By Junyi Liu
Late by seven years, I finally finished David Cage’s masterpiece, Detroit: Become Human. The animation and cinematics still hold up today, but as I put down my controller, I was especially fascinated by the ethical questions raised by the developers: What are these androids? How different are they from us humans? With the Greeks and Beyond SSHI program not being too distant a memory, Parfit’s Theory of the importance of Relation-R came up in my mind. For those who don’t know or remember, the theory of Relation-R states that continuous Psychological Connectedness is the criterion for whether two people are closely enough identical (Parfit 1989). In such a case, a person would technically live on forever by having replicas of themselves with this connection in memory, beliefs, values, etc. In the course, much of our discussion was oriented around physical transfers and fissions. While these thought experiments are interesting and will be further discussed, the game raises further questions about emotion and directives when judging identity using Relation-R, with the consideration of androids. Perhaps, in some cases, Relation-R might not be enough to make a judgment. Further, I will bring in another example of Alien: Romulus, the summer blockbuster last summer that raises similarly interesting questions with the android Andy.
Consumption of the game and movie would give a much better understanding of the contexts behind the following thought experiments, and I would strongly recommend both, though I’ll try to explain as thoroughly as possible the details behind each example for those who have no idea what happens in either story. There will be spoilers as well, so reader discretion is advised.
The year is 2038, and Detroit is a modernized city with androids everywhere that help private owners do their tasks, work on the street, and even in police investigations. Androids generally aren’t supposed to show emotion, although they are able to simulate it for social interactive purposes. For example, Alice, Todd’s adoptive child, is an android that tries to be a good companion for parents wanting the “perfect” kid. Then, the question becomes, is this simulation of emotion the same as human emotion? Is there a difference between simulation and actuality if the perception from a third person is the same (in reality, would it be the same? The Kamski Test, which I strongly recommend the reader look into, is a separate question of whether robots react differently than humans in terms of empathy)? Then, hypothetically, if all the physical attributes, memories, and “psychological connectedness” are uploaded into an android, with the only difference of simulated emotion vs human emotion, is it still the same person? What makes human emotion more real, if it even is? Of course, in the game, the emotions of androids are still quite jagged, so one can say that it’s not (which is also the point of the Kamski Test), but as androids progress and there’s no difference in perception, what would happen?
In fact, in the game, deviants start appearing, androids that disobey their owner’s commands, actually display emotion, and think for themselves beyond the directive. As is revealed later, androids become deviants when “they are overwhelmed by irrational orders”. For Luther, a butler robot who decides to save one of the three main characters, Kara (at least this is one of the possibilities if the player makes it to this path), he said that seeing Alice, the little girl, being maltreated, “opened his eyes” for the first time. No longer are his emotions bound, but he more freely emulates a human. Just considering the emotional aspect (the directive element will be later examined), the difference between simulated vs human emotion even goes away. Then, back to the hypothetical case of uploading consciousness, does what matters, or Relation-R, still hold true? Is there a discontinuity in the jump from human emotion to android emotion, or is it more of a mental block in us humans? I’d argue it’s the latter since every aspect of psychological connectedness is kept the same. Albeit, one can argue that physical connectedness, which isn’t conserved here because of the jump from human to android, is also part of the criterion (overturning Relation-R).
Androids can also be rebooted, which gives way to a continuation of the discussion at SSHI regarding identity. Let’s start with a trivial case. In one of the final scenes, two Connors face off in a weird form of combat. One of the options is for the “good” Connor to gain control over the “bad” Connor when it is about to die. In other words, the “good” Connor transfers its consciousness over to the other Connor. Does Relation-R hold true? Seemingly yes, so what matters is conserved.
Alternatively, if the android is wiped clean and rebooted, though it picks up fragments of memories from its previous past, and the people still treat it like before, is Relation-R still conserved? In the game, Kara experiences this when Todd, her abusive and psychopathic owner, kills her before and fixes her up, though she has lost most of her memories. However, her life is still pretty much the same as before, as is later revealed. Thus, the question becomes, how much psychological connectedness is required for effectual identity (Relation-R would be a hundred percent)? A different question we can ask is whether treatment by others is part of our identity. Say if Todd switched up his behavior from an abusive owner to a sober and reformed addict before Kara’s reboot, but after her death, would that change her psychological connectedness? While “yes” does seem like an attractive answer, since Kara’s life would be different, it would be weird for an outside factor to change something else’s identity. Perhaps we can reconcile it by saying that it is our reaction to outside treatment that is changed, which would be part of psychological connectedness.
Now, we pick back up on the Renaissance, the abandoned spaceship that the main characters of Alien: Romulus arrive on in the hopes of going to Yvaga III. Andy, the android, initially receives instructions from Rain, his owner. Formally, he has a “directive” from her, which is to be a protector and surrogate brother of sorts. He does a good job of this, fighting off the Xenomorphs, until the crew reactivates Rook, an incapacitated android on the Renaissance. He gives a new directive to Andy, that is to complete the research regarding the Xenomorphs for the sake of the Weyland-Yutani, the original crew of the ship that was massacred by the Xenomorphs. This changes Andy’s behavior majorly, now not as protective and kind to Rain, let alone the rest of the crew. In such a case, can we say that Relation-R is conserved (and what matters in identity)? Although retaining his past memories, this alteration in directive seems to be a change in the psychological continuity. This explanation would match up with the intuition of the crew in thinking that something weird changed in Andy.
Similar to the thought experiment of simulated emotion vs human emotion, if a normal human being’s directive, which can be thought of as its natural tendencies and biases, and the rest of Relation-R are uploaded into a humanoid, does identity still hold? Relation-R seems to be, but there is another mental block. Perhaps it is due to our uncertainty in knowing how well our tendencies will transfer over to a “directive”. Perhaps we think that having a directive will limit cognition. But aren’t human biases to the same effect (although we can argue that biases can change while “directives” are harder to change)? Even ignoring this technological gap, for outside perceivers, this change may definitely not feel like identity. Is there something else in humans that is lost during this transfer? The aforementioned physical connectedness that matters to us? Alternatively, we can also just say that Relation-R is not conserved. A directive is an extra command that the person had never had before, which is different from human tendencies.
In our SSHI course, Relation-R was also used as a potential argument to make us feel more comfortable with knowing that we will live on in passing over our psychological connectedness to another being. Yet, as science fiction slowly approaches reality, we are confronted with new questions of identity concerning artificial intelligence. Granted, AGIs and ASIs still are very much theoretical, conjectures regarding how emotions, directives, and other forces certainly complicate our understanding of Relation-R. Perhaps the criterion can be refined, though times may change our perceptions quicker than we think. For many, a change from human to android may seem unsettling and foreign. On the other hand, as Lieutenant Anderson said in the game, “Maybe androids aren’t as different from us as we thought.”
References
Alvarez, Fede, et al. Alien: Romulus. 20th Century Studios, 2024.
Detroit: Become Human. Directed by David Cage, Quantic Dream, 2018, PlayStation 4.
Parfit, Derek. Reasons and Persons. New York: Clarendon Press, 1989.
Quantic Dreams. Detroit: Become Human, https://www.quanticdream.com/en/detroit-become-human.