Nobody needs to gamble in Fallout: New Vegas. The slot machines pay out in caps you will never spend, the blackjack tables exist purely as set dressing, and the whole casino strip can be skipped without consequence. Players still sit there for hours. The same thing happens with the poker tables in Red Dead Redemption 2, the dice in Kingdom Come, and every roguelike that hands you a random reward at the end of a run.
The stakes are fictional. The pull is not. And once you understand why, a lot of modern game design stops looking like a coincidence and starts looking like a deliberate borrowing from a century of behavioural research.
That borrowing is also why the line between a game and a gambling product has become so hard to draw, a question that matters for players comparing free-to-play online gaming formats where virtual currency, not cash, is what moves on the screen. The mechanic is often identical. What changes is the plumbing behind it.
The mechanism has a name, and it predates video games
The engine here is the variable-ratio reinforcement schedule, mapped out in mid-century behavioural psychology long before anyone had a games console. The principle is simple and slightly unflattering: a reward delivered on an unpredictable schedule produces far more persistent behaviour than a reward delivered every time.
If a chest always contains a common item, you stop caring about chests. If it contains something excellent one time in forty, and you cannot know which time, you keep opening chests. The uncertainty is not a side effect of the reward. The uncertainty is the reward.
This is why the loop survives the removal of money. Your brain is not responding to the value of the payout, it is responding to the pattern of delivery. A slot machine that pays in worthless in-game caps still runs the same schedule as one that pays in cash, which is precisely why New Vegas can hold you at a table where nothing is at stake.
Near-misses do a lot of quiet work
The second ingredient is the near-miss. Two matching symbols and a third that lands just above the payline. A legendary drop that appears in the loot pool and goes to someone else. A card that would have completed the hand showing up one draw too late.
Structurally, a near-miss is a loss. Experientially, it registers as progress, which is why designers reach for it constantly. Boss fights that end with the enemy at a sliver of health, gacha pulls that show the rare animation before resolving into something common, and card games that let you see the deck you almost drew from are all built to make failure feel like it contained information about future success.
It usually does not. Each pull, spin or draw is independent. But the feeling of having nearly arrived is strong enough that players will fund another attempt with time, and sometimes with money.
What the research actually established
The evidence base here is narrower than the discourse suggests, and worth stating precisely.
The landmark work is a 2018 study by David Zendle and Paul Cairns, published in PLOS ONE, which surveyed 7,422 players and found a significant association between how much people spent on loot boxes and the severity of their problem gambling. The link was substantially stronger than the association between problem gambling and other in-game purchases, which pointed at the gambling-like structure of loot boxes rather than at spending in general. The University of York's summary of the research is worth reading in full.
The authors were careful about what they had not shown. The study could not establish direction: it is unclear whether loot boxes act as a gateway to problem gambling or whether people already prone to gambling problems are drawn to loot boxes. Both could be true simultaneously. What the data does support is the narrower claim that the mechanic and the harm travel together, which is enough to justify regulatory attention without justifying the stronger claims often made in headlines.
Why the ratings boards keep contradicting themselves
The clearest illustration of how confused this space is arrived with Balatro. In 2024, PEGI rated the poker-themed roguelike 18+ for prominent gambling imagery and for teaching hands that transfer to real poker, a decision that briefly pulled the game from several European storefronts. Balatro contains no microtransactions and no real money whatsoever.
Its developer, LocalThunk, pointed out the obvious asymmetry: EA Sports FC 25, whose Ultimate Team packs are randomised purchases made with real money, carried a 3+ rating over the same period. After an appeal by the publisher, PEGI reclassified Balatro to 12+ in February 2025, reclassified Luck Be A Landlord alongside it, and began reviewing how it handles simulated gambling. Belgium, for its part, had already ruled certain loot boxes unlawful under gambling legislation back in 2018.
So the boards ended up treating imagery more harshly than transactions, then partially corrected. Coverage of the fallout is still accumulating on the Balatro hub, and the underlying policy question has not been settled anywhere.
The takeaway for players is less dramatic than the headlines and more useful. The loop that keeps you at a fictional slot machine is the same loop that keeps someone at a real one, so the presence or absence of a payment rail is what determines the financial risk, not the presence or absence of the mechanic. Knowing which one you are looking at is the entire skill.