time would owe much to the Greek origins of our civilization, with its emphasis on discursive thought."
These themes are further pursued in the paper 'On the Theory of the Objective Mind', with special emphasis on their application to the history of science. After arguing that Galileo's mistaken theory of the tides is best understood with reference to its (World 3) problem situation, rather than with reference to biographical factors, Popper proceeds to compare and contrast his own method, called "situational analysis", with the philosopher-historian R.G. Collingwood's method of "re-enactment", without noting that the two methods had been explicitly equated by Popper's student Joseph Agassi. Although concurring with Collingwood on his emphasis on problems and problem situations, Popper then dismisses the re-enactment doctrine as a subjective method focusing on World 2 and using the analysis of objective problem situations as a mere aid to subjective understanding. As I have argued elsewhere, this is somewhat of a misunderstanding of Collingwood, whose doctrine was much closer to Popper's than the latter realized. Briefly, what Collingwood held was that any statement is intelligible only as the answer to a definite question and that any action is intelligible only as the solution to a problem. In a passage quoted by Popper, Collingwood argues that, to understand a certain edict by Emperor Theodosius, the historian must re-enact in his own mind the experience of the emperor; but he also makes it clear that what this means is finding and attempting to solve the problem confronting the emperor, and arriving at the solution expressed in the edict. There is no suggestion that, once the problem has been matched with its solution, there is some further subjective understanding or experience required for historical knowledge truly to exist; Collingwood's method is thus no less objective than Popper's.
To summarize this section. In Popper's epistemology we find an explicit recognition of the dependence of thought - specifically of rational, critical thought - on evolved external artifacts, which are human-made, but which through their externality attain an essential autonomy vis-a-vis the conditions and purposes behind their creation. Popper repeatedly makes the important point that the externality of, say, linguistic formulations is essential to their criticizability, and hence to their ability to function as vehicles of rational thought. In Popper's often stated view, only once something has been written down does it becomes criticizable and hence epistemologically interesting. In addition to explicitly embracing the Whorfian hypothesis of the linguistic relativity of thought, Popper thus implicitly lends support to Engelbart's "Neo-Whorfian hypothesis" that our future intellectual evolution will take place chiefly through advances in the tools for our manipulation of external symbols. Popper also explicitly classifies the computer with the pencil as tools for the manipulation of World 3 objects - years before the appearance of the personal computer.
What we do not find in Popper is an explicit statement that some important World 3 objects - theories, arguments, books, journals, etc. - are physically embodied, and that their specific physical embodiment, as distinct from their mere externality, affects their epistemic role. Although the mind-expanding function of the pencil was noted, the dependence of that function on the pencil's weight and shape - a dependence noted by both Engelbart and Petroski - is not explicitly noted by Popper. This defect in Popper's characterization of World 3 has been aptly noted by Paul Levinson, in a study that is otherwise highly sympathetic towards Popper:
"The problem with Popper's model is that its World 3 is drawn in terms too primarily ideational: the fundamental criterion for World 3 citizenship is being a humanly produced idea, with the material expression of ideas awarded a second-class or derivative status."
Levinson proceeds to formulate a technologically emended version of Popper's World 3 epistemology and apply it to the development of a philosophy of technology in general, and of media technology in particular .
For a classic thematization of the dependence of thought on the specific physical characteristics of the implements and vehicles of thought, in which the material expression of ideas is brought to the foreground, we shall turn, next, to Popper's great predecessor Peirce.
A closer relationship than with Collingwood - meaning more points of contact - may be detected between Popper's World 3 epistemology and the semiotic tradition represented by Popper's teacher Buehler, as well as by Buehler's predecessor Peirce and Peirce's latter-day disciple Umberto Eco. When Popper says, as already quoted, "we can learn more about production behaviour by studying the products themselves than we can learn about the products by studying production behaviour," the very wording foreshadows Umberto Eco's later distinction between a theory of codes and a theory of sign production, as well as the claim that the latter presupposes the former. The autonomy of World 3 is also expressed with all desirable clarity in the following passage from Eco's A Theory of Semiotics:
A signification system is an autonomous semiotic construct that has an abstract mode of existence independent of any possible communicative act it makes possible. On the contrary (except for stimulation processes) every act of communication to or between human beings - or any other intelligent biological or mechanical apparatus - presupposes a signification system as its necessary condition.
