Rats and electrical shocks
In one of their experiments – in 1908 – they placed a bunch of rats in a cage in order for them to explore one of two passages. One passage, labeled with a white card, contained a reward while the other one, labeled with a black card, led to an electrical shock. They wanted to test how quickly the rats figured out that white was a good thing, while black passages were to be avoided.
The one thing they varied was the voltage of the shock. Some rats received a mild shock which was not much more than a tickle, others were subjected to a shock of an intermediate level, while the third group was nearly fried to death each time the poor suckers merely sniffed at the black passage. Then Yerkes and Dodson compared the results.
Initially, as expected, higher shocks stimulated the rats to learn quicker; rats that received an intermediate shock were quicker to put into their tiny rat brains that black was to be avoided than the animals that received only a minor electrical stimulant. However, then it got interesting (rather than only cruel). The rats that received the high voltage fast-frying shock were much slower at performing the task. It took them much longer to figure out that white was good and black was bad – although they clearly had the highest incentives of all.
Apparently, making the stakes very high hampered these animals’ task performance, rather than making it better. And I could not help but wonder whether this same relationship as found in rats might not apply to, say, bankers.
Bankers and bonuses
Professor Dan Ariely, from Duke University, together with some colleagues designed a series of experiments with humans (rather than with rats) and with monetary rewards (rather than – to my slight disappointment – electrical shocks), to test what the relationship is for us humans between high incentives and task performance. That’s interesting because there are of course a lot of jobs around with pay-for-performance, for which we assume that high pay provides an incentive that leads to higher performance. However, it did not. Apparently, we’re the same as rats.
Dan and colleagues performed most of their experiments in India, where the financial rewards they gave for successfully completing a task could really represent a small fortune for a respondent. Believe me; these people were really motivated to get it right. As expected – and as for the rats – increasing the “pay for performance” relationship initially led to better task results. However, when the stakes were really high, people crumbled and could not get the job done. Apparently, a very stiff pay-for-performance relationship, paying people a near obscene amount of money if they succeed, actually make them less likely to perform well. It is not that these people are not motivated; they are too motivated. The very high pay killed their creativity, clogged their memory, and made them unable to solve the problem at hand. People who received less money for the same task did much better.
We assume that pay-for-performance motivates people, and motivation leads to better performance. However, the last step appears to be a flawed assumption. Large pay can surely motivate people, but too much motivation decreases their task performance. Hence, large task incentives like bonuses – used for traders, bankers an in many other industries – can be counterproductive. They cost you lots of money, but don’t get the job done.