Logan:  Basic Principles of Learning

                       PRINCIPLE: PRIMARY PUNISHMENT

          If a response is shortly followed by (a) the onset of 
          an innately aversive event or (b) the withdrawal of an
          innately emotionally-positive event, there will result 
          a decrease in the future probability of that response.

     As with reinforcement, there are two classes or events that fall 
in the category of punishment.  POSITIVE PUNISHMENT refers to the onset
of an aversive event, such as an electric shock applied to the paws of 
a rat that is bar-pressing for food.  NEGATIVE PUNISHMENT refers to the
withdrawal of an emotionally-positive event, such as the taking of food
out of a rat's mouth.  Note that in the latter case, we are talking about
withdrawing a positive event that is actually present.  Not giving a
customary reward constitutes extinction rather than punishment, although
these operations have very much in common.  They both lead to a decrease 
in the future likelihood of the response and exposure to one of them
increases an organism's persistence in the fact of the other.  Occasional
punishment increases resistance to extinction, and occasional
nonreinforcement increases resistance to punishment.

     Positive punishment is, by all odds, the most common procedure in
the aversive control of behavior.  This is especially true when the
objective is to get an organism not to make some response.  As would be
expected, the effectiveness of punishment increases the greater its
intensity: strong shocks are more effective than weak shocks, and the
greater the amount of reinforcement that has been given for making a
response, the stronger the punishment must be in order to suppress it.
Less obvious but no less important is the fact that the effectiveness of
punishment also depends on what the aversive event causes the organism to
do.  Punishment is most effective when it leads to the making of
responses that are incompatible with the reinforced response.  A shock
that causes a rat to jerk back from the goal works better than a shock
that causes a rat to jump forward towards the goal.  

     One common way to describe the effect of punishment uses the 
Principle of Secondary Motivation (fear) and the notion of
response-produced feedback stimuli.   Such stimuli occur as antecedent
stimuli for the classical conditioning of fear.  That is to say, the
organism starts to feel afraid whenever it feels itself starting to make
the punished response.  Stopping the response is reinforced because it
terminates the response-produced feedback stimuli and accordingly reduces
the organism's fear of punishment.  The second determinant of the effect
of punishment is based on the overt responses elicited by the punishment.  
According to the Principle of S-R Contiguity, these responses will become
associated with the initial stimulus.  In this case, the effect will
depend on the extent to which this response that is produced by
punishment is compatible or incompatible with the response that produces
the punishment.

     Many psychologists are concerned about the high frequency of using
punishment in the control of behavior.  This is because fear is not only
conditioned to the response-produced feedback stimuli or the undesirable
behavior itself, but fear also becomes associated with all of the
contextual environmental cues at the time punishment is given.  In the
laboratory a severely punished rat resists being placed in the apparatus
and may even attempt to bite the experimenter.  However, this concern is
at least to some extent relieved by research that shows no ill-effects
when the organism is concurrently provided with a rewarded alternative
that is not punished.  If a rat is rewarded for pressing either of two
bars, and tends to prefer pressing one of them, this preference is
readily reversed by punishing the favored bar.  This is true even if the
favored bar has been given a somewhat larger reward; the rat simply works
faster on the unpunished bar.

     We can summarize this analysis of punishment by stating several
dicta that should guide the use of punishment when one must resort to
aversive control: Use ADEQUATE punishment; too weak punishment may fixate
rather than eliminate the behavior.  BE SURE there is a rewarded
alternative; otherwise, one creates intense conflict.  Punishment should
"suit the CRIME;" unless it elicits an incompatible response, it may
actually potentiate the behavior.  DON'T WAIT; delayed punishment works
on the wrong behavior, if at all.  Finally, ELIMINATE rewards after
punishment; punishers can become secondary reinforcers if associated with
special rewards.  It should be clear from these dicta that there are many
ways to misuse punishment.

     The high frequency of punishment in modern society hardly requires
proof.  The reason is simple: it works.  To be sure, it does not always
work as effectively as it could because the people administering the
punishment do not follow the aforementioned dicta.  Furthermore, the
widespread (but generally unwarranted) fear of the police is testimony to
the side-effects of aversive control.  The worst offenders are those who
punish natural responses without providing an alternative; for example,
condemning masturbation by a person with no other socially-acceptable
outlet for natural sexual needs is not only ineffective, but cruel.

TERMS: Passive avoidance, context, response-produced feedback, fixation,
punishment (primary negative, primary positive, secondary
negative, secondary positive, varied).