Wanting vs. Liking

In the AI Story

The distinction required neuroscientists and clinicians to abandon an intuitively satisfying idea: that addicts use drugs because drugs feel good. They don't. Or more precisely, the continued compulsive use that defines addiction is driven not by pleasure but by wanting that has been decoupled from pleasure. The addict at the bar at two in the morning is responding to cues — the smell, the location, the emotional state — that the sensitized dopamine system has tagged with overwhelming motivational urgency. The reward, when it arrives, is often disappointing. The disappointment does not reduce the wanting.

Three decades of experimental evidence have confirmed and extended this framework. Robinson and Berridge's 2025 retrospective in the Annual Review of Psychology, marking thirty years of incentive-sensitization theory, documents a research program that has survived every major challenge. The finding holds across species, across substances, across experimental paradigms. Wanting and liking are separate. They can be separated. And the separation is the neural architecture of compulsive behavior.

Applied to AI tools: the prompt-response architecture constitutes a near-optimal activation system for the mesolimbic dopamine pathway. Each prompt is a cue predicting reward. Each response delivers variable magnitude. The speed of the cycle maintains dopaminergic activation at levels natural environments almost never sustain. The liking system, meanwhile, depends on different conditions — effort, mastery, embodied engagement — that the AI workflow structurally eliminates. The result: escalating wanting paired with static or declining liking, predicted by Berridge's framework with the reliability of a chemical equation.

The distinction illuminates productive addiction with diagnostic specificity. Edo Segal's description in The Orange Pill of nights when "the exhilaration had drained out hours ago" but "the grinding compulsion" continued maps onto the wanting-without-liking state with textbook precision. The exhilaration is liking. Its departure leaves pure wanting — the motivational drive to continue, indifferent to whether anything satisfying remains to be gotten.

Origin

Berridge's 1989 experiments at the University of Michigan tested what should have been a settled prediction. If dopamine was the pleasure chemical — as every textbook implied — rats whose brains had been depleted of dopamine should show diminished pleasure responses to sweet tastes. They did not. The facial hedonic reactions (rhythmic tongue protrusions, relaxed expressions) were preserved. What vanished was the motivation to pursue the sugar. The rats sat beside food and starved. Berridge spent the next three decades demonstrating that this dissociation was not an artifact of extreme dopamine depletion but a window onto the fundamental architecture of reward.

The framework was formalized with Terry Robinson in their 1993 incentive-sensitization theory of addiction. The theory predicted — and subsequent research confirmed — that repeated drug exposure produces two opposite processes simultaneously: tolerance (diminished liking requiring higher doses) and sensitization (amplified wanting in response to drug cues). The addict does not simply enjoy the drug less. The addict wants the drug more, and the wanting has become independent of the liking.

Key Ideas

Separate neural substrates. Wanting is mediated by mesolimbic dopamine; liking by opioid and endocannabinoid hotspots in the nucleus accumbens shell and ventral pallidum. These are different circuits with different neurotransmitters.

Dissociability under stress. The systems can be pulled apart by drug sensitization, by chronic stress, by environments that amplify cue-driven wanting while suppressing mastery-driven liking.

The signature of addiction. Compulsive pursuit of rewards that no longer produce pleasure is not weakness or moral failure. It is the predictable output of a wanting system that has decoupled from the liking system that would ordinarily terminate pursuit.

Indistinguishable from the inside. The wanting system does not label its outputs as distinct from pleasure. Wanting feels like enjoyment. The distinction becomes visible only in the afterglow — or its absence.

AI as structural decoupler. The prompt-response loop activates wanting at maximum while eliminating the friction, effort, and mastery that activate liking. The architecture produces the dissociation as a byproduct.

Debates & Critiques

Some neuroscientists have argued that the wanting-liking distinction overstates the functional separation of systems that ordinarily cooperate, and that the dissociation is visible only under experimental extremes irrelevant to everyday motivation. Berridge's response, extended across decades of papers, is that the everyday experience of coupling masks a latent separability that becomes clinically consequential wherever environments amplify one system while suppressing the other — from drug addiction to eating disorders to, now, the AI compulsion loop. The 2025 retrospective treats the framework as robustly vindicated across species and paradigms.