Mark Hammer

Back when I was studying animal learning and conditioning, one of the more intriguing ideas in the literature was that, although positive reinforcement of desirable behaviors was generally more effective in directing an organism to do what you wanted than punishment of undesirable behaviors, one of the unintended side-effects of positive reinforcement was that it induced “behavioral stereotypy” (see 30 year-old work by Barry Schwartz at Swarthmore). In other words, it introduced a certain inflexibility and narrowness in what the organisms being trained were doing. Where punishment essentially conveyed “Do anything BUT this“, reinforcement could inadvertently convey “Do ONLY this, and nothing BUT this”.

The general takehome message was that, whether rats, pigeons, dogs, or people, organisms are constantly learning, and very often learning things OTHER than what you wanted them to learn. And often, this happens because one is unintentionally rewarding them for behaviors other than what you were aiming for. Rewards programs are generally not much more intelligent than automatic feeder mechanisms dispensing food pellets to pigeons, and reward superficial indicators of some desired outcome, as opposed to more fully-fleshed comprehensive indicators of some abstract quality of performance.

Much the same way that managers can come to manage to the performance indicators, employees will respond to the rewards/awards.

This is the perpetual challenge of applying incentives for performance: finding a balance between indicators that are easy to measure and keep track of but too narrow in scope, and those that are burdensome to apply but more fully capture the entire scope what it is you want to reward and acknowledge. Lean too much in the one direction, and you have perplexed managers and employees. Lean too much in the other direction, and you have staff who are gaming the system and behaving more or less like pigeons who think they need to hop in circles, standing on one leg, in order to get fed.