This semiotic tradition goes back, through the intermediary of Charles Morris, to the great nineteenth-century American astronomer, mathematician, logician, and philosopher Charles Sanders Peirce, who defined a sign as "something which stands to somebody for something in some respect or capacity." Peirce's various "anticipations" of Popper have, of course, been widely documented. As we have seen above, Popper himself traces his indeterministic cosmology back to Peirce, the common element being their shared conviction that classical mechanics is compatible with indeterminism. In addition, Eugene Freeman and Henryk Skolimowski have instanced: fallibilism (a label used by both Peirce and Popper); Peirce's analysis of abductive reasoning and Popper's method of conjectures and refutations (first equated, I believe, by A.J. Ayer ); Peirce's pragmatic maxim and Popper's principle of falsifiability; and their shared evolutionary conception of science as a living and growing enterprise. In a paper that is somewhat critical of Freeman and Skolimowski, as well as of Popper, llkka Niiniluoto has also pointed to Popper's rediscovery of Peirce's propensity theory of probability, and has documented that the common features listed by Freeman and Skolimowski are traceable to a shared ancestrage in the thought of the nineteenth-century philosopher and historian of science William Whewell, an indebtedness repeatedly acknowledged by Peirce. Finally, Israel Scheffler has noted that Peirce's rejection of foundationalism is echoed by Popper, in a wording that closely recalls Peirce's substitution of the bog for the bedrock as a metaphor for the basis of scientific knowledge.
One parallel, however, has not been noted by any of the above. While Freeman and Skolimowski discuss both Peirce's doctrine of signs and Popper's World 3 epistemology, they do not note that the entities comprising Popper's World 3 are signs in the Peircean sense, or that Peirce's doctrine of signs represents a World 3 epistemology that in many respects is more detailed and developed than Popper's; nor has this been noted by any other writer that I am aware of. Again, I do not mean simply that World 3 consists of objects which Peirce would have classified as signs - that would be a rather trivial observation, given the ambitious scope of Peirce's doctrine of signs. Rather, as we have already seen, Popper's own statement that a World 3 object, such as a book, is constituted by its "dispositional character of being understood or interpreted," is what recapitulates Peirce's definition of a sign as whatever is capable of being interpreted.
Peirce's doctrine of signs was first introduced in a series of articles published in The Journal of Speculative Philosophy in 1868. Peirce here specifically targeted Descartes and the Cartesian spirit in philosophy, and claimed that we have no ability to think without signs. This claim presupposes a prior argument that all self-knowledge can be accounted for as inferences from external facts and that, therefore, there is no reason to posit any power of introspection. We are directed, therefore, to look to external facts for evidence of our thoughts, thus, we encounter a near-tautology if we conclude that the only thoughts so evidenced are in the form of signs: "If we seek the light of external facts, the only cases of thought which we can find are of thought in signs." This doctrine is repeated in 1909 in the following words, which recall Socrates and Aristotle: "All thinking is dialogic in form. ... Consequently, all thinking is conducted in signs that are mainly of the same general structure as words;..."
We have just seen one of Peirce's definitions of 'sign'; in 1868 he gave this formulation:
"Now a sign has, as such, three references: first, it is a sign to some thought which interprets it; second, it is a sign for some object to which in that thought it is equivalent; third, it is a sign, in some respect or quality, which brings it into connection with its object.
What brings the sign into connection with its object is some material quality it possesses, which enables it to represent its object in a particular way. A picture signifies by means of an association that connects it, in the interpreter's brain, with its object (the picture resembles the object in some way). A weathervane or a tally signifies simply by being physically connected with its object by a chain of causation. Words, finally, can function as signs only because they are capable of being connected into sentences by means of a copula (what we would call an abstract or convention-based connection)."
Later, Peirce was to classify all signs as: icons, which signify by virtue of resemblance, indices, which signify by virtue of a physical connection with the object, and symbols, which signify by virtue of the existence of a rule governing their interpretation.
The exact physical embodiment of symbols is of course largely, but not entirely, a matter of convention. The choice of Roman versus Old English lettering may be purely conventional, but that is certainly not the case with one's preference for Arabic over Roman numerals for the purpose of performing long division - a point also made by Peirce, and later by Engelbart . The important observation Peirce made in 1868, and was later to develop in depth, as we shall see below, is that the existence of symbols depends crucially on the existence of a notation that is capable of symbolic interpretation, and moreover that our thinking is facilitated or impeded by the specific physical features of our notation.
