This article was downloaded by: [University of Sussex Library]

On: 05 February 2015, At: 10:29

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK

The Journal of General

Psychology
Publication details, including instructions
for authors and subscription information:
http://www.tandfonline.com/loi/vgen20

The Reinforcing Effect of a

Differentiating Stimulus
a b
B. F. Skinner
a
Biological Laboratories, Harvard University
b
Society of Fellows , Harvard University
Published online: 06 Jul 2010.

To cite this article: B. F. Skinner (1936) The Reinforcing Effect of a

Differentiating Stimulus, The Journal of General Psychology, 14:2, 263-278,
DOI: 10.1080/00221309.1936.9713154

To link to this article: http://dx.doi.org/10.1080/00221309.1936.9713154

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of

all the information (the “Content”) contained in the publications
on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as to the
accuracy, completeness, or suitability for any purpose of the Content.
Any opinions and views expressed in this publication are the opinions
and views of the authors, and are not the views of or endorsed by
Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of
information. Taylor and Francis shall not be liable for any losses,
actions, claims, proceedings, demands, costs, expenses, damages,
and other liabilities whatsoever or howsoever caused arising directly
 or indirectly in connection with, in relation to or arising out of the use
of the Content.

This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access
and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [University of Sussex Library] at 10:29 05 February 2015
 THE R E I N F O R C I N G EFFECT OF A D I F F E R E N -
T I A T I N G STIMULUS*
F r o m the Biological Laboratories, Harvard University

B. F. SKINNBR~

T h e present paper continues a series in which the investigation

of a selected example of behavior has been reported. T h e example
is a chain of reflexes constructed in the following- way. ( S b R I )
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

is a reflex of long standing in the history of the organism, in which

the stimulus SI is composed of any or all of the energies arising from
a bit of food and the response RI is the seizing of the food. Subse-
quent members of the chain (which bring about the ingestion of the
food) do not concern us. A preceding member is attached as
follows :
S1I-D" + SLR',
where SKI is the stimulation supplied by a food tray and R" is the
approach to the tray in any way that will bring $1 into action. A
third member could be attached by allowing its response to produce
the whole of SII. In the present case a discrimination (2) is first
set up, so that Rn is evoked only in the presence of a differentiating
property or component. This is done by reinforcing Rn only when
SI1 includes the sound of a food magazine (D).Another member
is then added by allowing its response to produce D alone. In the
present case the third member is the pressing downward of a small
lever. T h e chain then stands
SIILRUI i, S L R I I + S L R I
D
where SII1is the stimulation supplied by the lever, RII1 is pressing
the lever, and where the first arrow is understood to connect RIn
with D only.
T h e present paper is concerned with adding another member to
this chain. Its purpose is not to see how far chaining may be carried
'Society of Fellows, Harvard University.
*Accepted for publication by Carl Murchison of the Editorial Board and
received in the Editorial Office, January 17, 1935.
263
 264 B. F. S K I N N E R

out, but to observe the rate at which the new member is added.
W h e n ( S1ll-R1ll) is conditioned there are several disturbing fac-
tors present, chiefly incidental effects of the new lever, which make
its study difficult (1, 3 ) . It is possible to avoid some of these in
attaching a fourth member by using the same kind of response and
especially by using the relatively stable method of periodic recondi-
tioning (2).
A discrimination is first established by reinforcing every response
in a stimulus complex which includes the lever and a differentiating
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

stimulus L and by extinguishing all other responses to the lever in

the absence of L (2, 5, 7 ) . It is then necessary to set up a situation
in which some arbitrarily chosen response will lead to the presenta-
tion of L, after which the discriminative response to the lever plus L
(SAB. . L f . ) is immediately reinforced. T h i s could be done by put-
ting another lever in the experimental box or by using some other
kind of response, but the result for another lever would be unclear
because of induction (5, 6 ) , and it would be hard to obtain any
response that was different enough to be certainly free of this com-
plication. T h e present procedure has been to use the same reflex.
T h e experiment thus represents only one case-that in which the
interference between reflexes I11 and IV will be maximal. A first
response to the lever produces L ; a second response in the presence
of L produces the sound of the magazine; and the response to the
tray plus the sound of the magazine leads to food. T h e chain may
be written
S I V -R I V + 8111 - R I I I + 811 - R I I + 81 -RI
AB . AB.. L . . D
T h e following method, which utilizes the relatively stable state of
the rat under periodic reconditioning and can best be described by
turning to an actual experiment. A group of twelve white rats, from
which seven complete series were obtained,2 approximately 140 days

