Four Theories That Try to Explain Why You Like What You Like

The deeper I go into the sciences, the more I realize how hard it is to build a theory that actually works. Beauty is a good example.

Not a theory that sounds good — plenty of those exist. But one that makes predictions specific enough to be tested, general enough to be useful, and honest enough about where it fails.

Psychology has produced four major theories of aesthetic preference, each attempting to explain why we like what we like. And each one captures something real about the experience — while also breaking down in ways that are just as instructive as the theory itself.

The four theories are mere exposure, arousal dynamics, prototype theory, and fluency. Together they cover a lot of ground. None of them covers everything.

First: what makes a theory scientific?

Before diving into the theories, there's a concept from Karl Popper that changes how you see all of them: falsifiability.

For something to count as scientific, it needs to make predictions that could, in principle, be shown wrong through observation. Not that they will be wrong — just that there exists some possible outcome that would disprove the theory.

Popper contrasted Einstein's general theory of relativity — which made a very specific prediction about how light should bend near massive objects, later confirmed during a 1919 solar eclipse — with Freud's structural model of the psyche (id, ego, superego). Freud's model is interesting, maybe even insightful. But it doesn't make predictions specific enough to be disproven. Whatever happens, you can retrofit it into the framework.

Fortune cookies work the same way. "Serious trouble will bypass you." If nothing bad happens, the cookie was right. If something bad happens, well, it bypassed you — you just didn't notice. There's no way to falsify it.

This matters for aesthetic theories because some of them suffer from exactly this problem. They're flexible enough to "explain" any outcome after the fact, but they don't predict anything specific enough to be tested.

But first: are people objectivists or subjectivists?

Before getting into why we like things, there's a prior question: do people think there's a right answer to aesthetic questions?

Researchers tested this by presenting statements like "Shakespeare is better than Dan Brown" and asking whether the statement was an opinion or an objective fact. They manipulated whether the domain was aesthetic (art, music), moral, or factual, and asked in several different ways.

The finding is that people are consistently subjectivists about aesthetics. When asked directly, they concede that aesthetic judgments are opinions, not facts. This challenges Kant's claim that we make a "claim of universal validity."

But there's a gap between what people say and what they feel. Intellectually, they take the subjectivist line. But once someone's taste gets personal, the feeling of objectivism creeps back in. You feel right, even while knowing that, in principle, it's subjective.

The studies also revealed that how you frame the question matters enormously. Compare Shakespeare to a five-year-old's drawing instead of to Dan Brown, and suddenly people shift toward objectivism. Make the comparison less abstract and more personal, and the "claim of universal validity" resurfaces.

Theory 1: Mere exposure — you like it because you've seen it before

The mere exposure effect , coined by Robert Zajonc in 1968, is the simplest of the four theories: the more you encounter something, the more you tend to like it.

No reasoning required. No deeper appreciation developing. Just raw frequency of exposure driving preference.

One of the strangest demonstrations of this was at Oregon State University, where a student attended class for an entire semester wearing a black bag over their body. At first, other students were hostile. Then curious. Then friendly. By the end, classmates were defending the black bag person. Mere exposure turned a genuinely bizarre stimulus into something familiar and even liked.

Real-world examples are everywhere:

• Your laptop's default wallpaper — you didn't choose it, but after months you don't want to change it
• Canvas (the learning software) — at first it looks ugly, but after a while it becomes visually familiar enough to feel normal
• Music your parents played when you were young — you grow up hearing it, and now it's your music too
• Fashion cycles — fanny packs went from embarrassing to cool in ten years, driven largely by repeated visual exposure
• Buying things you love, then losing interest — the trend exposed you to the item everywhere on social media, but once the trend fades, your external exposure drops even though you still own it

Cutting's art exposure study

James Cutting (2003) applied this to art. He found that the paintings people consider "great" Impressionist masterpieces are disproportionately the ones that appeared most frequently in art history books. People prefer the paintings they've seen more often in print.

