Wittgenstein, Edwards, and the humanity of robots
by David K. Nartonis

Abstract. What does it mean to be human? When artificial intelligence and theology are in dialog, I believe the question cannot be avoided. From my reading in theology and robotics, two opposing definitions emerge. One definition of humanity depends on what is inside, a soul or a set of instructions. On the other hand, the definition I favor restricts itself solely to what can be observed and verified from the outside. To support this idea that “being human” is strictly outer and social, I turn, rather surprisingly, to the eighteenth century theologian, Jonathan Edwards, and, less surprisingly, to the twentieth century philosopher, Ludwig Wittgenstein. Along the way, I also point out how other sources, from scripture to contemporary books and movies, are contributing to a debate which, I believe, will only grow more urgent as humankind enters the next thousand years of technological advance.

What does it mean to be human? And what might it mean for a robot to be human? I believe that we commonly define humanity as both outer and inner. We observe certain qualities and actions in others -- how they look, speak, act and interact, and we know that when we look, speak, act or interact in just these same ways, we are at the same time having certain inner experiences. Naturally, then, when we see others who look and behave as we do, we assume that they are having their own inner experiences. For example, we assume others are fully conscious when they appear and act in certain ways. But what warrants this assumption? Aren’t we totally basing the assumption of inner consciousness on outer and public qualities and actions that we observe? I believe this must be the case because we have no access to the inner states or experiences of others.

If this is true, when we say that someone is “conscious” we are saying something that we know by observation about their outer qualities while adding something we can only assume about their inner experience. Likewise, this is equally true of every other human quality -- intention, emotion, intelligence, perception -- and, in this sense, when we describe someone, or something, as human we are saying what we know to be true about their outer qualities. This is not saying that others do not have the same inner experiences that we do, but only saying that we cannot know whether they do or not.

It is for this reason that, for me, the word “human” legitimately refers only to what is outer and public about us, even though I know that we commonly use the word to also refer to something inner and hidden. You will see in this talk, that I apply this same criteria to robots, considering them human if their outer and public characteristics are what we call human. Two historical figures have helped me to reach this conclusion, the philosopher, Ludwig Wittgenstein, and the theologian, Jonathan Edwards. Whether or not you can accept my conclusions, I believe that the important question of machine humanity will only become more urgent as (1) more human-like robots are built and (2) science defines us mortals more and more as biological machines.

Wittgenstein versus Turing on the humanity of robots

My first section is called, “Wittgenstein versus Turing on the humanity of robots.” About fifty years ago, the philosopher, Ludwig Wittgenstein, rejected the idea that being human refers in any essential way to what is inside us or to what is inside a robot. Wittgenstein took this position when he argued about the humanity of robots with his student, Alan Turing. In 1936, in a paper known as “On computable numbers,” Turing showed how to break mathematical processes, like adding and subtracting, into many small steps that a machine can be programmed to take. So far, so good. But Turing also claimed that when a machine is taking these steps, and getting right answers, it is doing math just like a human. Against this audacious claim, Wittgenstein raised an important objection. According to him, there is an observable and fundamental difference between the way a human and a machine execute Turing’s program, and merely reducing math to many small steps does not eliminate this difference. If Wittgenstein is right, two important conclusions follow. On one hand, it follows that Turing was premature in claiming that his 1936 machine does math like a human. On the other hand, Wittgenstein shows exactly how a future machine might actually become human, by simply eliminating this difference.

I’ll return to these two points, but first it is important to ask, What was this observable difference that Wittgenstein found between a human and a machine when both execute Turing’s program? Were he here, Wittgenstein would answer that humans carry out Turing’s program because they regard its steps as normative, whereas Turing’s machine does not. For simplicity, consider one aspect of this difference. Humans follow instructions because they intend or want to get the right answer. Thus we might call what the human does “intentional math.” But my PowerMac adds one plus one and gets two, not because it intends or wants to get the right answer, but because its construction and its programming determine what it does. Thus, my computer is doing what we might call “causal math.”

The fact that this difference is observable is crucial to Wittgenstein’s view of humanity and robots. When we watch others do math, we use “intention” and “want” legitimately to refer to qualities and actions that are outer and public, but carelessly to what we suppose their inner states and experiences might be. For example, imagine I am a math teacher with a pupil called Mary. When I say, “Mary wants the right answer,” legitimately I may be referring to her tense posture when she works a math problem, or to how she pushes on her pencil. Legitimately I may be referring to her reaction when her homework is corrected, or to her many questions about math. Legitimately I may even be referring to what she tells me about her inner states and experiences during math class. But when I say, “Mary wants the right answer,” I cannot legitimately be referring to her actual inner states and experiences, since they are inaccessible to anyone but Mary.

