Psy 551  Learning Theory
Handout on Spence

     As research was carried out during the heyday of the Hullian
system (late 1930s to end of the 1940s), four research findings
became particularly challenging.  These were: (a) the PRE; (b)
the importance of such "cognitive factors" as instructions and
expectancies; (c) the latent learning experiments; and (d) the
manipulations of reinforcement (e.g., the reinforcement shift
experiments by Crespi and Zeaman).  Hull had devoted much effort
to deal with these findings, but no one was really satisfied. 
Spence modified aspects of the Hullian system during the 1950s
and 1960s to better come to grips with these findings.

     First, Spence included cognitive and emotionally-based
drives, generally linked to anxiety and frustration, in addition
to Hull's metabolic, physiologically-based drives [cf. (a) and
(b) above].

     Second, Spence emphasized the contribution of incentive,
through K and the r g - s g mechanism [cf. (b), (c), and (d)
above].

     These modifications, which again were made to account for
the research findings mentioned above, meant that Spence's
conception of how reinforcement contributed to learning and
performance in classical and instrumental conditioning procedures
was fundamentally different from that of other theorists.  Recall
that for Hull, reinforcement by drive reduction operated in what
was called a classical conditioning procedure as well as an
instrumental conditioning procedure.  More specifically, in the
Hullian analysis of performance generated by an instrumental
conditioning procedure, reinforcement was viewed as the number of
trials in which there had been drive reduction ( s H r ).  For
Spence, performance generated in an instrumental conditioning
procedure became more contiguity-based.  Food per se was a matter
of incentive, which affected performance and s E r.  Learning per
se became a matter of the number of pairings between stimulus and
response.  Thus, learning became based on contiguity between
stimulus and response, whereas performance became based on
incentive, the actual delivery of the food.  Drive contributed to
the motivational complex, but for instrumental learning,
reinforcement by drive reduction was technically no longer
necessary.  Performance generated by a classical conditioning
procedure remained based on reinforcement by drive reduction, as
it had been for Hull.

     Hull adopted a one-process approach to learning:  drive
reduction.  Spence and Skinner actually adopted a two-process
approach.  However, Spence's approach differed from Skinner's. 
For Spence, reinforcement as drive reduction operated in
performance generated by a classical conditioning procedure (as
it had for Hull), but not in performance generated by an
instrumental conditioning procedure.  The food presentation was
still important for two reasons: (a) as incentive motivation, and
(b) as the means for getting the response to occur in the first
place.  It was just that the food presentation was no longer
important as a means of reducing drive in instrumental behavior. 
For Skinner, reinforcement operated in operant but not classical
conditioning.  [Of course, Skinner eventually dropped the use of
"Drive," and his conception of reinforcement wasn't linked to
drive reduction anyway.]


Theorist              Classical             Instrumental

Hull                  reinforcement         reinforcement         
                      by drive reduction    by drive reduction

Spence                reinforcement         contiguity            
                      by drive reduction

Skinner               contiguity            reinforcement*


*not related to drive reduction