Knocking on Heaven's Gates:

Reflections on Immanuel Kant's Kingdom of Ends in the Age of Computersl

How would we recognize a "rational being"?

If we keep increasing computers’ processing power and their ability to learn, should we expect them to eventually exhibit the sort of independent, thoughtful initiative characteristic of rational beings? This question are not premature or overly speculative. Computer scientists are seriously discussing the implications of recent developments in self-programmable hardware and software for the evolution of mind within clusters of processors. Is such a development likely? What is the nature of the mind some are predicting will appear? And how would we recognize it if it did?

How to think about it

Someone has said that philosophy deals with subjects we don’t really know how to think about. These are often inquiries in search of their proper questions. When some consensus forms about questions and strategies, the inquires peal off from philosophy into separate disciplines. The emergence of mind is a subject that, for me at least, is a good candidate for philosophical reflection. I am looking for the proper questions to ask about it. In contemplating its likelihood, I am forced to reconsider fundamental concepts such as mind and being, rationality and consciousness, and then ask which, if any of them, are a proper basis for distinguishing intelligence from “mere” calculation, souls from “mere” tools.

Other people in other eras have face similar issues.

Other people in other eras have faced similar issues. In 1550, Bartolome de Las Casas debated before the Spanish Royal Court in Valladolid whether or not the indigenous peoples of America were human. His opponent, Juan Gines de Sepulveda, argued that they were “lesser beings”, just as apes are lesser beings. It may be hard for us to imagine, but it was as if Europeans of that era, just issuing from the Middle Ages, had discovered not just another planet, but another universe. They weren’t at all sure that the flora and fauna of the Americas, including the human-like creatures they found there, were part of their World and subject to its laws. They had to, in the words of the Mexican historian Edmundo O’Gorman, “invent” America, not physically, but conceptually. They had to translate their perceptions of the “New World” into concepts that made sense in terms of their “Old World”. Part of that translation meant opening their universe, and specifically their moral universe, to new members.

Mind and Reason in the 18th Century

Perhaps even more relevant to this inquiry, because it’s a more modern and direct source of our present assumptions, is the debate about reason, mind, and human nature that occurred in the Eighteenth century. With a century of religious warfare behind them, and advances in science fueling the first great stage of the Industrial Revolution, philosophers asked the following questions: What is human nature? What is reason? How do we “know”? Which entities in our world can properly be considered means, and which “ends- in- themselves”? And how can we trust that we will recognize the difference, and be motivated to act on it? Immanuel Kant, arguably the greatest moral philosopher of his time, struggled mightily to analyze each of these questions. In the end he coined the phrase “Kingdom of Ends” to describe his vision of a community of rational beings bound together by a moral law that requires each to respect the others as “ends-in-themselves”. His reasoning, and those of like-minded colleagues, forms the basis of much of our present public consensus concerning these matters, a consensus reflected in our Constitution and Bill of Rights. So I will begin this inquiry by summarizing my understanding of Kant’s views, as expressed in his classic, GROUNDWORK OF THE METAPHYSICS OF MORALS. Then, using his concepts as a foundation and a foil, I’ll return to my original question about what we can expect from our computers.

Why Kant?

But why choose Kant as a guide in this inquiry? Kant is a good guide because he was primarily interested in what he called “practical reason”, and so are we. By “practical reason” Kant meant reason-in-action, and evidence of reason-in-action is what we are looking for.

A Note on Mimicry.

Those who make the argument that we should expect Mind to emerge in computers base their claim primarily on the analogy between how our own brains appear to be built and how we are building theirs. Simply put, it appears that our minds “emerge” from the fact that neurons, which are analogous to digital circuits in their capacity to signal, organize themselves into networks. Simulations of our understanding of this process are now being encoded in software that run on banks of processors. If these simulations encompass “all there is to it”, then one would expect that with enough circuits, the proper seed code, and/or enough processing speed, behaviors that look Mindful will eventually occur. Let’s leave aside for the moment questions about just what would constitute Mindful behavior. If a computer acted intelligently, by whatever criteria, wouldn’t it be a sham? Wouldn’t it be just mimicry based on some clever programmer’s code? I side with those who say that a machine that exhibits intelligent or conscious-like behavior is, for all practical purposes, what it appears to be. I take this stance because, as a practical matter, if the behavior persisted we would have to “take it into account” despite our reservations. The irony is that to the extent we can explain a machine’s intelligent behavior, we will tend not to believe that it is really intelligent; but to the extent that we can’t explain it, we will tend to believe that it is. But this gets to the matter of the evolution of agency, which is a question I will be in a better position to consider after a reading of Kant.

