Logan:  Basic Principles of Learning

                         STIMULUS ASYNCHRONISM

          A conditional response tendency is acquired in classical
          conditioning only if the antecedent stimulus precedes
          the unconditional stimulus in time. 

     The time separating the antecedent stimulus event and the
unconditional stimulus is referred to as the INTERSTIMULUS INTERVAL (ISI)
and the principle is called stimulus asynchronism because classical
conditioning is not observed if the two stimuli are perfectly synchronized
such that they occur simultaneously.  Rather, the antecedent stimulus
must precede the US.   There is not evidence of conditioning if the onset 
of duration is sufficiently long that the antecedent stimulus precedes
its termination.  Specifically, the sound of a metronome will not become
conditioned to salivation if the food is given before the metronome
starts, at the same time that the metronome starts, or if the metronome
is sounded while the doge is eating.  Only if the metronome starts before
the food is given will conditioned salivation be observed.

     Of course, there is no time for anticipatory salivation to occur if
the stimuli are presented simultaneously or in backward order.  In order
to deal with this problem, test trials are run without the US.  That is
to say, after a metronome and food have been presented simultaneously a
number of time, the metronome is sounded without the food to find out
whether conditioning had occurred but could not be observed because of
unconditioned salivation to the food.  There is no evidence of a CR on
such test trials, supporting the principle that the antecedent stimulus
must, indeed, be antecedent in time.  At the other extreme, however, it
is probably not surprising that no conditioning is obtained if the
stimulus asynchronism is very long; the antecedent stimulus may be gone
and forgotten before the unconditioned stimulus arrives.

     An especially difficult issue concerns the length of the inter-
stimulus interval for optimal conditioning.  With the human eyeblink, the
optimal ISI is about a half second and no conditioning occurs if the ISI
is as long as three seconds.  With salivation in dogs, several seconds is
ideal and good conditioning occurs with ISI's of 30 seconds or longer.
The most extreme observations have been made with learned taste aversions
in rats.  The phenomenon of bait-shyness has long been known: A wild
animal that survives being poisoned will thereafter avoid the taste of
that poison.  In the laboratory, a rat can be exposed to a novel (and
highly preferred) flavor such as saccharine sweetened water. 
Then, as long as two or three hours later, the rat is given a sub-lethal dose of poison.  After the rat has recovered from the sickness, it will
avoid anything tasting of saccharine.

     The phenomenon of learned taste aversions does not disprove the
Principle of Stimulus Asynchronism; even in that context, conditioning is
best if sickness occurs within a few minutes of tasting the saccharine,
and conditioning does not occur if the ISI is over three or four hours.
But where the necessity for very short intervals, as in the human
eyeblink, trace of the stimulus, the occurrence of conditioning with very
long intervals suggests that conditioning can occur to a long-lived
memory of the stimulus.  It may also be the case that organisms are
innately predisposed to form some types of associations, in which case
long intervals are possible, whereas short intervals are necessary for
more arbitrary connections.

     The difficulty with the proposition that organisms are innately
predisposed (biologically prepared) to form some associations and
are perhaps contra-prepared (biologically prepared NOT) to form other
associations, is determining these predispositions in advance.  For
example, although it seems intuitively reasonable than an organism would
associate a poison-induced sickness with a taste stimulus, it is
not obvious why even better taste-aversion conditioning occurs with an
unnatural X-radiation US.

     There is an important corollary to the Principle of Stimulus
Asynchronism.  The latency of the CR, which is the time after the onset of
the antecedent stimulus before the CR begins, tends to be adjusted such
that it occurs just before the US.  That is to say, the longer the
interstimulus interval, the longer the CR latency.  Actually, the latency
of the CR may be quite short during the early stages of conditioning, but
it gradually lengthens until it almost equals the ISI.  And if the ISI is
changed, the latency of the CR gradually changes appropriately.

     This is a good context in which to alert the reader to the fact that
our knowledge of the Principles of Learning is still far from complete,
the interestimulus interval in classical conditioning is one of the most
obvious variables for experimental analysis, but many of the details of
this principle are still unknown.  A glaring omission is the effect of
trial-to-trial variation in the interstimulus interval.  The reason for
this omission is that no one has satisfactorily solved the dilemma of how
properly to measure a possible conditioned response under such conditions.
As we just noted, it is not clear when a CR should occur if the
interstimulus interval varies.  We are inclined to believe that
conditioning would occur with variable interstimulus intervals, but we do
not yet know how to study this variant of the basic procedure.

     In any event, the practical significance of the Principle of
Stimulus Asynchronism is simply that, to insure that an association is
formed, the stimuli should occur close together in time.  For example, if
a person has done something commendable, the time for commendations to be
given is as soon as possible.

TERMS: Backward conditioning, simultaneous conditioning, interstimulus
interval,  memory, stimulus asynchronism, antecedent stimulus,
conditioned stimulus, stimulus trace.