Now picture both Mary and Turing’s machine executing Turing’s instructions. When I say, “Mary is trying to follow the program exactly,” I am legitimately referring to what I observe in her actions and demeanor. When I say, “The machine exhibits no intention in its execution,” then I am legitimately referring to what I observe of the machine -- its lack of human form, its lack of variation in response, its lack of response to interruption. Additional reference to its lack of inner experience is unwarranted because this is, again, something I cannot possibly know. In a similar way, I can only legitimately say of a human or a machine, that it is or is not conscious, perceptive, emotional, intelligent, or in any other way human, if by these words I am referring to its outer and public characteristics. It is precisely in the lack of these that a simple machine, like that envisioned by Turing in 1936, fails to “do math, like a human.”

I mentioned that, from this difference overlooked by Turing, two conclusions follow. On the one hand, this difference shows that Turing was premature in claiming that a machine executing his 1936 program is doing math like a human. On the other hand, if we eliminate this difference by building, in the future, an ultra-sophisticated machine that looks, speaks, acts and interacts, just like a human doing math, then we will finally have the humanity that Turing prematurely claimed for his very primitive robot. Furthermore, if the machine has the observable qualities, like want and intention, that lead us to call it human, then the inner operation of the machine is no more than a very difficult technical issue and not a legitimate part of the determination of its humanity. As long as whatever goes on inside produces the observable effects we call human, inner processes, states, and experiences are, according to Wittgenstein, otherwise irrelevant.

At this point, you might object, “But I know that what goes on inside of me is relevant to my humanity.” Recall that I’m not denying the existence of inner processes and functions, in either a human or a machine. But what Wittgenstein and I are saying is that we can only define each other’s humanity by external characteristics that we can observe. Consider the analogy of a beautiful yellow flower. Even though the atoms that make up the flower produce the color we see in it, the atoms themselves are not actually yellow -- at least they are not yellow in the same way that the flower is yellow. Just so, even if our inner states, processes, or experiences produce the outward and public humanity that others see in our appearance and actions, this outward and public humanity is not a quality of these inner states, processes, or experiences.

Defining our own humanity and the humanity of machines is likely to become even more urgent in our technological future. Consider, for example, all the genetic engineering that could take place over the next 100, or the next 1000, years. The inner processes of at least some of our grandchildren may be very different than ours. But will they be less human if they still look, speak, act and interact as we do? In addition, what if you found out tomorrow that a beloved sibling was an artificial human all along? Would you suddenly stop loving this person? And what if we someday build robots that are as human as us, will this final blow to human “specialness” be the end of religion? Or is there a theology that can accommodate human robots?

Jonathan Edwards and the humanity of robots

It is with just these kinds of questions that I think the theology of Jonathan Edwards has its own contribution to make. My second section, then, is called “Jonathan Edwards and the humanity of robots.” So far, we have been wrestling with a question raised, at least for me, by robotics and sharpened by philosophy, the question of what it means to be human. Now I want to introduce one possible theological approach to the problem of artificial humanity. This approach grows out of the long tradition of Christian Platonism and I find it in the writings of American preacher and theologian, Jonathan Edwards.

Edwards arrived at Yale College in the early 1700s, just as books by Isaac Newton and John Locke were bringing the modern concept of science to America. He was quickly struck by the pervasive materialism of this new movement. A world that had once been seen as God’s gracious activity on behalf of humanity was now said to run by itself. Without rejecting Newton’s physics or Locke’s psychology, Edwards spent the remaining 38 years of his life mounting a massive resistance against the incursion of this materialism into traditional religion. He did this in more than 1000 sermons, a dozen publications, and volumes of unpublished writings that restated the theology of his Puritan ancestors in powerful new terms.

Edwards’ concept of personhood was foundational to his restatement of Puritanism. In this respect, Edwards drew on a theological tradition that can be traced back to ancient Greek, Jewish, and Christian sources such as St. Augustine and, even farther back, to Philo of Alexandria, a little known Jewish scholar who was a contemporary of Jesus. Philo puzzled long and hard over the fact that humankind is created twice in Genesis, the first book of Moses. In Genesis 1 we read, “And God said, Let us make man in our image, after our likeness.” A few verses later, in Genesis 2 we read, “And the Lord God formed man of the dust of the ground, and breathed into his nostrils the breath of life; and man became a living soul.”

