Liking vs Wanting in Food Reward Psychology

Anyone who has finished a bowl of chips and then wondered, mid-crunch, why they kept eating has bumped into a gap that psychologists have spent decades trying to name. The eating felt automatic. It happened before the person noticed wanting it, and only afterward came the puzzled thought: I didn't even really need that. That gap, between what a food does to us and what it makes us do, is the territory Professor Graham Finlayson has spent his career mapping. He is a professor of psychobiology at the University of Leeds, and his research looks at human appetite, food reward, and the mechanisms that push some people toward losing control over how much they eat (04:26).

Finlayson's first move is to break apart something most people treat as one thing. When someone calls a food rewarding, they usually mean it is pleasurable, full stop. Finlayson's research splits that pleasure into two separable pieces.

"Liking is how pleasurable we actually experience the food to be, the sort of hedonic impact, it's sometimes called. And then another component of reward, which is separable and separate from liking, is the extent to which we want it." — Graham Finlayson [07:02]

Liking, in this framework, is the felt pleasure of a food. Wanting is the motivational pull, the force that grabs attention and drags a person toward choosing and consuming something, independent of how much they actually enjoy it. Most of the time the two move together: what people like, they also want. But Finlayson has found specific situations where they split apart. People with a binge eating phenotype, meaning a consistent pattern of loss-of-control eating episodes, show unusually strong wanting for combinations of high-fat and sweet food, and it is that wanting, not their liking, that best predicts how much they overeat (10:20). A subtler version of the same split happens after a big meal. Eating suppresses hunger, and it also reduces the pleasure of the specific food just eaten, a phenomenon called sensory specific satiety (11:17). Liking drops. But wanting can linger, a kind of residual pull that keeps predicting further eating even after the pleasure has faded.

Wanting is also harder to study, because it resists a basic research tool: asking. People can rate how much they like a food without much trouble. Wanting is different.

"Wanting is much more difficult to, I think, get your head around as a consumer in everyday experience. So it's not as clearly something you can introspect on." — Graham Finlayson [13:16]

Because wanting runs partly below conscious awareness, Finlayson's lab measures it indirectly, by timing how quickly people respond when choosing between pairs of foods rather than asking them to rate their desire on a scale (14:28). That reaction-time approach turns out to be a strong predictor of which food a person will choose, though it is a less reliable predictor of exactly how much they will eat once they start (15:43).

Why a nutrition label is not the whole story

The most striking part of the conversation concerns a 2023 paper Finlayson published, called Food-Level Predictors of Self-Reported Liking and Hedonic Overeating. To build it, his team started with a database of 60,000 foods and narrowed it to a representative sample of 300, chosen to span the range of nutrient and sensory variation in a typical diet (18:40). That let them test broad patterns across an entire food landscape rather than drawing conclusions from a single lab meal.

The team asked participants not just how pleasant each food was, but also to rate what Finlayson calls hedonic overeating: the extent to which a food's sheer pleasure would lead someone to eat too much of it (19:39). Then they tested what actually predicts that outcome. First they built a model using only hard nutrition numbers: energy density, protein, fiber, and similar values. That model captured about 20% of the variance in how much a food was liked (21:59). Then they added perceptual and sensory characteristics, meaning how the food is perceived rather than what it is chemically made of, and the added explanatory power jumped by another 40% to 60% (22:35).

"You can understand how much the food is liked to a certain degree from its nutritional characteristics, but you can understand much more if you combine those nutritional characteristics with more kind of psychological, perceptual sort of aspects." — Graham Finlayson [22:35]

The implication is not just methodological. It suggests that liking, and the risk of hedonic overeating, is not baked into a food the way its calorie count is. It is a property that emerges from the food, the eater, and the setting, all at once (24:29). The same slice of cake can be irresistible at a birthday party and forgettable eaten alone at a desk. Beliefs about a food, cued in advance, can shift how pleasurable it actually turns out to be (26:01).

What simple labels hide

That finding sets up the episode's sharpest turn: a direct test of the NOVA classification, the widely used system that sorts foods into groups by degree of industrial processing, with "ultra-processed" as its most scrutinized category. Ultra-processed foods are often blamed for driving overeating through what researchers call hyperpalatability, an unusually strong rewarding quality. Because Finlayson's team already had detailed nutrition and perception data on hundreds of foods, they could test that claim directly, running the analysis across three separate large survey studies. Once nutritional properties were already accounted for, NOVA's ultra-processed label added almost nothing: between 0% and, at its strongest, 7% of additional variance in hedonic overeating (30:12). A related idea, that an unnatural, roughly even ratio of carbohydrate to fat makes food unusually rewarding, fared even worse, adding only 0% to 2% (30:44).

"Ultra-processed food and hedonic overeating... actually added very, very little additional predictive variance in our analyses." — Graham Finlayson [30:12]

Finlayson is careful not to dismiss the broader link between ultra-processed diets and poor health outcomes, which shows up consistently in observational data. His objection is narrower and, in a way, more useful: a label that lumps enormous numbers of foods into one bucket can hide the actual mechanism causing harm.

"The danger is that in looking at huge numbers of groups of food under one label or one classification, you're losing a lot of the information that can tell you about mechanism." — Graham Finlayson [32:50]

A simple category is persuasive. It is easy to put into a public health message. But if the goal is understanding why a food leads someone to lose control over their eating, a single label trades away exactly the detail needed to answer that question.

Finlayson closes by pointing to what he still wants to know: whether it is possible to design foods or eating situations that preserve the enjoyment of eating without also triggering the motivational pull that leads to overeating (38:09). It is a fitting question from someone whose whole body of work argues that pleasure and compulsion, while they usually travel together, are not actually the same trip. Asked for one small daily habit, he mentions something almost domestic by comparison: writing in a notepad by his bed most nights, working through what went well, what did not, and what he might carry into the next day (39:38). It is a small, deliberate act of separating experience from automatic reaction, which is more or less the whole argument of his research, just done with a pen instead of a reaction-time test.