Contingent rewards can reduce intrinsic motivation

Mark Lepper, David Greene, and Richard Nisbett (1973) conducted research on this question. They divided children into three groups:

Group 1 was the Expected group. The researchers showed the children the Good Drawing Certificate and asked if they wanted to draw in order to get the certificate.

Group 2 was the Unexpected group. The researchers asked the children if they wanted to draw, but didn’t mention anything about a certificate. After the children spent time drawing, they received an unexpected drawing certificate.

Group 3 was the Control group. The researchers asked the children if they wanted to draw, but didn’t mention a certificate and didn’t give them one.

The real part of the experiment came two weeks later. During playtime the drawing tools were put out in the room. The children weren’t asked anything about drawing; the tools were just put in the room and available. So what happened? Children in the Unexpected and Control groups spent the most time drawing. The children in the Expected group, the ones who had received an expected reward, spent the least time drawing. Contingent rewards (rewards based on specific behavior that is spelled out ahead of time) resulted in less of the desired behavior if the reward was not repeated. Later the researchers went on to do studies like this, with adults as well as children, and found similar results.

Before the certificate, drawing meant drawing. Afterwards it could mean “the activity adults pay me to tolerate”. The reward changed the interpretation before it changed the behaviour.

That is the awkward part of contingent incentives. A spike in activity can conceal a collapse in motive. Close ten support cases for points and difficult cases become professionally inconvenient. Reward five helpful replies and five thin replies appear with admirable punctuality. The measurable behaviour rises while judgement, curiosity and care are pushed outside the target. Management sees movement and declares victory; participants infer that the activity must be grim if bribery is required. When the prize disappears, the new reason disappears too, and returning to the old reason can feel like working for less. A sales competition can raise calls for one month while teaching everyone to guard promising leads. A learning badge can increase completed modules while making the shortest video unusually attractive. The visible gain arrives immediately; the changed interpretation appears after the celebration.

Compensation remains essential for tedious or valuable labour. Unexpected recognition can make contribution visible. The danger comes from assuming a temporary prize can purchase permanent conduct without altering what the conduct means. Progress, mastery, autonomy and social value may already support the activity. An incentive can strengthen those reasons, compete with them or replace them entirely. The distinction will rarely appear in the first week’s count. Raw volume is particularly corruptible: the person solving one ugly case loses to the person disposing of six easy ones, while the metric congratulates itself on productivity.

Two weeks later, the drawing tools were available and no certificate was promised. The children who had expected one spent the least time drawing. The reward had ended. Its interpretation remained.