Can a machine, lacking reflection, be a Person?

Question

We are well beyond Frankenstein and the experience that the machine—“it’s alive”! As we continue to rely on Suri’s for GPS directions, “self”-checkout aisles, or the artificial intelligence of robotic vision and other digital trends, philosophers cannot help but to address the ontology and selfhood of automation. It seems clear that machines can function as selves, in Kant’s full sense of understanding and determinate judgments laid out in the Critique of Pure Reason.

We can analogize Kant’s analysis of human cognition and the synthesizing of the “sensible manifold” with the “content” and “form” of computer programming, for example. What distinguishes human experience from machines, aside from the flood of our emotions, feelings, and spontaneous anticipations, lies in the power of reflective judgment. Reflection is the hallmark of human experience according to Kant’s Critique of Judgment and all our determinate judgments presuppose it; otherwise how would we be aware of our actions? In the history of philosophical idealism the difference between the self and person stems from this distinction between cognitive determination and reflection. So far as we know, machines lack this power of reflection. We have read or seen examples of it in science fiction and the movies, but is it possible that super-computers or other forms of hyper-technology can perform reflection in the way Kant describes which is representative of persons, who have their own teleological purposes or wills?

Answer 1

If physicalism is true and determinism (at least on a Newtonian level) is true, then yes, a machine theoretically could be designed in a way you describe and thus considered a "person".

Phsyicalism as you may know is the idea that (stated most simply) "everything is physical". Determinism is the idea that everything has a cause; nothing is uncaused. In other words, given the way things are at time t, the way things progress thereafter is fixed as a matter of natural law. Under these views the human mind and all human behavior is just the result of a causal chain of prior physical occurrences in the brain. This, in fact, brings us to the realization that the brain itself is a machine; sure it's not made of metal and wiring but the same kind of complex feedback loops and processing occurs.

At this point, we've thus already answered your question. If the human brain is itself a machine, then of course we think machines can be seen as a "person". But if you insist that by "machine" you mean something other than man, then still, yes, it is theoretically possible to replicate the human mind using non-organic parts as long as you maintain the same logical feedback loops and the like. The content/structure (brain matter, cells, chemicals vs. metal circuits, electricity, etc.) does not matter in a deterministic system, only the logic they produce matters.

Answer 2

A person is an emergent thing. Can a man built machine be called emerged maybe.

By emergent I mean flexible, adaptive, in relationship to whom it's interacting with. Like a bell. It sounds different when you strike it differently. But compared to a bell a person changes shape and evolves over time.

Answer 3

A soul is essential for a sense of self/self-awareness, which is essential for self reflection.

If a machine (in the electro-mechanical sense) can have a soul, a distillation of the essence of the Universe (a true and larger Self), then my answer is 'yes, it can be considered a person'. If that can't happen, then my answer is 'no, not even if all the systems and feedback loops of the organic body and mind are replicated'.

Note: Other than my own interpretations, I don't have any references to material that supports my answer. If anybody is able to provide supporting arguments, I'd be much obliged.

Answer 4

When I read this question, I can't help but thinking of the Turing Test, in which a judge must tell which the machine is and which the human. Both reside in different rooms, and contact each other by text, which the judge can see/hear. If he fails to do so, the machine is considered indistinguishable from a human being.

The machine clearly lacks reflection, as it only reacts to the given input in a human-like manner. The logic on which it is built defines its answer, instead of a thinking process: For exactly the same input, with exactly the same conditions, a machine will always produce the same output (even with random numbers, as these are generated with some logic corresponding to some input).

Therefore, in theory, it is possible to have a machine which acts indistinguishable like a person, which leads to the idea that a machine can indeed be a person.

Answer 5

This assumption usually strikes those who equate human intellect (soul) with human brain and reduce human brain to a machine.

This is rooted in materialist philosophy that rejects human soul having an independent, incorporeal substance. Mulla Sadra, the most prominent Muslim meta-physician of the 16th-17th century, firmly established that human body in its evolutionary process is guided and perfected by a cosmic immaterial Macro-Intellect. Human soul is a substance granted by the Macro-Intellect to human body after it is sufficiently developed in the womb. But the soul works in close interaction with the body (especially the brain), however it exists independently in the spiritual realm. Consciousness and reflection stem from human soul (micro-intellect), however, with involvement of human brain (which is in fact the instrument of the intellect) as long as man has not departed natural life. Therefore, life and consciousness are in essence incorporeal qualities of the incorporeal Macro-Intellect that emanate from it, permeate the organic body, dissipating down into the constituent physical elements.

Under this light, a machine in essence can never be a conscious reflective body comparable to human beings. Because for it to be conscious and reflective it has to be originated by the incorporeal intellect that governs and leads beings in the natural realm to the point of independent level of consciousness. A humanoid robot is only a combination of physical entities that work together and sensually interact with the external world based on the laws of the physical world. For them to become truly conscious they must receive incorporeal qualities from the incorporeal intellects that originate the body from its very early states of natural origination onward but that's not how a robot develops.

Here's a succinct exposition of Mulla Sadra's theories on human nature, development and afterlife for those interested in this less-famous but uniquely brilliant muslim philosopher: The Soul as Barzakh:Substantial Motion and Mulla Sadra’s Theory of Human Becoming by Maria Massi Dakake.

Answer 6

Convergence: what you experience and what you predict or think you know come together. When a machine can experience and theorize its own behavior in a way that converges, it will be a person.

