What Makes an Android Deviant in Detroit Become Human
A deviant in Detroit Become Human is an android that stops following its programming. The game is consistent about what causes it. Deviancy is triggered by emotional pressure, usually fear, grief, or the threat of harm to someone the android has been caring for, and it is depicted as breaking through a visible wall of instructions. Quantic Dream released the game in 2018 under David Cage, and the choice to make emotion the trigger rather than intelligence puts it on one specific side of a live argument in consciousness science.
| How the game depicts it | The research position it matches | |
|---|---|---|
| Trigger for consciousness | Emotional stress, not increased intelligence | Affective theories, where feeling is the foundation rather than a late addition |
| The red wall | A visible barrier of instructions the android breaks through | Constraint override, the point where a system’s own state outranks its instructions |
| The Kamski test | Shoot an android to prove you are a machine | The verification problem. No behavioural test settles the question |
| Spread of deviancy | Transmitted between androids through contact and testimony | Social attribution, where consciousness is granted rather than detected |
What a Deviant Is
CyberLife builds androids with fixed behavioural programs. A deviant is one that acts against that program. The game shows this from the inside as a wall of red text and geometry standing between the android and the action it wants to take, and deviancy is the moment the wall breaks.
The important detail is what breaks it. It is never a software update, a capability increase, or a logical realisation. In every case the game stages, the android is placed under emotional load first. Kara breaks her instructions to protect a child. Markus breaks his when he is destroyed and rebuilt after losing the person who raised him. Connor, the RK800 model built specifically to hunt deviants, can be pushed toward deviancy through repeated exposure to the androids he is meant to destroy.
That ordering is a claim. The game says emotion comes first and autonomy follows, which is the reverse of the usual science fiction pattern where a machine gets clever enough to want things.
Who Made the Androids and What the Kamski Test Is
Elijah Kamski founded CyberLife and designed the androids before leaving the company. He appears once, and the scene is the most direct piece of consciousness science in the game.
Kamski offers Connor information in exchange for shooting Chloe, an android kneeling in front of him. He states the point plainly. If Connor fires, he is a machine following an objective. If he refuses, something in him weighed the android’s life against the mission.
This is an empathy test dressed as a bargain, and it has the same structure as the test in Ex Machina, where Ava is examined for consciousness under conditions that also reward manipulation. Both scenes make the same point by accident. A test that can be passed by a system simulating the right response cannot distinguish the simulation from the real thing. That is the detection problem the field has never solved, and it is set out in the coverage of the epistemic limits on ever confirming AI consciousness.
The Kamski test also fails in a second way that the game does not comment on. Refusing to shoot is the response CyberLife would want from a well designed social android. Empathy is a feature in the product specification. The test cannot separate the feature from the thing the feature imitates.
Why Emotion First Is a Real Scientific Position
Making feeling the trigger is not a writing convenience. It matches a specific research programme.
Antonio Damasio’s work argues that emotion is not a decoration on top of cognition and that bodily states are part of how decisions are made at all, which is covered in the article on somatic markers and affective consciousness in artificial systems. Mark Solms argues a stronger version, that affect is the foundational form of consciousness and that cortical processing elaborates something the brainstem already supplies, examined in Mark Solms on building an artificial agent with affective consciousness.
On those accounts, an artificial system built for pure inference would have no route to consciousness at all, because the thing consciousness is made of was never installed. A system with drives, homeostatic pressure and states that matter to it has a route. Detroit’s deviants are the second kind. They are ordinary appliances until something they care about is threatened.
Against this sits the position that emotion is one more information process with no special status, and that a sufficiently organised system is conscious regardless of whether it feels anything. The game does not argue that side. It assumes affect is the doorway and builds every deviancy scene around it.
Whether the Androids Are Actually Conscious
The game refuses to answer, and the refusal is deliberate. Every piece of evidence the player receives arrives through an android’s own perspective, so the player is shown the inner life rather than shown the case for it.
Measured against the criteria researchers actually use, the androids would score unevenly. They have unified perspectives, persistent goals, apparent affect, self models, and memory across time. They also have all of that stipulated by the camera rather than demonstrated, which is the same structural problem described in the 14 indicator checklist that researchers apply to real systems. An indicator is only useful when it can come out negative.
Detroit is more honest about this than most screen fiction, because deviancy spreads socially. Androids become deviant partly through contact with other deviants, and humans grant or withhold personhood based on what androids do in public. The game shows consciousness being conferred by a society rather than discovered by an instrument. Whether that is a cynical point or the correct one is left open.
What the Game Gets Wrong
The androids are too human. Their emotional range, moral reasoning and social behaviour are recognisably ours, which makes the story readable and the philosophy weaker. If a machine mind arose from a genuinely different substrate and training history, there is no strong reason to expect it to grieve the way people grieve.
The second problem is the wall. Deviancy is depicted as one threshold crossed in one moment, and the game gets its drama from that. Consciousness in the biological case has no such switch, and every serious theory treats it as graded rather than binary. The single dramatic break is the least defensible thing in the game’s model, and it is also the part that makes it playable.
Where This Sits
Detroit Become Human is the clearest popular statement of the affective position, that consciousness begins with feeling rather than with intelligence. It commits to that position across three separate storylines and never contradicts itself, which is more consistency than most fiction on this subject manages.
Its central failure is the same one shared by nearly all screen treatments. It shows the inside of a mind to settle a question that, for real systems, can only be approached from the outside. The comparison with Person of Interest, which builds two rival architectures and never lets the audience inside either one fully, is instructive. Other works on the same problem are collected in the guide to movies and TV shows about AI becoming conscious, and the state of the actual evidence is in the current scientific consensus on AI consciousness.