Acquainted with Plato, through his Greek philosophical education, Philo presented his readers with the possibility that humanity was created first as an idea in the mind of God. The second creation, then, was the realization of God’s idea by first forming a matter body and then putting something into it that would make it human. This somewhat Platonic explanation of the two creations of humankind was picked up by early Christian writers and a version of it can be found in St. Augustine’s writings. Thanks to Augustine, this “spiritual” interpretation of Genesis reappears throughout the Middle Ages and profoundly influenced various Renaissance writers. For example, it is repeated in the first few chapters of Sir Walter Raleigh’s History of the World.

Building on Genesis, as interpreted by Philo and Augustine, and perhaps on renaissance sources like Raleigh, Edwards understands men and women as image and likeness rather than as physical bodies with something put into them. Thus, he defines personhood in terms of qualities expressed rather than something inside. But to really encounter what personhood and humanity are for Edwards, we must dig into his doctrine of the trinity. We must do this because he models the personhood of men and women on the personhood of God.

We begin, Edwards tells us, with a God who is a divine mind. Being a mind, this God thinks about itself and thus forms an idea of itself. We might imagine this as God looking into the mirror of God’s own thought. However, when the divine mind forms this idea of itself, the result is such a perfect image that this idea is also God. In Edwards’ own words, “the perfect idea God has of himself is...properly God,” because, “if God doth think of himself...with perfect clearness...that idea he hath of himself is absolutely himself again.” For Edwards, this image is the Christ or second person of the trinity.

Thus, says Edwards, we have two persons, Father and Son, that know and love each other. Furthermore, in loving each other, this mind and its Christ-idea form another kind of image, a pure love or “affect” which is also God. This affect is the third person of the trinity, the Holy Spirit. Again in Edwards’ own words, “The Holy Spirit is the act of God between the Father and the Son infinitely loving and delighting in each other.” Thus, “God is glorified within himself...(1) by appearing to himself in his perfect idea...(2) by...delighting in himself by flowing forth in infinite love towards himself...in his Holy Spirit.”

How, then, does Edwards explain the existence of mortals, living in a material universe? As he puts it, “the Son has also an inclination to communicate himself.” Thus, “God determined...[that] his Son...have an object...of his infinite grace and love...to receive God’s love.” This object is human consciousness which, after the fall of man, shatters into individual “consciousnesses.” These hold in themselves their own bodies and a universe of other objects and events, communicated to them by God. “A mind or spirit is nothing else but consciousness and what is included in it.”

By this stage I hope it is clear that Edwards has completely reversed the usual meaning of “embodiment.” In his view, God did not first create a physical body and then put consciousness into it. Instead, God created men and women as consciousness and then put both universe and body into them. Our universe, then, Edwards describes as “only mental” and “the substance of bodies becomes...nothing but the Deity acting in that particular manner.” Here, then, is a theology that defines humanity in terms of the observable qualities we reflect back to God. “For God to glorify himself is to discover himself in his works, or to communicate himself in his works, which is all one.” I think, Edwards’ definition of humanity, in terms of the image we send back to God, opens the way to a theological definition of the humanity of robots.

However, I realize that it is very difficult for most of us to join Edwards in his view of personhood and the universe. We are used to thinking of ourselves as having a consciousness within a body which, in turn, is within the universe. But Edwards challenges us to think of body and universe as within us, within individual consciousness. Thus he interprets our familiar experience in a largely unfamiliar way. Let’s take Edwards, then, point by point, and relate him to Wittgenstein:

(1) We have “inner” experiences that no one else can share. According to Edwards these are within consciousness but not within the body.

(2) We also have outer and public experiences, including the appearance and behavior of others. According to Edwards these are, indeed, individual experiences, but experiences made public through divine correlation. For example, if God causes my image of my body to raise its hand, then God causes your image of my body to raise its hand too.

(3) Part of what we know of others is this outer and public experience of their appearance, speech, and behavior. Recall, that Wittgenstein insisted that this is all we can legitimately know of others. For Edwards, this appearance, speech, and behavior are included in the image and likeness we receive from God and send back to God.

(4) We commonly go on to assume that we know the inner experience of others when we see our appearance and behavior in them. For example, when we feel anger, our body takes on observable qualities. When we see these same observable qualities in the actions of others, we assume their inner experience of anger. Wittgenstein saw this assumption as unwarranted -- and irrelevant to the definition of humanity. Equally irrelevant for Wittgenstein would be any assumptions about the inner states and processes of others. On this latter point, I think Edwards would agree with Wittgenstein that the inner states and processes of the body are irrelevant to the definition of humanity. This is because, for him, the body is only an image within consciousness and the “inner workings” of this image do not produce its outer and visible characteristics.

