Rewards Punish

Definition

Rewards and punishments are commonly framed as opposites — carrots versus sticks — but Kohn argues they proceed from the same psychological model: both are attempts to control someone's behavior by manipulating what happens to them. "Do this and you'll get that" and "Do this or here's what will happen to you" differ in tone, not in kind. Because a reward is by definition conditional, the possibility of not earning it is always present, and missing out on an expected reward is, in practice, indistinguishable from being punished. As Kohn puts it, "the stick is contained in the carrot."

In the Book

Chapter 4 opens Kohn's five-reasons-rewards-fail argument with this claim. He cites a 1991 study of elementary teachers across thirteen schools showing that reward use and punishment use were positively correlated — the same teachers relied on both — and a survey finding a similar link between reward-giving and physical punishment among mothers of kindergartners. He walks through the parent who threatens "Keep this up and you can forget the circus tomorrow" as functionally identical to a punishment threat, and notes behaviorists' own "response cost" terminology — withholding something good — as punishment by another name. He concludes there are only two ways to avoid this: give rewards unconditionally (in which case they aren't really rewards) or present them only after the fact as a surprise — but almost no real-world reward system works that way, because the whole point of promising a reward in advance is to control behavior.

Why It Matters

This collapses a distinction most incentive design treats as foundational — that positive reinforcement is a kinder, safer alternative to punishment. Once you see that any conditional reward carries an implicit threat of its own withdrawal, "positive" motivation systems (bonus plans, gold stars, leaderboards, loyalty tiers) can be audited for the same coercive dynamics usually reserved for punitive ones: does missing the target feel like a loss, does the recipient feel controlled, does the system quietly train people to manage the reward-giver rather than do the underlying work.