Ask someone to accept a coin flip: heads you give them a hundred pounds, tails they give you a hundred. Almost nobody takes it, though the expected value is zero. Now ask what the gain would have to be, against a possible loss of a hundred, before the bet becomes attractive. The typical answer is somewhere between a hundred and eighty and two hundred and fifty.
That ratio is loss aversion, and Daniel Kahneman and Amos Tversky built it into the formal architecture of prospect theory in 1979, in a paper that eventually won a Nobel Prize and that remains the most cited work in economics of its era.
The core of prospect theory is that people do not evaluate outcomes as final states of wealth, which is what expected utility theory assumed. They evaluate changes, measured against a reference point — usually the status quo, but not always. Above that point the value function rises with diminishing sensitivity; below it, it falls, and it falls more steeply than it rises. The asymmetry is commonly summarized as losses hurting about twice as much as equivalent gains please, though the exact multiplier varies by domain and by measurement method, and a large multinational replication published in 2020 found the qualitative pattern robust across nineteen countries with substantial variation in magnitude.
Two consequences follow that are worth more than the number.
The first is that the reference point is a variable, not a constant, and whoever sets it controls whether an outcome is experienced as a gain or a loss. A salary of sixty thousand is a gain against fifty and a loss against seventy. A price is a discount or a surcharge depending on which number you were shown first. This is the mechanical link between loss aversion and the framing effect in the next chapter, and it is why so much manipulation consists of arranging what you compare something to.
The second is stranger. Because the loss branch is steeper, people become risk-seeking in the domain of losses. Faced with a certain loss and a gamble that might avoid it, most people take the gamble even when it is unfavorable. This is not a bug in a few people; it is the reliably observed pattern, and it explains behavior that looks otherwise inexplicable. The trader who doubles down to avoid booking a loss. The government that escalates a failing war rather than accept a defeat. The gambler who chases. The person who stays in a collapsing venture or a bad relationship because leaving means admitting the loss. Sunk cost, two chapters from here, sits directly on top of this.
For a manipulator, the operational instruction is simple: never describe what the target will gain by acting. Describe what they will lose by not acting. This is why churn-prevention offers list the features you are about to lose rather than the features you would keep. Why insurance is sold on catastrophe rather than on coverage. Why political messaging concentrates on what is being taken from you. Why the limited-time offer is framed as missing out rather than as declining. Each of these is the same edit: move the same fact across the reference point.
The defense is not to become indifferent to losses, which would be irrational in the opposite direction. It is to notice when a choice has been described exclusively in the language of loss, and to restate it yourself in terms of end states. What will I actually have, in each case, when this is over? That question ignores the reference point, which is the only part of the setup the other side controls.
Defense: The Money Is Already Gone: Escaping Sunk-Cost Entrapment