A single engineering result illustrates a deep distinction in the philosophy of mind between intelligence and its functional supports.
The Ghost in the Repetition
We ran an autonomous marketing agent across seven live campaigns. Each time, it was tasked with selecting a persuasion trigger, an emotional affect, and a Gestalt principle for the creative. The results were oddly consistent: across seven runs, it chose Authority Bias six times and Cognitive Ease five times. The immediate diagnosis was a failure in reasoning — a model stuck in a rut.
The diagnosis was wrong. Each run was a separate, stateless API call. The system had no memory of its previous choices. It wasn't repeating itself; it was beginning anew every single time, with the same inputs, and arriving at the same locally optimal conclusion. When we supplied the agent with its own recent selections and the current funnel stage, the behaviour diversified immediately. The agent wasn't reasoning badly. It simply couldn't learn from its own history because, from its perspective, it had no history.
Functionalism and Its Fault Lines
This engineering outcome serves as a concrete illustration for a set of ideas from the philosophy of mind. In his PHIL3022 lectures, Dr. Bruce Long draws on Putnam-style multiple realizability to argue that intelligence is a functional property. Any system, regardless of its physical substrate (silicon, protein, or otherwise), that can functionally emulate the complex, information-processing patterns of human cognition is intelligent. There isn't some extra, non-physical quality of 'real' intelligence left over.
Long also draws a critical asymmetry between intelligence and phenomenal consciousness. A perfect functional emulation of pain behaviour — flinching, crying out, reporting distress — might still lack the subjective, first-person experience of pain. We can't say the same for intelligence. A perfect functional emulation of flexible problem-solving, generalization, and reasoning under uncertainty simply is that capacity. There is no further fact to be realised. This position has roots in the work of thinkers like Dennett and Papineau.
Our agent's behaviour maps directly onto this asymmetry. What the system lacked was not intelligence, but a functional precondition for exercising it effectively over time: episodic persistence. As Long’s argument suggests, if intelligence is precisely the functional capacity for reasoning, then removing a functional support (like memory) degrades the observable performance without implying the underlying system reasons badly within its operational constraints. The logic of each individual choice was sound; the failure was in the architecture that kept each choice informationally isolated.
An Illustration, Not a Proof
It is important to state the limits of this inference. The behaviour of a single marketing agent cannot settle a thesis in the philosophy of mind, a field with a rich history of debate around intentionality from Brentano and Stich to more recent work by Millière and Buckner. Our observation doesn't prove Long's position. But it does provide a compelling, concrete instance of it in action.
It suggests that when we evaluate these systems, we should be precise about what we mean by 'intelligence'. Is it the capacity for logic in a single moment, or the ability to string those moments together into a coherent, learning narrative? As Nemoto suggests, the binary of real versus apparent intentionality may be the wrong frame entirely, with a spectrum of 'continuous intentionality' being a more useful model. Our agent, before the fix, wasn't unintelligent. It was intelligent, but only for an instant.
Method note. This piece was built in NeuraScene's Advertising Mode using an explicit psychological frame:
- Persuasion trigger: The Halo Effect
- Target affect: Cognitive Ease
- Gestalt principle: Law of Similarity
The article establishes credibility through an intellectually rigorous but clear explanation (Cognitive Ease) of a complex topic. This rigor is designed to create a positive impression of the agency's expertise that extends to the brand as a whole (The Halo Effect). The chosen visual principle, Law of Similarity, directly represents the agent's repetitive choices and the subsequent, state-aware break in the pattern.