By claiming that all thinking is signification, i.e. the production and interpretation of signs, Peirce was not denying the psychological fact that thoughts can be subjectively experienced as internal states of mind; what he is claiming is that such experiences are merely psychological facts: "[Every] thought, in so far as it is a feeling of a peculiar sort, is simply an ultimate, inexplicable fact." The meaning or intellectual value of a thought lies in its potential for further interpretation, whether by my mind or by some other mind, that is, it lies in its signhood:
"Finally, no present actual thought (which is a mere feeling) has any meaning, any intellectual value; for this lies not in what is actually thought, but in what this thought may be connected with in representation by subsequent thought; so that the meaning of a thought is something altogether virtual."
In a footnote, Peirce adds: "just as we say that a body is in motion and not that motion is in a body we ought to say that we are in thought and not that thought is in us." These are the ideas which Peirce proceeded to develop into the famous and widely discussed theory of meaning known as pragmatism; on this occasion we shall look at some less known aspects of the doctrine of signs.
The recognition of the sign character of thought led Peirce repeatedly to emphasize the externality and artifactuality (if I may so call it) of knowledge. Thus, in 1902 Peirce wrote, in a passage echoed by Popper sixty-odd years later:
" I think it more accurate to say that there are stores of knowledge in a library than to say that there are books there which, if I was reading, having become acquainted with the languages they are printed in, would be conveying knowledge to my mind."
Similarly, in another manuscript from the same year: "In my opinion it is much more true that the thoughts of a living writer are in any printed copy of his book than they are in his brain." These are not just casual observations, but corollaries of the doctrine of signs.
A sidelight on just how seriously Peirce took the physical embodiment of knowledge as an essential constituent of it is provided by his obituary of Charles Babbage, the inventor of history's first programmable computer, the Analytical Engine. After praising Babbage's engine as "the most stupendous work of human invention," Peirce goes on to instance Babbage's publication of a volume of logarithms, where Babbage tried fifty different colors of paper, and ten of ink, before settling on the combination that was easiest to read.
From Peirce's perspective, Babbage's choice of the right paper-ink combination for his table of logarithms was itself as much a contribution to scientific knowledge as the computation of the logarithms themselves. Knowledge, in Peirce's semiotic doctrine, consists less in states of mind ("ultimate, inexplicable facts") than in the potentiality of external objects to induce certain states of mind, and this potentiality depends on the specific physical characteristics of said external objects.
Peirce never denied the existence of consciousness, but he did deny that consciousness is an essential attribute of mind. In an especially topical passage Peirce goes on to emphasize the dependence of our language faculty on external tools for linguistic expression:
" A psychologist cuts out a lobe of my brain (nihil animale a me alienum puto) and then, when I find I cannot express myself, he says, 'You see, your faculty of language was localized in that lobe.' No doubt it was; and so, if he had filched my inkstand, I should not have been able to continue my discussion until I had got another. Yea, the very thoughts would not come to me [emphasis added]. So my faculty of discussion is equally localized in my inkstand."
As is indicated by the emphasized sentence, Peirce is not making the trivial point that without ink he would not be able to communicate his thoughts. The point is, rather, that his thoughts come to him in and through the act of writing, so that having writing implements is a condition for having certain thoughts - specifically those issuing from trains of thought that are too long to be entertained in a human consciousness. This is precisely the idea that, sixty years later, motivated Engelbart to devise new technologies for writing so as to improve human thought processes, as well as the idea that motivated Havelock's interpretation of Plato.
As was pointed out by Freeman, Peirce's doctrine of signs implies that all reasoning is diagrammatic, a corollary repeatedly made explicit by Peirce himself, e.g. in this passage quoted by Freeman: "For reasoning consists in the observation that where certain relations subsist certain others are found, and it accordingly requires the exhibition of the relations reasoned within an icon." The value of the iconic representation, Peirce repeatedly insisted, lay in the possibility it afforded of performing experiments on our thoughts, by changing some elements in the diagram and literally seeing new relations appear. Since Popper has opined that there is "a big gulf" between Peirce's dictum "all reasoning is diagrammatic" and his own "all reasoning is critical", we need to note that, in Peirce's view, the chief benefit of the externalization of thought is the opportunity it offers for logical criticism through the performance of the type of experiments just noted. It is worth noting, also, that Peirce regarded the self- critical use of language as the chief difference between humans and other animals. People not only use signs; we think of signs as signs, and this mode of thinking allows us to control our thinking through criticism. In Peirce's words:
"All thinking is by signs; and the brutes use signs. But they perhaps rarely think of them as signs. To do so is manifestly a second step in the use of language. Brutes use language and seem to exercise some little control over it. But they certainly do not carry this control to anything like the same grade that we do. They do not criticize their thought logically."
