Original title: Inferring Affective Consciousness in an Artificial Agent: A Case Study
Mark Solms, St John Grimbly, Bruce Bassett, Evert Boonstra +6 · September 2026
Many scientists believe animals have feelings if they seek out places with pleasure-giving substances that lack food value, like morphine or cocaine. Researchers assume this preference cannot just be an unconscious reflex. This paper tests that assumption using a simple computer agent. The authors created a program that acts like an emotional system. It tracks uncertainty about its internal needs compared to resources in its environment. Through this seemingly subjective way of handling data, the artificial agent develops the same place preference for pleasure. Yet, the agent remains completely deterministicOperating in a completely predictable way based on fixed rules, with no randomness or free choice., meaning its code follows fixed rules. This shows that such behavior does not prove real feelings exist, raising questions about the physical roots of consciousness and the feeling of free will.
You pick up a candy bar even though you are not hungry, simply because you enjoy the taste. Now, imagine you could not tell if your choice was driven by a genuine craving or just a series of automatic habits. In this paper, the artificial agent behaves as if it has such a craving, but its actions are entirely programmed, which makes us question what real desire actually is.
This changes how scientists think about what feelings really are, affecting studies of mind and artificial systems. If machines can mimic emotional behaviour without being alive, it questions how we define consciousness in humans and animals.
The abstract does not explicitly state limitations. It presents a case study based on a simple artificial agent, so the findings are theoretical and may not directly apply to complex real-world systems or prove consciousness in machines.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Inferring Affective Consciousness in an Artificial Agent: A Case Study — Mark Solms, St John Grimbly, Bruce Bassett et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.