JF AAS Sept. 1997

If your proposal for changing behavior ends with "that will make them think twice," you are not thinking like Skinner.

If you are talking about rearranging incentives, you are not talking Skinner.

A news item reports that computers are now being fitted with software that counts the time that data is actually being entered. This is hailed as a great breakthrough which reduces monitoring costs and allows office workers to be paid according to work. Is piece work pay an example of positive reinforcement? No. It is the usual hedonism of the future - workers calculating future payoffs - work now to get something in the future. It may result in more work, but Skinner would suggest that it runs the risk of suppression and subterfuge. If the monitoring system breaks down, workers may slack again. They are waiting for the monitor to sleep.

Skinner's is a hedonism of the past. Workers subconsciously work to a high level without calculation (and that means no calculation of the probability of being caught for slacking). There may have been a reinforcement in the form of money when the desired work quality is achieved. Or, some other reinforcement may have occurred, e.g. Praise.

What difference does a hedonism of the past or future make? If it is of the past, the worker is more likely to continue the desired behavior if the monitor sleeps. Of course, if the monitor can't reinforce the desired behavior, the result may also be deterioration of the behavior - but the decay may be slower.

Empirically, how can we tell the difference between a reinforcer that works subconsciously or a reward that works in a calculation if the material is the same? Could we ask people? Are you working hard to get the bonus and promotion, or are you doing it because you want to, seems the right thing to do, or have not considered any other alternatives?" Can we trust the answers?

What behavior would you predict when office workers are monitored by their computers and rewarded on a piece rate basis. With the usual rational calculus we would predict greater effort-less on-the-job leisure. Skinner might predict resentment, searching for ways to beat the system, e.g. enter a lot of mistaken key strokes, learn how the counter works and corrupt it, etc. The person with inherently low skill and low speed may become discouraged and not perform even up to their capacity.

Could a Skinnerite do any better? The controller would still have to know when the worker did as desired in order to deliver the reinforcement. But the Skinner-type controller might not choose the perfect computer monitoring technology knowing that the monitoring is part of the environment and might be viewed by the workers as negative reinforcement-big brother is watching you.

What about the material content of the reinforcement? Skinner would not eliminate money, but he would not be fixated on it either. He does not worry much about "why" something is a reinforces, but experiments and tries to observe what works. Maybe that is the usefulness of the Skinner perspective in economics. It increases our options.