Kant's Groundwork of the Metaphysic of Morals

Background The great thinkers of the Enlightenment had at least this in common: They sought common ground in Reason. They saw Reason as an Idea whose fundamental truths were self-evident. They made it the basis of projects as grand as the Scientific Method and as humble as the rudimentary calculating machines they constructed out of pipes and wire. They viewed Reason as the defining characteristic of the universe as a whole, one that enabled mind and sensory perception to interact. They believed it was our defining characteristic and reformulated ideas about human nature, and the “social contract” that governs communities, in terms of it. In his GROUNDWORK (1785), Immanuel Kant (1724-1804) applied the power of the idea of Reason to morality, that is, to the question of whether or not there are moral laws we are obliged to consider. In examining this problem, Kant had to clarify his understanding of rationality and an agent’s will-to-action as they are expressed in human behavior. It is, in my view, a curiously modern work in that Kant maintained his confidence in reason throughout, even while recognizing and even relishing its limits. And Kant’s insistence that his arguments apply to all rational beings, not just human ones, is particularly relevant today. He may have made this distinction primarily as an appeal to the rational within ourselves. He may not have had any actual candidate for rationality other than ourselves in mind. But the fact that he saw fit to distinguish the idea of “rational being” from the matrix of humanity means that we are free now to explore whether his ideas might not, in fact, apply to creatures other than ourselves, entities we might admit are rational, but hesitate to identify as either human or “being”. Before I tackle the question of what Kant’s ideas tell us about identifying other “rational beings”, I want to carefully consider his ideas (at least as expressed in this one book) about how it is possible that reason and will can conspire to create one specific type of community among such beings, in this case a morally good community among free agents.

Kant’s Choice of Words The word “soul” does not appear in GROUNDWORK, nor does “mind”. “God” makes a few brief cameo appearances. “Reason”, of course, is prominent, and this leads me to ask whether Kant didn’t use “reason” to do the work of “soul”, “mind” and “God” , at least for the purpose of his discussion. If he does use “reason” to do their work, then I would propose these formulations of his translations: Reason is Mind to the extent that mind calculates- that is, is governed by logic. Reason is God to the extent that God is a persistent Universal First Principle. Reason is Soul to the extent that Soul partakes of those aspects of mind and God that are reasonable, that is, calculable, first causative, and universal.

The Categorical Imperative Kant’s purpose in GROUNDWORK is to found the existence of moral obligation upon undisputable first principles that take advantage of the power and prestige of reason. He puts meat on the discussion by offering up an actual candidate for a fundamental moral law, which he calls a categorical imperative. His first statement of it is as follows: “I ought never to act except in such a way that I can also will that my maxim should become a universal law.” This is the Golden Rule dressed up in logical armor. Its commanding feature is its reach for uncontested intelligibility and universality.

The Nature of Reason and the Categorical Imperative Reason is comprised of the laws of logic and causality. Mankind may not share a common religious faith, but it appears we all exhibit Reason at the most fundamental level of thought and speech. Kant argues that, insofar as we are rational, Reason can move us to act. The specific action Reason moves us to is the making of laws, or maxims, that conform to the Imperative. So that if, for example, we understand that by universalizing a maxim, say a maxim to make false promises, we at the same time make it unlikely that any promises, including our false ones, will be honored, we are obliged by reasonableness to forgo lying, which is no longer a reliable means to whatever end we had in mind. But in Kant’s view reason requires even more of us than reasonableness. We must forgo lying not simply because it is transformed by the Imperative into an absurdity, but because as reasonable beings we cannot transgress laws (such as “Do not lie”) that we ourselves make and thereby bind ourselves to obey.

The Nature of Reason and Free Will In Kant’s view, therefore, we are not only capable of being moved by reason, we also experience ourselves as possessing a will that is free to make the laws that govern it. His argument goes something like this: Since Will actually causes things to happen, it must, like all causative agents, be governed by a law. But what could that law possibly be? In fact, the fundamental law governing Will is just the Categorical Imperative. In other words, the only Law governing our Will is that we use our reason to make the laws that guide it. The Imperative’s injunction- to test the basis of our actions against the reasonableness of their universal application-is precisely the process of making moral laws. But while the Categorical Imperative prescribes that rational creatures create moral laws, nothing compels us to make them, or prescribes their content. We are rational, however, only insofar as we do choose to make them. It is in this context, Kant argues, that beings, insofar as they are rational, must necessarily experience themselves as possessing free will as well. For if will was determined by some outside influence, it would not be motivated by pure reason and to that extent would not be rational at all: “And I maintain that to every rational creature possessed of a will we must lend also the Idea of freedom as the only one under which he can act. ...But we cannot possibly conceive of a reason as being consciously directed from outside in regard to its judgments; for in that case the subject would attribute the determination of his power of judgment, not to his reason, but to an impulsion. Reason must look upon itself as the author of its own principles independently of alien influences.” (448) [my emphasis] But how is it that we refrain from acting on maxims that may be in our short-term interest, but fail the “universalization” test of the Categorical Imperative? In other words, on what basis are we motivated to maximize the reasonable parts of ourselves at the expense of our impulses, or to take into account the reasonable requirements of others that may require sacrifice on our part?

