Professional Documents
Culture Documents
The Effect of Verbal Reinforcement Combinations On Conceptual Learning'
The Effect of Verbal Reinforcement Combinations On Conceptual Learning'
Two previous studies (1, 2) have cancelled by the greater negative rein-
investigated the effect of verbal rein- forcement of the incorrect response by
forcement combinations on learning Nothing-Wrong; therefore the Right-
and extinction. The specific rein- Nothing and Nothing-Wrong combi-
forcement combinations were: the E nations should lead to similar rates of
says Right for a correct response, learning.
Wrong for an incorrect response Since these expectations were not
(Right-Wrong); Nothing for a correct borne out in the first study (1), a
response, Wrong for an incorrect re- reformulation of the verbal reinforce-
sponse (Nothing- Wrong); and Right ment continuum was attempted (2).
for a correct response, Nothing for an It was assumed that Nothing is a
incorrect response (Right-Nothing). nonreinforcer and that Right is a much
There were two major findings: (a) weaker positive reinforcer (approach-
Right-Nothing results in significantly ing Nothing) than Wrong is a negative
slower learning than either Right- reinforcer. Thus Right-Wrong is
Wrong or Nothing-Wrong, and (b) similar to Nothing-Wrong, and both
Right-Wrong and Nothing-^rowf of these combinations offer more
yield similar rates of learning. differential reinforcement than does
These data cannot be explained by Right-Nothing In other words the
the commonly accepted verbal rein- critical component is Wrong, and the
forcement continuum, which assumes combinations with Wrong (Right-
that Right is as strong a positive Wrong and Nothing-Wrong) should
reinforcer as Wrong is a negative yield faster learning than the com-
reinforcer. According to this con- bination without Wrong (Right-Noth-
tinuum Right-Wrong offers more posi- ing). These predictions coincide with
tive reinforcement of the correct the results reported previously (1, 2).
response than Nothing-Wrong and The theoretical reinforcement con-
more negative reinforcement of the tinuum was tested in a very limited
incorrect response then Right-Noth- experimental learning situation. The
ing. Since more differential rein- paradigm was that of stimulus gen-
forcement leads to faster learning, the eralization : the dimension was height,
Right-Wrong combination has a dis- and during acquisition only stimuli of
tinct advantage over the other two a given height were presented.
combinations, and it should yield Thereafter generalization was tested
significantly faster learning. Right- without extinction (1) or with extinc-
Nothing offers more positive rein- tion (2) on stimuli of varying heights.
forcement of the correct response than Furthermore, S was limited to making
Nothing-/F«mg, but this advantage is only one of two possible responses.
1
The writers acknowledge the helpful criti- Obviously the proposed verbal rein-
cism of Jerry Wiggins. forcement continuum can have little
283
284 ARNOLD H. BUSS AND EDITH H. BUSS
generality because of the severe limi- cards was exhausted, he was handed another
tations of the experimental conditions (identical) deck. If the color concept was not
learned to the criterion by 100 trials, S was
under which it was tested. There- stopped.
fore, the present experiments attempt There were three groups of Ss, each with a
to apply the theoretical continuum to different verbal reinforcement combination:
Right for a correct response, Wrong for an in-
a conceptual learning situation in correct response (Right-Wrong); Nothing for a
which S has more than two possible correct response, Wrong for an incorrect response
responses. The predictions for this (Nothing-^Frowg); and Right for a correct re-
new situation are essentially the same sponse, Nothing for an incorrect response (Right-
Nothing). The same combination was employed
as for the previous experimental in the learning of both concepts.
situations: (a) the Right-Wrong and
Nothing-Wrong groups will learn new Results
concepts significantly faster than the The response measure used is the
Right-Nothing group, and (b) the number of trials to learn, excluding
conceptual learning of the Right- the 10 consecutive correct trials. The
Wrong and 'Nothing-Wrong groups mean and median number of trials to
will be approximately the same. learn the shape concept is zero for all
three groups. Only three Right-
EXPERIMENT I Wrong Ss, one Nothing-Wrong S, and
no Right-Nothing S required more
Method than zero trials to learn the shape
Subjects,—The Ss were 45 neuropsychiatric concept. Since 41 out of 45 Ss
patients of both sexes who were not receiving started sorting for shape immediately,
shock treatment, not mentally defective, and
not diagnosed as having organic brain damage.