The confound is obvious: maybe the better paintings got reproduced more because they're better. Exposure might be a consequence of quality, not a cause of preference.

Cutting tried to address this with a manipulation: he showed participants high-frequency paintings once and low-frequency paintings four times, essentially reversing the exposure pattern. This "cancelling out" approach is clever, but it raises questions about whether short-term lab exposure can really counteract decades of cultural familiarity.

Where mere exposure breaks down

The theory has a fundamental problem: infinite regress. If you like something because you've been exposed to it, then why were you exposed to it? Often because someone else liked it. But why did they like it? Because they were exposed to it. You never get to the original cause.

Mere exposure can explain the perpetuation and amplification of a preference, but it can't explain where the preference came from in the first place. Something had to be appealing before the exposure cycle started.

Also: mere exposure has limits. There are things you can see a thousand times and never like. Some things start out mildly negative and grow positive with exposure. Others start positive and become less appealing with overexposure (think of a Cocomelon song on repeat for weeks).

Does mere exposure make bad art worse?

Philosophers tested this using Thomas Kinkade — wildly popular with the general public, despised by art critics — versus John Everett Millais, whose work carries much more canonical prestige. Over a seven-week period, they manipulated how many times students saw each artist's work.

The result was mixed but interesting: repeated exposure reduced liking of Kinkade, while it didn't reliably increase liking of Velázquez. This suggests that mere exposure might interact with quality — repeated exposure to something simple or "low-brow" can make it more annoying rather than more appealing, while complex art might already be at a ceiling.

This matches a pattern many people recognize intuitively: complex music that you initially don't enjoy can grow on you over months of listening. But a simple, repetitive pop song can become maddening with overexposure. The theory doesn't account for these quality-dependent variations well.

Theory 2: Arousal dynamics — the Goldilocks zone of stimulation

Daniel Berlyne (1971) proposed a more complex theory: aesthetic pleasure depends on arousal, and arousal follows an inverted-U curve.

Too little stimulation → boring. Too much stimulation → overwhelming. Somewhere in the middle → peak pleasure.

Berlyne identified three types of variables that drive arousal:

The theory proposes two competing neural systems:

1. A reward system that activates as arousal increases — this generates positive affect
2. An aversion system that kicks in when arousal gets too high — this generates negative affect

The reward system saturates before the aversion system, which is what creates the inverted-U shape. Pleasure rises, peaks, then falls.

The Yerkes-Dodson parallel

This pattern shows up everywhere in psychology, not just aesthetics. The Yerkes-Dodson Law describes the same curve for performance: a little stress helps you perform better, but too much stress hurts performance. Drug effectiveness follows a similar curve. Even ecological biodiversity shows this pattern — moderate disturbance promotes the most species diversity.

Where arousal dynamics breaks down

The theory is hard to test across individuals because different people have different optimal levels of arousal. What's boringly simple for one person might be pleasantly engaging for another. A novice looking at abstract art might be overwhelmed where an expert finds it perfectly stimulating.

This makes the inverted-U prediction hard to falsify in practice. If you find someone who prefers high complexity, you can always say they have a higher optimal point. If someone prefers low complexity, their optimum is lower. The theory accommodates almost any individual result.

There's also a range restriction problem in the studies that have tried to test this. When researchers use representational paintings and classical piano solos as stimuli, they're sampling from things that already made it into museums and recordings — things that are already in the aesthetically pleasing range. If the inverted-U exists but you're only sampling from the middle, your data will look like a flat line or a weak trend, not a clear curve.

A better test would use the full range: from a Cocomelon jingle to avant-garde experimental music, from a child's crayon drawing to a Kandinsky. Nobody has really done this convincingly yet.

Expertise shifts the curve

One part of this theory that does feel deeply right is the way expertise shifts the sweet spot. A musician who's been studying complex jazz for decades needs more complexity to hit their optimal arousal than someone whose most demanding musical experience is pop radio. The curve doesn't change shape — it shifts rightward with expertise.