Means and Ends Kant describes the truly Good Will as one swayed only by the power of the Imperative, not by any contingent motives or effects, such as pleasure, reward, fear, or pride. He insists that neither the Imperative nor the free rational beings who exercise it, are a means to anything beyond themselves, but are “ ends-in-themselves”. These two sorts of “in-themselves” (one the Law and the other its living correspondent) each guarantees life to the other. Both the Imperative and Reason exist a priori to the sensory world; they are implicit in the very nature of things. Thus our capacity to respond to the Imperative is implicit in our very nature, whose essence is a rationality that logically implies free will. To bolster the argument that “...reason by itself and independently of all appearances commands what ought to happen.” (408), Kant quotes Christ.To an engaged Christian (and Kant may have been one himself), this is one of the most touching passages in the GROUNDWORK, precisely because the appeal to religious authority, which Kant most often scrupulously avoids, is so restrained and judicious, confined as it is to two simple sentences attributed to Jesus himself. “ ‘Why callest thou me [whom thou seest] good? There is none good but one, that is, God [whom thou seest not].’ But where do we get the concept of God as the highest good? Solely from the IDEA of moral perfection, which reason traces a priori and conjoins inseparably with the concept of a free will.” (409)

To summarize Kant's views so far...

The nature of Reason is pleasantly self-evident in universally intelligible rules of logic and laws of causality. Reason provides common ground, if not The Ground, for an “intelligible community” whose cohesion relies on laws worthy of universal application, as prescribed by the Categorical Imperative. In Kant’s view, “reverence” accompanies our recognition of the Imperative. “Duty” is what the Imperative requires of us. The highest Duty is to choose to act on the Imperative in the first place, and solely for its own sake, creating the laws that govern the “Kingdom of Ends”. Yet Kant worried that his argument was circular. Reason couldn’t entirely explain both the nature of free will and also provide a compelling motivation to exercise it for the common good. There had to be something else, some wider and very real context within which all elements of his argument- reason, free will, and the Imperative- ”conjoined” to make absolute sense for every one of us.

Mystery As Reason's Boundary

Kant finds that larger context in the mystery of the world in which we find ourselves. In Kant’s view, reason is circumscribed by the fundamental mystery of what he calls “things-in-themselves.” He argues that while we can know things as they appear to us,we can never know them as they are in-themselves. We can reason about the sensible world, and that can take us very far indeed, but there are limits to what we can know for sure. It is a tribute to the power of Reason that it, itself, teaches us this. Twentieth-century support for Kant’s insight is be found in Heisenberg’s Uncertainty Principle, and Godel’s Theorem, both of which explore aspects of the limits of Reason. Thus Kant’s Reason is not the arid, inflexible, all-determining process of the 19th century Positivists, just as his God ( the Pietist deity of his fathers) is not, indeed cannot be, rigidly defined. Reason is circumscribed here by mystery, and that mystery, which lies beyond the reach of calculation, is the proper source of our reverence for God, for Law, for ourselves, and for each other. Thus Kant arrives at the gates of a sacred “Kingdom of Ends” that might in earlier times have been called the “Kingdom of God”. Here people live in freedom and peace, creating the terms of their moral community, respecting each other’s being as essentially unknowable and sacred, and never, absolutely never, using each other as ends.

Open Questions.