we may conclude that these Ss were
Fifteen Ss were assigned to each of three groups "set" for shape. Virtually all of the
in the order of their appearance. Right-Wrong Ss and all of the Right-
Material.—The Wisconsin Card Sorting Test Nothing Ss received only the Right
(4) was used. In this test there are stimulus component of their respective rein-
cards and response cards. The stimulus cards
(1 red triangle, 2 green stars, 3 yellow crosses, forcement combinations; all but one
and 4 blue circles) are placed in the upper set of of the Nothing- Wrong Ss received
four divided compartments. The S sorts a deck only the Nothing component of their
of 64 response cards into the lower set of com- reinforcement combination. Since
partments under the stimulus cards on the basis
of the shape, the color, or the number of the
most of the Ss did not receive both
figures printed on each card. components of the reinforcement com-
Procedure.—The S was handed the deck of binations, the "learning" of the shape
response cards and told to sort them into the concept cannot be used as a basis for
bottom set of compartments according to where testing our predictions.
he thought they belonged. To all queries E
replied, "Just go ahead and sort them where After making 10 consecutive correct
you think they belong," shape responses, all Ss were required
The first concept to be learned was shape. to shift to a color concept. The
Placing a response card under a stimulus card learning data for color are presented
having figures of the same shape was correct;
all other sortings were incorrect. The criterion in Table 1.
of learning w.as 10 consecutive correct trials. The Right-Nothing group differs from
When S had learned the shape concept to the the other two groups in that its mean
criterion, the concept was changed to color is much larger and its SD is much
without S being told of this change. Again the
criterion of learning was 10 consecutive correct
smaller than those of the Right-Wrong
trials. If S did not make 10 consecutive color and Nothing- Wrong groups. The
sortings by the time the deck of 64 response smaller Right-Nothing SD may be
VERBAL REINFORCEMENT COMBINATIONS 285
this latter group should not learn any learning of the color concept the Right-Wrong
faster than the former. and Nothing-Wrong groups learned at similar
In the two experiments reported here rates, and both learned faster than the Right-
an attempt was made to extrapolate from Nothing group.
In Exp. II (patients and nurses) the Ss
a generalization situation to a conceptual learned a number concept first and then a color
learning situation, and from data ob- concept. Only the Right-Wrong and Nothing-
tained solely with neuropsychiatric pa- Wrong combinations were used, and there were
tient <5"s to data on normals. It was no differences in their learning of either concept.
assumed that the theoretical continuum The results of both studies confirmed predictions
was not limited to either the generali- made on the basis of the proposed verbal rein-
zation paradigm or to patient >9s. The forcement continuum: Nothing is a non-
fact that the predictions were confirmed reinforcer and Right is a weaker positive rein-
in both experiments suggests that the forcer (approaching Nothing) than Wrong is a
assumptions were correct and leads to negative reinforcer. It was concluded that this
the conclusion that the stated theoretical theoretical reinforcement continuum possessed
verbal reinforcement continuum pos- some generality,
sesses some generality.
REFERENCES
SUMMARY 1. Buss, A. H., WIENER, M., & Buss, E.
A theoretical verbal reinforcement continuum Stimulus generalization as a function of
that had been derived from generalization data verbal reinforcement combinations. /.
obtained with neuropsychiatric patient Ss was exp. Psychol, 1954, 48, 433-436.
applied to a conceptual learning situation 2. Buss, A. H., BRADEN, W., ORGEL, A., &
(Wisconsin Card Sorting Test), with both Buss, E. Acquisition and extinction
patients and student nurses as Ss. The rein- with different verbal reinforcement com-
forcement combinations were: E says Right for binations. /. exp. Psychol., 1956, 52,
a correct response, Wrong for an incorrect 288-295.
response (Right-Wrong); Nothing for a correct 3. COCHRAN, W. G., & Cox, G. M. Experi-
response, Wrong for an incorrect response mental designs. New York: Wiley, 1950,
(Nothing-Wrong), and Right for a correct Pp. 92-93.
response, Nothing for an incorrect response 4. GRANT, D. A. Perceptual versus analytical
(Right-Nothing). In Exp. I (patients only) Ss responses to the number concept of a
were required to learn a shape concept and then Weigl-type card sorting test. /. exp.
a color concept. There were no differences Psychol, 1951,41, 23-29.
between groups (reinforcement combinations)
in their learning of the shape concept, but in the (Received October 3, 1955)