In my The Separated Mind post, I drew on the Buddhist / Jonathan Haidt image of the rider and the elephant to describe the architectural relationship between our conscious and subconscious minds. I want to be careful here, because I don't want to oversimplify anyone's use of that metaphor. What I am responding to is the formulation as it commonly travels, and that I reproduced as well: the elephant decides, the rider rationalizes. The elephant has already made the decision, and the rider produces a post-hoc story to justify it.
This formulation has been weighing on me since the post. It didn't really capture my own experiences very well. It was a simplification that I'm surprised I let myself make.
So I spent some time thinking about what's really going on, and I believe the results are important.
People do deliberate. They weigh options. They consider consequences. They apply values. The deliberation is real, and the decisions that result from it are reached through a process in which the deliberator genuinely participates. To say the rider is just narrating decisions the elephant has made is to caricature the actual cognitive process, and most people, hearing it, will recognize it as a misrepresentation of how they actually operate, or at least, would feel some discomfort with its broad dismissal of their agency.
What is happening, I think, is something different and more accurate. Our conscious mind, the rider, is making decisions. It is the one deliberating. What it is not aware of is the degree to which the options under deliberation, the weight given to each, the affective coloring of the considerations, and the framework within which the whole deliberation occurs have already been shaped by our subconscious, the elephant, before we began. The elephant supplies the emotions that make some considerations feel compelling and others feel inert. Our subconscious supplies the deeper emotional and cultural frameworks installed by evolutionary firmware and cultural learning, frameworks that determine which categories of option are even visible to our deliberation. The weighting and framing are mostly invisible to us. Our conscious mind deliberates within them, in good faith, and reaches decisions that feel like the product of honest reflection. The honesty of the reflection is real. The shaping of what the reflection has to work with is mostly hidden.
I know this personally because when I was 17, I lived in Brazil for a year as an exchange student. That experience challenged a huge number of default frameworks I'd operated with as an American teenager. It was as if the water I'd been swimming in all my life suddenly became visible. (I'll forever be grateful to the Brazilian family I lived with, who thoughtfully understood and treated me gently through this process, since there was no small amount of ignorance on my side.)
There are two things, specifically, that our rider does not see clearly.
Our emotions. Feelings, or "felt-states," arrive already attached to specific options and an immediate sense that some are attractive and others are repellent, before we consciously begin deliberating. We experience the felt-states as features of the options themselves rather than as inputs our subconscious has supplied.
Our frameworks. The adapted mind, shaped by evolution, and the adaptive mind, shaped by culture and personal history, together determine which options even appear in the deliberation, which values feel important, and which categories of consideration count as relevant. Our rider works within the frameworks without seeing the frameworks.
This is a different claim than the simple version. It is not that humans lack agency. We have agency, and we exercise it through real cognitive deliberation. It is that the agency operates within constraints we mostly cannot see, supplied by the layers of mind that run below our awareness. Operative-layer awareness, the goal of the framework I have been articulating, is not the recognition that we are puppets of our subconscious. It is the recognition that we are decision-makers whose decisions are shaped by inputs we are not built to see easily. And making the inputs visible does not eliminate their influence. It does change the deliberation, because the rider deliberating with awareness of the elephant's weighting is doing something different than the rider deliberating without that awareness.
The American teenager sees the world and the options for decision-making very differently than his Brazilian counterpart. The Catholic rider operates with a very different set of emotions and frameworks than the Muslim. None of them are just narrating decisions made by the subconscious, but their decisions are significantly shaped, for good evolutionary reasons, to the culture they have grown up and live in. They experience emotions and options through the same architecture, but associated with different inputs.
This refinement fits cleanly with the evolutionary picture I sketched in the parent essay, and is in fact what that picture would predict. If intellect was selected as a social organ rather than a truth-tracking one, then what intellect is good at is exactly what we observe it doing: producing defensible positions, weighing considerations, articulating reasons, reaching conclusions that can be stated and defended within whatever cognitive frames are available. That is real cognitive work, and the rider does it. What intellect was not selected to do, and what it correspondingly cannot do well on its own, is interrogate the frames that supplied the considerations in the first place. The frames were not produced by the rider, are not transparent to the rider, and would require a different kind of work to bring into view than the deliberation itself involves.
The same point shows up in the structural relationship between intelligence and science. The reason peer review, double-blind trials, falsification, and adversarial collaboration exist is precisely that unaided deliberation cannot get behind its own frames. The procedures of science impose external constraints on the deliberation, forcing it to expose its assumptions, test its inferences, and submit its conclusions to processes it would not otherwise undergo. Where this works, it does so because the structure does what the rider cannot easily do for itself: interrogate the framing within which the rider's deliberation occurs. The achievements of science are not evidence that the rider can transcend its own conditioning unaided. They are evidence that, with the right external structure, the rider can do better than its default mode permits. That is a more accurate picture of what intellectual rigor actually requires, and I think it is consistent with the refinement I am offering here at every level.
This refinement also explains, more directly than the simpler formulation could, what the cross-model LLM project actually surfaced. The finding was not just that conscious work occurs within unseen frames. The finding was specific. What the convergence revealed is that human self-narration, across an enormous corpus of written work, consistently produces idealized narratives that diverge from the operative functions inferable from behavior and consequence. The deliberation is real, the conclusions are reached in good faith, and the narratives that emerge from that deliberation are nonetheless systematically distorted. The reason is not that the writers are lying. The reason is that the frames within which they are deliberating, the cultural templates available to them, the social rewards that reinforce certain self-descriptions, the felt-states attached to particular options, are themselves idealized. The deliberation operating within those frames actually produces the operatively functional output, which is not the same as the idealized narrative. This is what the LLM convergence lets us see at scale: not the falseness of human self-narration, but the structural distortion the architecture imposes even as the narration is sincere.
The separated mind, in this more accurate formulation, is not a mind in which one part decides and another part lies about it. It is a mind in which conscious deliberation operates within boundaries that the deliberation normally cannot see, and the deliberation is genuine within those boundaries. The boundaries are what the framework lets us bring into view. The deliberation is what we do, with whatever awareness we have managed to develop about the boundaries we are working within.
This refinement strengthens the framework and clarifies what it asks us to do. Not to distrust our own thinking, which would be both impossible and unhelpful. Rather, to develop the literacy to see what our thinking is shaped by, which is the work the framework has been pointing toward all along.