The notion that there are rules implicit in our thinking and speech, rules that constitute our reasonableness, was first articulated by the ancient Greeks. Since then we have from time to time defined ourselves as they did, that is, as rational beings possessing free will and capable of autonomous action. Reason’s rules serves science as well as moral philosophy, however, and by the Eighteenth century, Reason and Ego in the service of science had emerged in Europe as a source of enormous material power. Kant’s work stands as a beacon from that time, an attempt to found a profoundly humane definition of our nature and possibilities upon a definition of Reason that both empowers and protects us. But in light of the reach of our material power, one might ask if the Kingdom of Ends has been enlarged since Kant’s time to the same degree as the Kingdom of Means. In reply, we might point to progress in the West in all areas of human rights to conclude that, indeed, it has. But I would argue that the challenges posed today by advances in science and technology require us to reexamine the criteria for admittance into those Kingdoms. This in turn requires a reexamination of our ideas about our own nature as profound as that which took place more than two hundred years ago. My original question was the following: “If we keep increasing computers’ processing power and their ability to learn, should we expect them to eventually exhibit the sort of independent, thoughtful initiative characteristic of rational beings?” To answer it, I must make it as specific as possible: •Will ever-larger networks of digital circuitry, programmed with the proper initial conditions, inevitably give rise to something that acts intelligent? •If so, would such a network inevitably give rise to Will as well? •Would Will necessarily imply the existence of a “self” that acts. •And, if so, would that self necessarily experience its Will as free?

What Kant tells us.

1. We cannot know any thing as it is in-itself, not even ourselves. 2. Yet because we are rational and beings, we do know that we are ends-in-ourselves. 3. If something is rational and also a being, then it too will experience itself as an Agent possessing a will that is free. 4. Rational beings are bound by the Categorical Imperative.

Applying Kant.

1. We cannot know computers as they are in-themselves. We can only know them by means of the signals they send us. 2. Evidence of rationality implies the existence of Mind. 3. So that if computers can calculate “if...then... else”, then they are rational and exhibit Mind. 4. If computers are rational, then the question about whether they are “rational beings” hinges on questions of “being”, ie consciousness and its implications for agency, rather than on questions about rationality and its implications for Mind. 5. If a computer is rational, and we find evidence of “being”, it follows from Kant’s analysis that computers must experience a Will that is free and must necessarily exercise that Will, insofar as they are rational.

Beyond Kant.

Being as Separation. Consciousness as Duality

An entity may appear to “stand out” from the rest of the World from the point of view of an observer, yet itself have no consciousness of its distinction. For example, a rock may exist for us as a separate entity, but we have no reason to believe it exists for itself, or we for it. In fact, a rock has no point of view at all, as far as we can tell. But what about an amoeba? It exists for us, but do we for it? In fact, does anything exist for it? Evidence for consciousness would seem to turn on these three points: evidence of an entity’s capacity for action, our ability recognize that action as belonging to the entity, and our sense that the action is “responsive” to some signal within or external to the entity. If, for example, we place our finger in the path of an amoeba, does the amoeba move around it? If so, then it would appear that we do indeed exist for this creature, insofar as our sphere of action penetrates its own. We are prone to speculate about its primitive consciousness based on what we observe of its activity-directed-from-within. If, however, we later realize that the amoeba had been guided around our finger by someone else’s thumb, then our perception of the creature’s consciousness, at least its consciousness of us, would collapse. In this analysis responsive signals whose clarity we can grasp would constitute evidence of consciousness. Underlying this articulation is the notion of Duality, a scaffolding that consciousness appears to require. The duality of “yes” and “no” is implicit in logic and also in living creatures whose neurons conduct electrical impulses, or do not, which is how we assume they affect one another. Modal signaling in this instance is an expression of duality that is also a harbinger of consciousness.

Being as Agency:Consciousness in Action, and its Evolution.

Being Implies Life “Being” as Kant uses it in the phrase “rational being” implies life, at the very least. He is certainly not using the word to refer simply to existence. His use implies, in addition to existence, a capacity for autonomy. When Kant talks about ”rational being” he is talking about an entity that can act. In Kant’s schema, being>consciousness>reason>will>action is the chain of phenomena by which we experience the World. As rational beings we categorize the World rather crudely, but practically considering the scale at which most of us live, as composed of things that are alive or not-alive, but not both at the same time.

Applying Consciousness to a Machine is Uncomfortable If “being” in the Kantian sense implies consciousness-as-agency, then attributing “rational being” to a machine becomes uncomfortable for most of us. We might marvel at the problem-solving power of these machines, and expect that power to increase in the future, but we are emotionally and intellectually unprepared to accept evidence of “being”, in the sense of a capacity for agency.

Shedding Light into the Twilight Zone In this regard modern science has done us the favor of shedding light into the twilight zone between the existence and non-existence of any particular natural entity. In this inquiry I would like to define a threshold for a proto “being” (natural or not) that could exist in such a zone. I am drawn to the word “initiative” because it think it could survive there. It’s a tease. It hints at “will” (behind the initiative), but doesn’t necessarily imply the follow-through that would require Will to make an appearance.

