Looks Ripe to Me
When surface cues stop meaning what they used to, good decisions get harder to make.
A shorter version of this post was published on Psychology Today.
You can also hear AI Matt’s summary of the piece below.
Back in college, I worked in the produce department of a local grocery store. I spent a lot of early mornings unloading cases of apples, oranges, and every other produce imaginable from the back of a semi and then ensuring shelves stayed stocked throughout the day. It wasn’t glamorous, but it was oddly satisfying (and I had a terrific manager). Over time, I got good at something that felt like a quiet superpower: I could pick fruit with the best of them1.
Not all fruit, to be fair. Apples and oranges I was a bit meh at. But tomatoes? I had them down to a fine art. I could predict with eerie accuracy whether a tomato would be flavorful and whether it would be ready in one day, two days, or never. It was part learned skill, part intuition—and it held up pretty well into adulthood.
Until it didn’t.
Somewhere along the line, my tomato radar (my tomadar?) started giving me false positives. I'd pick a beautiful one off the pile—deep red, a seemingly perfect tomato—and cut into it later only to find... disappointment. Grainy texture. Mealy flesh. No flavor. The tomato looked ripe. It felt ripe. But it wasn’t. And it kept happening.
At first, I assumed I’d just lost my touch. But it turns out something else had changed: the tomatoes.
Turns out it wasn’t me that changed—it was the tomatoes.
Most of us use a handful of sensory cues—generally visual, but they may also include tactile, olfactory, or auditory (in the case of melons)—to decide whether a fruit is worth purchasing2. With tomatoes, it might be a certain richness of color, the slightest give when pressed, maybe even a smell. These are the heuristics we use to judge the quality of fruit.
And for most of human history, these heuristics worked. A red tomato on the vine really was ripe. A peach that gave a little when you squeezed it really was juicy and sweet. Our shortcuts matched the underlying biology of the fruit.
But not anymore.
In today’s grocery store, many of those signals have been decoupled from the thing they’re supposed to signal. Tomatoes are bred for durability, uniform color, and shelf stability—not for flavor. Some varieties have even been genetically tweaked to eliminate the green shoulders that used to indicate an uneven ripening process. Why? Because consumers prefer tomatoes that look perfectly red, even if it means sacrificing taste. In essence, choice architecture is being used to cater to our preferences, even with all the downsides they come with.
Our Behaviors Belied Our Biases
Produce that’s manipulated to look riper or tastier than it is didn’t happen because some shadowy agricultural cabal decided to gaslight the nation with bland tomatoes. It happened because we—collectively—showed up to the store, stared at a pile of fruit, and picked the prettiest ones. Over and over.
Retailers and growers noticed…because of course they did. Faced with two apples, most of us reach for the shinier one. We skip the peach with a small bruise, even if it smells amazing. We equate rich, uniform color with flavor, and any surface blemish with decay. So over time, the entire system—farming, harvesting, storing, displaying—optimized around that behavior.
A case study from the University of British Columbia notes that much of the produce rejected by retailers is tossed not because it’s unsafe or unripe, but because it’s too visually inconsistent. Imperfect fruits are treated as economic dead weight. And that mindset filters all the way back to how crops are bred.
Take tomatoes: modern breeding programs favored a trait called uniform ripening, which produces smooth, evenly red fruit that looks great on a store display. But it wasn’t just about looks. Uniformity made harvesting more efficient and sorting more automated. At the same time, breeders prioritized higher yields, longer shelf life, and resistance to bruising—all of which made tomatoes easier to grow, pack, and ship at industrial scale. The trade-off? Flavor. Or more precisely, the genes that support flavor—like those that influence sugar levels and aroma compounds—were slowly bred out or deprioritized (Wang & Seymour, 2017).
The result? A tomato that looks the part but lacks any deeper flavor. It’s a classic case of unintended consequences—a decision made to solve one problem that quietly created another.
Our behavior told the system what we wanted—even when our taste buds might’ve said otherwise. So the system responded by optimizing for what we reach for, even though, in doing so, it compromised flavor. The end result is more uniformity, and ultimately less waste—even if we had to sacrifice taste to get there.
This is choice architecture doing what it does best: picking up on our habits and then designing environments to serve them back to us. Not to trick us necessarily. Just to sell us what we keep reaching for.