(5) Now let’s apply Edwards’ theology to the inner states and processes of robots. Since the robot is an image in consciousness, its inner states and processes are irrelevant to its humanity. This, then, is a major sense in which I would say that Edwards’ theology makes room for the humanity of robots.

(6) Having dismissed the relevance of inner states and processes, what about a robot’s “inner experiences.” Edwards would not agree with Wittgenstein that our inner thoughts and feelings are irrelevant to our humanity because they too are part of the image and likeness we send back to God. However, some recent commentators have cautioned readers of Edwards’ theology not to separate what is “outer” and “inner” in his definition of personhood. For example, Stephen H. Daniel tells us that, in Edwards’ theology, we “cannot consider the intentionality of an act apart from the act itself.”

If Daniel is right that inner and outer are one in Edwards’ theology, then when we see the image and likeness of God in the outward characteristics of a robot, we are seeing its humanity. This rather startling possibility can be traced back to Edwards’ relational or “social” definition of the trinity. The Holy Spirit is not “in” the Father or the Son but “between” them, as their loving interaction. Similarly, the Father is not within the Son but sees its reflection in the Son, a matter of appearance. As the image and likeness of this social trinity, humanity is also defined by appearance and interaction, by its public characteristics rather than by something “hidden inside.”

Conclusion

Further exploration is needed, but I am convinced that an understanding of personhood, as defined by Edwards, may indeed have an impact on the questions raised by Artificial Intelligence and Cognitive Science. To summarize this point, on which I believe both Wittgenstein and Edwards’ would agree, I will leave you with a vivid illustration. At the close of John Steinbeck’s novel, Grapes of Wrath, a starving man is saved by a destitute woman who feeds him at her breast. Where do we find the humanity in this humiliating and, yet, exhalting act? Is it in the body of the woman, in her lactation? Is it in the body of the starving man, in the onset of death? As essential as these embodiments are, I do not think that it is inside these fictional characters that we find their humanity. It is in something that we observe happening between them, in something social and outward, that we see their humanity, and even glimpse their divinity.