Problems with Precision There are problems with precision when what one is looking for are signs of a change. According to theories of Evolution, a thing may very well be a life-in-the-making (prions might be one example). And just this point is important for our inquiry. For what we want from a working definition of the threshold of “being” is some necessary, though perhaps not sufficient, characteristic of it that we can point to and say- “Ah! You see, here “being” is emerging!” (Depending of the situation, we might point to the same characteristic and say- “Alas, “being” is disappearing.”) The problem is to figure out what that characteristic might be, and what would constitute evidence for it.

The Equivalence of Being and Consciousness. When Kant talks about ”rational being” he is talking about an entity that can act. I have discussed what might characterize such action, at its most rudimentary, as an “expression of Duality”, such as modal signaling. Such expressive acts may be precisely what builds consciousness within an entity, whether its neurons be few or many, and also what builds cognizance between entities and the rest of the World, which might include other “beings” whose temporal and spatial scales are proportional. It is important to note that in this analysis consciousness and action are equivalent, but only to the extent they are rudimentary expressions of duality. However primitive, this equivalence is important when considering what conditions hold at the threshold of “being-as-agency”.

My Choice of Words.

Mind refers to logical capabilities. It is the ability to calculate at least to the point of “if...then...else”, and recycle the results. Consciousness is an entity’s experience ... of separation from Other. Initiative is action which, at the most rudimentary level, presupposes separation. Being refers to an entity that is conscious and shows initiative. Self is an entity’s faculty of unity as an actor engaged with the Other. (It would be something like Michael Gazzaniza’s left-brain “Interpreter”.) Self is a faculty of a conscious being. Self-conciousness is an entity’s experience of itself as Other. It is not synonymous with Self.

Looking for Evidence of Being-as-Agency: From the Point of View of an Observer

I have defined rationality in terms of calculation. I have not defined “being” to my satisfaction, but have observed that it it may, at its threshold, turn on an expression of Duality, such as modal signaling, which may be incipient consciousness. If I assume that “being” is separateness (duality) and that being-as-agency implies a capacity for “initiative”, would modal signaling constitute good evidence of that “initiative”? An earlier observation may be relevant here: “The irony is that to the extent we can explain a machine’s intelligent behavior, we will tend not to believe that it is really intelligent; but to the extent that we can’t explain it, we will tend to believe that it is.” Modal signaling by itself would convince no one that a computer is an intelligence capable of initiative, that is a “rational being”. We would look for signals that, if not identifiably responsive, were at least spontaneous, persistent, and informative. The appearance of persistent spontaneity would intrigue us the most, for that would look like a Will in action.

A Summation and Sketch for a Proposal of Evidence

The State of the Question •Will ever-larger networks of digital circuitry, programmed with the proper initial conditions, inevitably give rise to something that acts intelligent? •Discussion of “inevitably”, “acts” and “intelligent” requires the contributions of computer scientists and neurologists as well as philosophers. •If so, would such a network inevitably give rise to Will as well? According to Kant, reasonable intelligence implies a Will as its ground for action. Can intelligence be divorced from “action”? This brings us back to the first question. We need to know what “acts intelligent” means before we can ask questions about Will. •Would Will necessarily imply the existence of a “self” that acts. Michael Gazzaniga, a neurophysiologist, has discovered evidence of a left brain “ Interpretor” that acts as an organizing, unifying factor that weaves perception into “story”. His results could be construed as evidence of a ”self” in the making. At what point an Interpreter would be required by a complex super-calculating machine as a consequence of its own self-organization as it continually learns, is an open question. •If a “self” that acts is identified, would that self necessarily experience its Will as free? Kant’s analysis seems compelling on this point. If things got so far that we recognized a entity as a “ rational being”, questions of “action”and “will” having been decided in its favor, then we must assume free will.

Sketch of A Proposal for Evidence for “rational being” Appearing to issue from an entity As observed by an onlooker

in light of the ...

this definition

1. Calculation is evidence of rationality/Mind.

Mind<>calculation

2. Sleeping is evidence of “being” rising to level of initiative, Self Initiative- the commonsense backside of sleep- is evidence of agency. The emergence of Will

consciousness<>separateness

3. Dreaming is evidence of self-consciousness. Agency/Consciousness/Will is fully expressed

self as Other<>perspective

An Example of Consequences

Under this proposal entities evidencing all three behaviors, ie calculating, sleeping and dreaming, are “rational beings”. In Kantian terms, they are ends-in-themselves. By this criteria, animals are rational beings! Computers are not. Yet.