In short, we differ from other animals through a use of language that enables logical criticism of our thought - precisely the doctrine Popper inherited from Buehler.
In keeping with this view, Peirce developed his system of "existential" graphs as a new notation for logic. (That Peirce wrote extensively on topology and cartography is also not coincidental.) As has been observed by Martin Gardner, no doubt correctly, the graphs were not intended as a practical improvement over existing algebraic notations, but as a means of laying bare the diagrammatic essence of thought and laying before the reader, in Peirce's words, "a moving picture of thought". Such a moving picture can also be exhibited by a machine, as Peirce made clear in his 1887 article on the logic machines constructed by William Stanley Jevons and Allan Marquand:
"The secret of all reasoning machines is very simple. It is that whatever relation among the objects reasoned about is destined to be the hinge of a ratiocination, that same general relation must be capable of being introduced between certain parts of the machine."
To illustrate, the syllogism, 'If A then B; if B then C; therefore, if A then C', can be embodied in a machine where pushing the lever A activates the piston B, which in turn rotates the wheel C, so that pushing A in effect activates C. Machines, then, are capable of reasoning, in the sense of drawing inferences, but in Peirce's view this is not a peculiarity of logic machines but a general characteristic of machines, including here a wide variety of apparatus not generally thought of as machines: "Accordingly, it is no figure of speech to say that the alembics and cucurbits of the chemist are instruments of thought, or logical machines."
Without here venturing into speculations about what Peirce might have said about modern digital computers, we shall observe that he pointed to two differences between people and the logic machines of his day. First, machines are "destitute of all originality, of all initiative". In a machine, Peirce stresses, this is a good thing; it is precisely the machine's lack of originality that makes it predictable and hence useful; a balloon, for instance, has limited usefulness because it has too much initiative: "We no more want an original machine than a house-builder would want an original journeyman, or an American board of college trustees would hire an original professor." This is precisely the view of machines later echoed by Popper, as well as by Licklider. It was also expressed more than thirty years before Peirce wrote the above, by Charles Babbage's associate Lady Ada, Countess Lovelace, history's first computer programmer: "The Analytical Engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform."
Second, Peirce observes, a logic machine is limited by its design: "it has been contrived to do a certain thing, and it can do nothing else." Now, Peirce is not denying, but explicitly admitting, that there could be nondeterministic machines constructed like Jacquard looms, incorporating if-then loops, and capable of handling a great variety of problems. What is of chief interest here is not, however, Peirce's observations about machines per se, but what he is obliquely saying about people. People are like machines, in that the unaided individual is as limited by his/her design as is the machine. But people do not think unaided, and therein lies the difference: "The unaided mind is also limited in this as in other respects; but the mind working with a pencil and plenty of paper has no such limitations." (There are of course other limitations; sometimes you need tools for building three- dimensional models to visualize relations that cannot be represented two-dimensionally.) Peirce goes on to extol the virtues of algebra, in particular the empowering function of the parenthesis, which enables levels of abstraction that would be literally unthinkable without it; later in life, as he perfected his logical graphs, he was to express a preference for diagrammatic over algebraic notation, but the general point remains that people reason by means of parentheses, diagrams, alembics, cucurbits, and other machinery (what we would today call soft or virtual and hard), and that no one machine can capture this general machine-dependency of human reasoning.
Peirce's view of thinking as the production, development, and interpretation of signs led him to attribute reasoning to a wide range of artifacts, simple as well as complex. But the distinctly human type of reasoning consists in the use of external artifacts as aids to thought. For Peirce, as later for Popper, this externalization of the objects of thought is what made logical criticism, and hence rationality, possible. Peirce, however, showed a far greater awareness of the importance of the exact physical embodiment that this externalization took, an awareness that led him to spend a great portion of his life studying and perfecting various types of notation, diagrammatic as well as symbolic. Again, for both thinkers, there are no utilitarian or instrumentalist implications: that thoughts are instruments for survival (as some thoughts certainly are) is not primary; the point is rather that a variety of external things are instruments for our thinking. Further, our conception of how we think will determine how we structure those external things, just as their structure in turn guides the direction of our thinking.