This explains something that other theories struggle with: why experts and non-experts systematically disagree about what's "good."

Theory 3: Prototype theory — you prefer the most typical example

Rooted in Eleanor Rosch's (1975) research on categorization, prototype theory claims that we prefer stimuli that are the most typical examples of their category.

When you think "dog," you picture a generic medium-sized dog — not a Chihuahua, not a Great Dane. That mental image is the prototype. And prototypes, according to this theory, are inherently preferred.

This has been demonstrated across multiple domains:

• Colors — people prefer prototypical red over atypical reds
• Furniture — the most typical-looking chair is rated as most attractive
• Faces — averaged faces (as Galton discovered) are rated as more attractive
• Surrealist paintings — even in abstract art, more prototypical examples within the genre are preferred

Where prototype theory breaks down

The biggest problem is that supermodels aren't prototypical. If averaged faces are the peak of attractiveness, then the most beautiful people in the world should look maximally average. But they don't — they have exaggerated features (larger eyes, thinner jaws). The actual peak of facial attractiveness goes beyond the average.

Also, some things that are clearly not prototypical are deeply loved. Avant-garde art, experimental music, spicy food in cultures that didn't traditionally eat it. Prototype theory can't explain why people develop preferences for the atypical.

Theory 4: Fluency theory — you like what's easy to process

The most general theory is simple: people prefer stimuli that their brains can process easily.

Processing fluency is the sense of ease (or difficulty) you experience when perceiving, interpreting, or categorizing something. High fluency = pleasant. Low fluency = unpleasant.

Fluency theory is elegant because it can explain the other theories:

• Mere exposure works because repeated exposure makes processing easier
• Prototype preference works because prototypes are faster to categorize
• Even some arousal dynamics can be reframed in terms of fluency — things at the right complexity level are processable without being boring

Where fluency theory breaks down

Its biggest strength is also its biggest weakness: it's too flexible. It can explain almost any result after the fact, which means it doesn't make strong falsifiable predictions. This is exactly the Popper problem.

If someone prefers simple art → "they prefer fluent processing." If someone prefers complex art → "conceptual fluency from expertise makes complex art easy for them." Whatever happens, fluency theory can accommodate it.

More specifically, fluency theory predicts that preference should decrease monotonically with complexity — simpler is always better. But empirically, that's not what happens. People don't prefer the simplest possible stimuli. They prefer an intermediate level — which is what Berlyne's arousal theory predicts, not fluency theory.

Fluency theory also struggles with meaning. A painting can be easy to process visually but deeply meaningful — or easy to process but completely vapid. The theory doesn't distinguish between these cases.

The Gaussian Girl effect: when fluency can predict either outcome

There's a revealing example that exposes fluency theory's flexibility problem. In research on the "cheerleader effect," blurred faces were compared to sharp faces. Before seeing the data, fluency theory could predict either direction:

• Blurred faces are easier to process (less information, simpler) → should be preferred
• Blurred faces are harder to recognize (ambiguous, uncertain) → should be less preferred

The actual result: blurred faces were rated as more attractive (the so-called "Gaussian Girl" effect, which may help explain the appeal of Instagram filters and soft-focus photography). Fluency theory can explain that result. But that's exactly the problem. If blurred faces win, the theory can say less visual noise made them easier to process. If sharp faces had won, it could just as easily say clearer features made them easier to recognize. Once a theory can move comfortably in either direction after the fact, it stops making a strong prediction and starts sounding like commentary on whatever already happened.

How the four theories relate

These aren't four competing explanations where only one can be right. They're more like four perspectives on the same phenomenon, each highlighting a different mechanism:

The Thomas Kinkade test: applying all four theories at once

Thomas Kinkade is the perfect stress test for these theories. He was wildly popular with the American public — reproductions of his paintings appeared everywhere, from mall galleries to mugs and puzzles. Yet he was widely dismissed by art critics as low-brow, saccharine, and uninteresting.