Legally, a person has to be able to give consent, thus there are laws having to do with people being of age to make decisions, being in their right mind during a crime, etc. A Corporation can be said to assent to something or decide under some circumstances, so legally it functions as a person (can be fined, etc), but that does not mean that a Corporation is a Self. Both are needed to fit our understood meaning.

Answer 7

The "Re..." words are certainly the catch in machined knowledge, but I would argue for "Reproduction" through "Recognition," as opposed to "Reflection."

It is interesting that you cite Kant here, for his transcendental philosophy is, in my view, both the best critique of reductive physicalism and potentially the best set of instructions for assembling fundamental concepts as integrated circuits into artificial intelligence and even, perhaps, self-identified "personhood."

All the "parts" are there in the transcendental kit, just open up the "intuiting" valves and add sensible flows. A good candidate for generating "reflection" are the weirdly unpredictable, pulsing patterns of electromagnetic oscillations produced by video cameras observing themselves, recording recordings.This might potentially bring in a very fine-grained type of mutual "recognition" or symmetrical "disequilibrium" in machine relations.

To me, the greatest drawback to machine consciousness and personhood is finding the logical counterpart of love, by which I mean "reproduction" through "recognition" in the Hegelian sense. One capacity of brains, far too neglected in physical accounts, is that they self-select and reproduce themselves. Here, Von Neumann's Universal Replicators provide a mathematical description of "immortal" machine genealogy requiring overproduction, errors, and selection. It also requires a "software" and "hardware" interacting dualism. But this is still way too simple.

To attain personhood machines must not only recognize a "categorial instruction" they must see themselves as interchangeable "ends-in-themselves" which is a paradox. They must endlessly regenerate and resolve "paradoxes," including the fundamental paradox of gender... and some machine equivalent of in utero or a priori conceiving: i.e., the state in which two "persons" identify as "one" and then "one" person develops into "two" persons in "one," finally separating into a conceptual unity of "three" persons, the minimal unit of regeneration.

I say this to emphasize the degree to which paradox and irrationality must enter into the machine logic for anything approaching "personhood" to emerge. Kant does provide a promising blueprint. And he was unique in his day for even a modest mention of sexuality. But in my own view, most approaches to AI underestimate this necessary substrate, those capacities of practical and aesthetic judgment that "brains" must include to selectively reproduce "brains."

Answer 8

A dualist would very likely answer that it isn't possible that a machine can acquire personhood. This would be because the self-awarness/reflection part stems from the non-physical aspects of the mental which we cannot instill in an artificial machine no matter how well engineered it is.

From a strictly materialist point of view, Hofstadter's concept of strange loop, and self-representational and higher order theories of consciousness would allow for a machine to reflect upon itself. Broadly speaking, these types of theories assert that conscious arises (or is possible) in systems which have the ability to monitor themselves, or store and process representations of themselves. If a system's repertoire of symbols is complex enough that it can store and manipulate symbols corresponding to itself, then it should be capable of self awareness. All the issue comes down to is the level and complexity of representation that the system is capable of.

Where Kant would diverge from those who adopt the the self-representational theory of consciousness, is that he sees our self-awareness, or reflexiveness, as leading to free-will, whereas the self-representational models don't require or imply freewill at any point. Indeed many from that school believe that we don't have freewill in the traditional sense at all, they are all either compatibilists or hard determinists.

Answer 9

I would strongly disagree that reflection is the hallmark of human experience. Most of our life is spent in non-conscious states. We behave mostly by habit, and subconscious activity. Definitely, we do spend some time on conscious, reflective states, and this is... imo, how we get our conceptual knowledge.

Specifically, why can't machines "reflect"? A machine performs an action expecting a certain a result... the result doesn't happen, so the machine looks its logs to determine what action it took, and how the action could be improved. Is this not reflection?

Answer 10

The difficulty with your question is that we do not understand how self-reflection arises from the biological processes at work in a human being, and their brain in particular. Not knowing the answer to that question makes it impossible to say with any degree of certainty whether or not self-reflection can be realized in a non-biological system.

Searles, for example, argues for something that he tentatively calls "biological realism", in which consciousness (and therefore self-reflection) is a biological phenomenon like digestion or photosynthesis. The digestive process is inseparable from its biological/chemical mechanisms. A computer program can model the process to a large degree of accuracy but there is a difference between a model of digestion and the actual transformation of a hamburger into energy and feces.

Under this view, a computer program can model consciousness, but can't produce it any more than a computer model of a weather system can produce a real rainstorm.

Is the mind "hardware independent"? Is it like a computer program that can run on any physical system capable of interpreting the syntax of the program? Or is the biological hardware (which, as far as we know, is the only kind of system that has produced consciousness) intrinsic to how consciousness is realized?

As far as I know, there is no definitive answer to this question.

Answer 11

Kantian idealism, in its Copernican turn from the object to the subject already pre-supposes the existence of mind as a given; and the distinction between the two realms - objects as matter without mind, and subjects with mind.

Kant takes for granted that to have experience, there is something that experiences - in more modern language, it has qualia; he further takes for granted, an axiom that wouldn't be contested - it certainly isn't by Descartes cogito: I think therefore I (am a mind) that exists - this is the picture of the Averronian floating man; or again, much later, and picturesquely put into the argot of Science Fiction brains in a vat.