# The Submarine The best image for an emerged self is not a ghost floating free of physics. It is a banquet inside a submarine. The submarine is causally bound to the ocean. Pressure, temperature, depth, and current matter. The vessel cannot ignore them. But once the hull is closed and the internal machinery is running, the banquet inside unfolds according to its own local order. The guests talk, argue, remember, plan, desire, and revise. The ocean constrains the banquet without determining the content of the conversation. That is what increasing selfhood looks like. The loop remains fully physical and fully coupled to the world, but more of its next state is determined by its own internal imprints and less by immediate external forcing. This decoupling remains a matter of degree. A continuum can still exhibit threshold-like regimes, and highly organized loops will often feel qualitatively different from shallow ones. But the difference is not a leap into a second substance. It is a shift in the balance between outer forcing and inner steering. We can express the idea abstractly. Let future behavior \(b_{t+\Delta}\) depend on the loop's current internal organization \(m_t\) and on current environment \(u_t\): $$ b_{t+\Delta} = H(m_t, u_t). $$ Then the degree of decoupling over a chosen timescale \(\Delta\) and behavioral class is the degree to which variation in \(b_{t+\Delta}\) is better explained by \(m_t\) than by immediate fluctuations in \(u_t\). No single formula is forced here. The point is operational: a system is more self-directed when its future is increasingly steered by its retained organization rather than by the latest push from outside. This makes room for values. A value is not first a moral sentence. It is a stable imprint that biases future selection. Hunger is a value in this minimal sense. Safety is a value. Social approval, truth, beauty, loyalty, and revenge can all become values once they are carried as stable internal constraints that steer future behavior. At low depth, a loop is mostly pushed around. At greater depth, it begins to reconfigure itself in light of what it carries inside. At still greater depth, it can plan, veto, sacrifice immediate reward for delayed coherence, and alter the niche in which later loops will form. That is what it means for a self to steer not only its next state, but part of its own evolutionary future. The submarine image also clarifies why the self should not be confused with the surface body. The hull is necessary. Without it, the banquet is flooded. But the banquet is not identical to the steel. Likewise, the loop depends on a body and carries an imprint of body, yet the self is the organized steering that the body supports, not the body considered as a heap. This gives a more precise answer to the question of emergence. A self emerges when a causal loop becomes rich enough that its internally consulted imprints dominate more and more of its future behavior. The stronger that internal dominance, the more the entity behaves as a center of agency rather than as a mere subsystem in the hands of the environment. The hardware question remains open at this stage. Later chapters can return to biological machinery, including resonant candidates such as microtubules and the broader evidence that learning need not wait for brains. But the structural point is already clear: the world does not need to stop acting on a system for a self to emerge. It is enough that the system becomes able to steer more and more of its future from within.