effortless study of operant conditioning

May 30th, 2008

In psychology, the term operant conditioning is used to describe a simple type of learning. The aim is to create a certain desirable behavior by reinforcing (ie. rewarding) it. In the alternate case, an undesirable behavior may be extinguished by punishing this behavior (ie. withdrawing a reward, or actual punishment of some kind).

Psychologists have long studied this behavior in animals in laboratories. A small rodent (apparently this is the their optimal kind of animal) is put into a confined space. The desired behavior (just for kicks) is for the rodent to press on a lever in that space. This behavior is reinforced by releasing a food pellet everytime the lever is pressed. After repeating this experiment a few times, the animal (presumed hungry or gluttonous) will press the lever, eat the food and exhibit no other behavior at all, it will immediately press down the lever again. The behavior has now been taught.

But what if you don't have any rats or cardboard? Well, there's a simpler way. Suppose you live in a house with 4 other people. There is a bathroom you all share that has a lock on the door, but no explicit visual indicator of when it's occupied. So in principle there is no way of knowing whether it's vacant, thus something (another way of checking vacancy) has to be learned in order to make this judgment. Okay, let's set up the experiment.

Behavior to extinguish: Blindly pushing down the door handle.
Door doesn't open, it's locked from the inside.
Additional stimulus:
Visual cues:
a) the door is either ajar or shut,
b) the lock is open or shut, observed in the crack between the door and the frame,
c) the light on the inside is on or off, observed through the crack between the door and the frame.

The subject approaches the bathroom door, pushes on the handle, only to find that the door doesn't open. For a person occupying the bathroom this is perceived as an attempted intrusion, and the subject is aware of this, having been on the other side before. (In addition, certain norms in society dictate that attempts at breaking into an occupied bathroom are not in good taste.) This causes mild embarrassment in the subject, who offers an awkward apology and leaves. In other words, this behavior is undesirable to the subject, who will attempt to avoid repeating it.

So we have the behavior of pushing down the handle, and a punishment which is the locked door. In order to extinguish this behavior, we have to learn how to determine whether the bathroom is occupied. In basic operant conditioning the door is either always open or always locked, and the response (reinforcement or punishment) will create the behavior that corresponds. In this case, the door is sometimes locked, and we need additional stimulation to achieve the corresponding behavior.

This additional stimulus exists in the form of visual cues. First, we need to establish whether the cues are useful.

The door is ajar or shut

Observation shows that the door is left ajar 60% of the time when the bathroom is vacant. It is always shut when the bathroom is occupied, so this cue will never mislead by correlating an open door with occupancy.
In terms of observability, a subject intending to enter the bathroom will always detect the state of the door being open or shut, because we are sufficiently conditioned to look at the door of the room we plan to enter.

The lock is open or shut

The state of the lock perfectly reflects the state of occupancy at all times. There's no way to lock the door from the outside, and all occupants lock the door.
The state of the lock is observable only when standing right in front of the door. The space between the door and the frame is wide enough to clearly determine whether the lock is open or shut. The likelihood of spotting this cue is 30%.

The light in the bathroom is on or off

Observation shows that the light is on 15% of the time when the bathroom is vacant. It also shows that an occupied bathroom always correlates with the light being on.
This cue is much less likely to be observed, for two reasons. Firstly, it is not as readily visible as that of an open door because of ambient light. Strong ambient light (ie. daylight) will make this cue visible only when the subject is positioned directly in front of the door. Secondly, the incentive for lighting the bathroom differs from that of using a locking mechanism. A lock prevents the occupant from being accosted, which would be a source of embarrassment to the occupant. Therefore, this incentive relates the question of occupancy directly to the question of locking, whereas the incentive for lighting the bathroom is a purely functional one. In other words, the subject is much more likely to look for a lock than to look for light on the inside. The likelihood of detecting this cue is 20%.

The task of learning to determine the occupancy status of the bathroom is therefore defined as the ability to spot and correctly interpret the visual cues. In this case we have several cues to draw the correct conclusion on. The state of the door will always be seen, and has a 60% accuracy rate in terms of revealing occupancy. Then there is the state of the lock, which in itself is enough to make the right determination. Finally, there is the bathroom light, which has an 85% accuracy rate. If this seems like a trivial task to you, you'd be mistaken. Real world observations show that out of the four subjects, two learned the correct behavior, while the other two, even over the course of several months, didn't.

According to the principles of operant conditioning, such a case requires a stronger punishment to extinguish the undesired behavior. In many of the experiments performed on animals, there was an element of electric shock involved. But since I don't have the equipment for it... ;)

So there you have it, a study of operant conditioning purely out of the naturally occurring circumstances in a house. No work at all had to be done to set up the experiment or prepare the subjects. Consequently, no bias could have been introduced by tampering with the circumstances. There's also another added advantage. Skeptics like to make the accusation that laboratory experiments cannot be extrapolated to the real world, because they were performed in an artificial environment. That argument has no stand here.

Another common case of operant conditioning is banging on your tv to make it work. People try it, it sometimes work, so they keep at it.

:: random entries in this category ::

1 Responses to "effortless study of operant conditioning"

  1. erik says:

    "Another common case of operant conditioning is banging on your tv to make it work. People try it, it sometimes work, so they keep at it."

    Eugh :D

    Good read, I liked that.