Why You Can’t Win With a Tomato
Even if you try to learn from your mistakes, it often doesn’t matter—because there just aren’t any good heuristics left. Those heuristics we used to rely so heavily on simply no longer work very well. And so that tomato you picked—the one you were pretty sure was going to be a tasty piece of fruit—ends up being disappointing. And you have no way to know that until you slice it open.
What makes it all the more maddening is that even if you try to adjust, you’re still mostly guessing. Avoid that variety next time? Try a firmer one? Pick the one with the weird green shoulders? There’s no consistent logic anymore because the old cues—color, firmness, smell—have been decoupled from the thing they’re supposed to indicate3.
What used to be reliable signals of flavor and readiness have now been repurposed to serve entirely different goals—shelf stability, durability, visual perfection. Our brains are still wired to treat those cues as meaningful, but the fruit is no longer playing by those rules. The signals look the same, but they don’t mean what they used to, and so now it’s become nearly impossible to choose well. It’s become a sort of tomato roulette.
What Happens When the Heuristics Break?
If this were just about tomatoes, it would be mildly frustrating. But it’s not.
In a recent post, I wrote about how AI can act as a hyper-personalized choice architect—constantly reshaping decision environments to exploit our preferences and biases. But here, we’re seeing the reverse dynamic: technology being used to generate outputs that cater to our existing heuristics—the mental shortcuts we use to judge quality, fit, or value.
We see it in the produce aisle, where fruit is bred or treated to look ripe, even if it isn’t. But we also see it in job applications crafted by AI and sent out en masse (Rajan, 2025)—tools that, in some cases, have been used by bad actors to game the system (Microsoft, 2025)—in student essays that undercut the assumption that written assignments reflect genuine mastery (Kelly, 2024), and in the proliferation of AI-generated content filling up search results with plausible-sounding, low-substance slop (Caswell, 2025). In all these cases, the cues match what we expect “good” to look like—but those cues may no longer be nearly as valid as they once were.
And that’s the deeper risk. When the cues we’ve learned to trust can be so easily manufactured or manipulated, our ability to make sound decisions gets compromised—not because we’re lazy or irrational, but because the signals are misleading. The game has changed, but our brains haven’t had time to catch up. And even once we have time, such heuristics are often so firmly ingrained that they will take effort to unlearn.
We can’t stop using heuristics. They’re part of how human cognition works. But we can recognize when they’re being exploited—and we can start asking harder questions about what we’re really responding to.
That might mean different things depending on the context:
Accepting when heuristics just don’t work anymore. Some domains no longer offer trustworthy signals (purchasing certain fruits will remain a crapshoot).
Building better front-end filters. Sometimes we need smarter, earlier-stage heuristics to weed out noise before we even get to the familiar signals (e.g., identifying clearly AI-generated content before evaluating substance).
Creating new heuristics—or new systems. In some cases, we may need to rethink the decision environment itself (e.g., designing assignments that AI tools can’t easily complete without human insight).
Because whether it’s a tomato, a résumé, or a term paper, the surface might be just right. But sometimes that’s as far as it goes.
It helped that my produce manager also farmed, and I learned a lot from him.
One heuristic still works with a great degree of precision—the gustatory (taste) heuristic. Some fruits, like grapes or cherries are both small enough and packaged in a way that allows for a simple taste test. I seldom go wrong with a bag of grapes or cherries when I buy them because I have had a chance to sample them first.
This allegedly is not the case with heirloom varieties, but I have not tested or researched that myself so I cannot confirm the statement’s validity.
This reminds me of the shoot-the-messenger idea or the tendency in the medical industry to suppress the pain, even the symptom rather than cure the actual problem. And the phenomenon you're describing I'm thinking of as 'inflation' at play - not sure if anyone's used such a term, but I'd propose 'appearance inflation'.
And perhaps over time we'll develop newer intuitions to judge a good tomato and soon the industry will such a thing all over again. A cat and mouse game?
This phenomenon sometimes occurs naturally, when a signal becomes decoupled from the quality that it originally represented. A simple example is bright coloring for poisonous or unpleasant-tasting animals. Once this association is established in the minds of predators, other animals (primarily insects) start to evolve similar bright coloring without needing to also evolve to produce the poison or bitter chemicals. In evolutionary theory this phenomenon is referred to as "false signaling". Besides the realm of edibility, false signaling often occurs in the context of mate selection, whereby elaborate visual displays are a proxy for overall health in many bird species, but actually diminish the bird's odds of survival by making it more visible to predators and sometimes by impeding the bird's mobility.