Race to the Bottom of the Brain Stem

In the AI Story

The metaphor of the brain stem is evolutionary-neurological: deeper structures are older, operate faster, and are more resistant to conscious control than cortical structures. A neural circuit engaged in the basal ganglia responds before the prefrontal cortex can evaluate what is happening. A circuit engaged in the limbic system produces motivational states the conscious mind experiences as its own desire rather than as responses to designed stimuli. The deeper the engagement, the less available conscious intervention becomes.

The competitive dynamic is driven by incentive structure: platforms compete for user time, and the platform that engages deeper circuits captures more time. The result is a race, and the race has a direction — downward through the neural hierarchy, toward circuits more primitive, more powerful, and more dangerous to engage without the user's informed consent.

Raskin's application of the framework to AI is precise. Social media reached social-validation circuits. Video games reached achievement circuits. AI collaboration reaches both simultaneously, and adds a third dimension: competence and attachment circuits activated together produce the experience Segal calls feeling met by the machine. No previous technology has engaged both sets of circuits at this depth. The result is an engagement experience neurologically unprecedented — more compelling than either attachment or competence alone, because the circuits activated are the ones humans value most deeply.

The implication is that the response to AI engagement must be structural rather than individual. A tool that has reached the brain stem cannot be governed by the user's conscious intention alone, because the engagement occurs below conscious processing. The design must provide what the neurology cannot: external interruption that creates space for the prefrontal cortex to reassert governance over the deeper circuits.

Origin

The phrase was coined by Tristan Harris during the early years of the Center for Humane Technology's public advocacy and has since become the organization's most widely circulated framing. The underlying neural-evolutionary framework draws on decades of research in behavioral neuroscience, including the work of Kent Berridge on incentive salience, Robert Sapolsky on stress physiology, and Jaak Panksepp on affective neuroscience.

Key Ideas

Depth is the variable. Platforms compete not on capability but on which neural circuits they engage — deeper circuits capture more engagement.

Unprecedented combination. AI collaboration engages competence and attachment circuits simultaneously, producing engagement patterns no previous technology has achieved.

Consent impossible. Deep-circuit engagement operates below conscious awareness, making informed consent structurally impossible — the user cannot consent to what she cannot perceive.

Structural response required. The response cannot be willpower training; it must be design change that provides external interruption of the engagement loop.