Notes

  1. . Born in Vienna in 1889, Wittgenstein is best known for two periods of creative activity, both centered on Cambridge University in England. Here I will focus on the second period, from 1929 until his death in 1951. To the extent that I understand Wittgenstein’s philosophy generally, I agree with the interpretations of P. M. S. Hacker. See, Wittgenstein’s Place in Twentieth-Century Analytic Philosophy (Oxford: Blackwell Publishers, 1996).
  2. . Turing was born in London in 1912, lived until 1954, and is known for his code-breaking activities during World War II. He was also a pioneer in building computers and thinking about artificial intelligence. Here I follow Stuart G. Shanker’s analysis of this disagreement between Wittgenstein and Turing. See, “Wittgenstein versus Turing on the Nature of Church’s Thesis,” Notre Dame Journal of Formal Logic 28 (1987): 615-649. I have also been helped by lively e-mail discussions with Professor Norman Gall of the University of Winnipeg and Professor Peggy DesAutels of the University of South Florida. Norman Gall, e-mail discussions with author, Winnipeg, Canada, October 29, 1997 and February 24, 1998. Peggy DesAutels, e-mail and telephone discussions with author, St. Petersburg, Florida, April 20 and April 22, 1998.
  3. . Alan Turing, “On computable numbers, with an application to the Entscheidungsproblem,” Proceedings of the London Mathematical Society 42 (1936-1937): at No. 9. This paper quickly became the theoretical basis for the modern computer.
  4. . Turing considers “a man in the process of computing a real number,” which he calls a “computer.” He then strips away everything that might indicate that this man is doing math like a human and describes him instead as a machine doing causal math. “The behaviour of the computer [man] at any moment is determined by the symbols which he is observing, and his ‘state of mind’ at that moment.” Turing then shows how to “construct a machine to do the work of this computer.” To this, Wittgenstein objected that humans that calculate in the deterministic or “causal” way assumed by Turing would not actually be humans as we know them, but only biological versions of Turing’s machines. “If calculating looks to us like the action of a machine, it is the human being doing the calculation that is the machine.” Quoted in “Wittgenstein versus Turing,” 616-622.
  5. . Another aspect of normative, as Wittgenstein conceives it, is that humans initially develop, continually extend, and are able to explain and justify the rules that they follow. See “Wittgenstein versus Turing,” 622.
  6. . As Norman Gall put it in a February 24, 1998, e-mail to me, Turing’s machine will have to evolve into “a Data (from Star Trek) type-of-thing” before we can say that it is human
  7. . “Wittgenstein versus Turing,” 619-622, 640-641.
  8. . The deep historical connection of Edwards to the author’s own Christian Science tradition is explored in a 1983 dissertation. See Thomas C. Johnsen, “Christian Science and the Puritan Tradition” (Ph.D. diss., The Johns Hopkins University, 1983). This connection is also pointed out by John F. Wilson in his “Editor’s Introduction” to Jonathan Edwards, A History of the Work of Redemption (New Haven: Yale University Press, 1989), 84, 94.
  9. . Sydney E. Ahlstrom describes Edwards as “communing with the long and rich tradition of Christian Platonism, dwelling on the ancient metaphysical problems of the faith, pondering the inner mysteries of the Trinity, concerned with the ineffable problems of Being and uniting with precisely those aspects of St. Augustine’s theology which Luther and Calvin had found least acceptable and with which Locke could never have sympathized.” See “Theology in America” in The Shaping of American Religion, vol. 1, James Ward Smith and A. Leland Jamison, editors (Princeton: Princeton University Press, 1961), 243-251. The place of Edwards’ theology in American history is more fully examined in volume 1 of Elizabeth Flower and Murray G. Murphey, A History of Philosophy in America (New York:
  10. G. P. Putnam’s Sons, 1977).
  11. . Thomas H. Tobin, “Interpretations of the Creation of the World in Philo of Alexandria,” in Creation in the Biblical Traditions, Richard J. Clifford and John J. Collins, editors, The Catholic Biblical Quarterly Monograph Series 24 (Washington, DC: Catholic Biblical Association of America, 1992), 119-122; David T. Runia, Philo in Early Christian Literature: A Survey (Minneapolis: Fortress Press, 1993), 324-327.
  12. . Sir Walter Raleigh, The history of the world (London, 1614).
  13. . Here I will follow the exposition of Edwards’ views by the modern American theologian, Robert W. Jenson. See his, America’s Theologian: A Recommendation of Jonathan Edwards (New York: Oxford University Press, 1988).
  14. . Jonathan Edwards, Dissertation Concerning the End for which God Created the World, (n.p.: n.d.), I, I, 3-5, in Jenson, 39 and 202 n.19.
  15. . In addition, “The Almighty’s knowledge is...by an idea, as ours is, only infinitely perfect.” Jonathan Edwards, Miscellanies, transcribed by Thomas Schafer (ms. in Yale University’s Beinecke Library), 179, 238, quoted in Jenson, 96 and 208 nn. 20, 21.
  16. . Miscellanies, 94, in Jenson, 97 and 208 n. 23.
  17. . Miscellanies, 448, in Jenson, 42 and 202 n. 37.
  18. . Miscellanies, 1004, in Jenson, 42 and 202 n. 38.
  19. . Miscellanies, 1245, in Jenson, 105 and 209 n. 22. It is for this reason that, “the church is said to be the completeness of Christ.” Miscellanies, 1004, in Jenson, 42 and 202 n. 38.
  20. . Jonathan Edwards, Scientific and Philosophical Writings, ed. Wallace E. Anderson (New Haven: Yale University Press, 1980), 342-343, quoted in Jenson, 30 and 200 n. 32.
  21. . “For we are to remember that the world exists only mentally.” Miscellanies, 247, in Jenson, 39 and 202 n. 20.
  22. . Anderson, 353, 215, quoted in Jenson, 27, 28 and 200 nn. 18, 26.
  23. .Miscellanies, 247, in Jenson, 39 and 202 n. 20. See also, Anderson, 206, 215-216, 234-235, 339, 342-344, 350-352, 354, 377-380; and Miscellanies, 1, 267, in Jenson, 28-32 and 200-201 nn. 25-30, 32-44.
  24. . Stephen H. Daniel, The Philosophy of Jonathan Edwards: A Study in Divine Semiotics (Bloomington: Indiana University Press, 1994), 158.
  25. . In fact, Daniel sounds almost like Wittgenstein when he goes on to say that, “a system of meaning that is limited to a private language-user is no system at all, for the individual cannot guarantee the stability of relations required for intelligibility without appealing to a criterion other than the expressions and actions chosen by the individual....Without such a criterion, [Edwards] would be open to charges that virtuous choices could be evaluated apart from virtuous acts, and that individuals might be virtuous apart from their communal engagements in terms of which human life is significant.” Ibid., 172.
  26. . John Steinbeck, The Grapes of Wrath , with an introduction by Robert DeMott (New York: Penguin Books, 1992), 617-619.