How would each theory explain this split?

• Mere exposure explains the popularity: his images are everywhere. You've seen them without even trying. That repetition feeds liking.
• Prototype theory explains why the images feel "right" to many people: they match the prototype of a pleasant landscape painting — warm light, cozy cottage, peaceful nature. It's the prototypical American painting.
• Arousal dynamics might be the most interesting explanation for the expert/non-expert split. For the average viewer, Kinkade hits the sweet spot of complexity. For art critics whose arousal curves have shifted rightward through years of expertise, his work falls below the optimal complexity threshold — it's boring.
• Fluency theory explains the general appeal because the images are easy to process, but it can't explain why experts find them less appealing when they should be even more fluent for expert viewers.

No single theory handles both the popularity and the critical disdain. But arousal dynamics, with its expertise-shifted curves, comes closest to explaining why the same stimulus produces opposite reactions in different populations.

A few things I'm taking away

• Falsifiability is the minimum bar for scientific theories — if a theory can explain any outcome, it doesn't really explain anything
• Studies consistently show people are subjectivists about aesthetics when asked directly, but the feeling of objectivism returns when it's personal — how you frame the question matters
• The mere exposure effect is real and powerful but can't explain where preferences originate — only how they amplify
• Your parents' music becomes your music partly through mere exposure, creating a "second reminiscence bump" for songs from their era
• Berlyne's arousal dynamics capture something true about the Goldilocks zone of stimulation, but the inverted-U is hard to pin down because different people have different optimal levels
• Prototype theory explains baseline preferences beautifully but can't account for why the most attractive faces and most celebrated art often exceed the average
• Fluency theory is the most general framework but borders on unfalsifiable — its flexibility is a scientific weakness
• None of these theories alone explains aesthetic preference — they're complementary lenses, each capturing a piece of a complex phenomenon
• The ugliness-beauty scale might not be a single dimension at all — like sweet and sour are both taste but not opposites, beauty and ugliness might be separate experiences that can coexist

And the one I keep coming back to: the theories that sound most satisfying are often the ones that are hardest to disprove. And a theory you can't disprove isn't really doing the scientific work you need it to do. That tension between explanatory elegance and testable specificity is what makes this field so challenging — and so interesting.

Sources:

• Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, 9(2), 1-27. Used for the core mere exposure effect.
• Berlyne, D. E. (1971). Aesthetics and psychobiology. Appleton-Century-Crofts. Used for arousal dynamics, the inverted-U, and collative variables.
• Rosch, E. (1975). Cognitive representations of semantic categories. Journal of Experimental Psychology: General, 104(3), 192-233. Used for prototype theory and category structure.
• Reber, R., Schwarz, N., & Winkielman, P. (2004). Processing fluency and aesthetic pleasure. Personality and Social Psychology Review, 8(4), 364-382. Used for the basic fluency theory account.
• Palmer, S. E., Schloss, K. B., & Sammartino, J. (2013). Visual aesthetics and human preference. Annual Review of Psychology, 64, 77-107. Used for the critique of fluency theory, especially the complexity problem.
• Cutting, J. E. (2003). Gustave Caillebotte, French Impressionism, and mere exposure. Psychonomic Bulletin & Review, 10(2), 319-343. Used for the art-exposure example in Impressionism.
• Meskin, A., Phelan, M., Moore, M., & Kieran, M. (2013). Mere Exposure to Bad Art. British Journal of Aesthetics, 53(2), 139-164. Used for the Kinkade versus Millais repeated-exposure section.
• Chatterjee, A. (2014). The Aesthetic Brain: How We Evolved to Desire Beauty and Enjoy Art. Oxford University Press. Used as general background on aesthetic preference and beauty research.
• Popper, K. (1963). Conjectures and Refutations: The Growth of Scientific Knowledge. Routledge. Used for falsifiability and the scientific-theory framing.