In the above we have examined and documented a coherent set of ideas common to Engelbart's augmentationism, Popper's evolutionary epistemology, and Peirce's pragmatistic semiotic. Let us summarize these ideas in eight theses, not all of them explicitly shared by all three thinkers:
1. Human evolution proceeds exosomatically, through the evolution of external artifacts.
2. Human knowledge does not reside in the head, but in a subclass of our evolved artifacts: Engelbart's "external symbols", Popper's "World 3", Peirce's "signs".
3. The Whorfian hypothesis: Our fundamental world view is shaped by one of our evolved exosomatic organs, namely our language.
4. The Neo-Whorfian hypothesis: Our thinking is also shaped in a variety of ways by other exosomatic organs, specifically the tools by which we manipulate external symbols.
5. Abstract, critical thought is made possible only by the existence of written language. This thesis is also embraced by Buehler and Havelock.
6 The possible degrees of abstraction (as well as the possible kinds of intellectual projects) available to us depend on the logica or effective features of the symbols by which we represent abstractions - Arabic numerals, parentheses, and the like - i.e. not what the symbols look like, but what they can and cannot do.
7. The possible degrees of abstraction available to us depend on the physical features of the tools with which we produce and manipulate symbols - pencils, computers, and the like.
8. The digital computer is most fruitfully regarded as a means for augmenting human intelligent activity by automating and accelerating the production and manipulation of symbols, rather than as an independent source of intelligent behavior.
Theses 6 and 7 are not explicitly found in Popper, and thesis 8 is perforce not found in this exact form in Peirce, although his comments on logic machines come very close. All technology, as has been made clear by Paul Levinson among others, is the embodiment of ideas. But only some technologies are the embodiment of ideas about ideas (what Levinson calls "meta-cognitive" technology); such a technology is today's personal-computer technology. The digital computer itself may be considered the physical embodiment of Boolean algebra together with the recursive Church-Turing definition of computability.
The articulation and extrapolation of these ideas as an approach to understanding the computer as a reasoning machine may seem to lead to a mechanistic paradigm of reasoning and of mental activity in general - a paradigm actually embraced by Turing himself. Whether such a paradigm is in any way implied by the Church-Turing thesis is obviously too large a question to be raised here. But, although this question is not irrelevant to the subject of this paper, we need to emphasize that the ideas embodied in the construction of the inference engine are of little help in explaining what the computer is for, or how to use it for maximum benefit to the human race.
These are the questions that were raised by the augmentationists - Bush, Licklider, Engelbart, et al. - and answered with reference to human capabilities and to the design of the computer's user interface, rather than to the principles governing the operations of the inference engine behind the interface. And the thesis with which I shall conclude is that, by studying the revolution the augmentationists have accomplished in the computer and our conception of the computer, as well as the ideas motivating that revolution, we are led to a nonmechanistic, evolutionary, and semiotic conception of mind and knowledge as the most appropriate philosophical paradigm for understanding this latest revolution in technology.
* I am indebted to Howard DeLong of Trinity College for numerous helpful comments on successive drafts of this paper. Levinson calls "meta-cognitive" technology); such a technology is today's personal-computer technology. The digital computer itself may be considered the physical embodiment of Boolean algebra together with the recursive Church-Turing definition of computability.
The articulation and extrapolation of these ideas as an approach to understanding the computer as a reasoning machine may seem to lead to a mechanistic paradigm of reasoning and of mental activity in general - a paradigm actually embraced by Turing himself. Whether such a paradigm is in any way implied by the Church-Turing thesis is obviously too large a question to be raised here. But, although this question is not irrelevant to the subject of this paper, we need to emphasize that the ideas embodied in the construction of the inference engine are of little help in explaining what the computer is for, or how to use it for maximum benefit to the human race.
These are the questions that were raised by the augmentationists - Bush, Licklider, Engelbart, et al. - and answered with reference to human capabilities and to the design of the computer's user interface, rather than to the principles governing the operations of the inference engine behind the interface. And the thesis with which I shall conclude is that, by studying the revolution the augmentationists have accomplished in the computer and our conception of the computer, as well as the ideas motivating that revolution, we are led to a nonmechanistic, evolutionary, and semiotic conception of mind and knowledge as the most appropriate philosophical paradigm for understanding this latest revolution in technology.
* I am indebted to Howard DeLong of Trinity College for numerous helpful comments on successive drafts of this paper.