'In view of the delicacy of this method w e are not justified in assuming
the irrelevance of even slight technical variations, and it is therefore diffi-
cult to bring a perfect series to completion. In the present case three series
were eliminated when the differentiating stimulus w a s accidentally left
on continuously for a f e w minutes during the first day of the discrimina-
tion. In one other case L remained absent during the last day of the dis-
crimination because of a fault in the apparatus, and the rate returned to
that characteristic of periodic reconditioning. T h e records in these four
 REINFORCING EFFECT OF DIFFERENTIATING STIMULUS 265

old at the beginning of the experiment, were put on the usual sched-
ule of daily experimental periods of one hour each, during which
responses to the lever were reinforced every five minutes (for this
method see 2). All the rats assumed an approximately constant rate
of responding, in agreement with previous work. After two days
of periodic reconditioning a discrimination was begun by sounding a
buzzer (as the differentiating stimulus) whenever the next response
was to be reinforced. After four daily hours the rate of responding
in the absence of the buzz had fallen to less than half its original
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

value. On the next day the order of events was as follows. When
the rat was released, the buzzer was off. T h e first response turned
it on and at the same time completed the circuit to the food maga-
zine, making it ready for the next response. T h e second response,
in the presence of the buzz, was followed by a discharge of a
pellet of food into a tray where it was accessible to the rat. T h e
buzzer was then turned off and the magazine disconnected. Responses
during the next five minutes were ineffective. At the end of five
minutes the apparatus was again set (without supplying any stimu-
lus to the rat) so that the next response turned on the buzzer and
the next after that discharged a pellet of food. T h e light and
magazine were then turned off again. T h i s was repeated at five-
minute intervals during the hour. All responses to the lever were
recorded.
After the discrimination had been begun, all the responses to the
lever which produced food were in the presence of the buzz. Any
inductive effect upon the responses in silence was weak or lacking.
T h e responses still observed during the intervals in silence were,
according to previous work, due largely to the original periodic re-
conditioning (2, 5, 7 ) . These should have continued to disappear
as the discrimination curve was carried out if the newly arranged
correlation of one response every five minutes with the production
of the buzz were having no effect. T h e experimental result, how-
ever, is that immediately upon changing to the new condition the
rate of responding in silence increases until the original rate is
reached, or at least approximately.

series are entirely compatible with the present conclusion but cannot well
be included in an average. In the remaining case the rat developed a
labyrinth disorder occasionally observed in this strain.
 266 B. F. SKINNER

U
3
0
r
U
W
a
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

cn
W
cn
z
Bcn
W
oc
DISCRIMINATION p-L

1 2 3 4 5 6 7 8 1 9
1 1 l 1 1 1 1 1 1

DA I L Y P ER I 0DS
FIGURE 1.
A. T h e average rate of responding for seven rats, established at a
value of about 200 responses per hour under periodic reconditioning, falls
during four daily hours of discrimination to less than 100 responses per
hour. When undifferentiated responses begin to produce the differentiating
stimulus ( L ) , the rate rises (Days 5 , 6, and 7), but to a value somewhat
lower than that observed upon returning to the original procedure of periodic
reconditioning in the continued absence of L (days 8 and 9 ) .
B.C.D. Averages for groups of eight, four, and four rats, respectively,
showing the last two days of a discrimination'and three days on which
undifferentiated responses produce the differentiating stimulus.

T h i s is shown in Curve A, Figure 1, which gives the average for

the seven completed series. I n this figure no attempt is made to
follow the rate during each hour. T h e rates have been plotted as
horizontal bars as if they were constant. I n order to indicate the
course of the change from day to day, smoothed curves have been
drawn to the center of these bars. It will be seen from this figure
that the average rate of responding falls during discrimination from
202 responses per hour a t the beginning of the first day to 89 per
hour on the fourth day. On the fifth day, when responses to S A B . .
 REINFORCING EFFECT OF DIFFERENTIATING STIMULUS 267

now produce the differentiating stimulus periodically, the rate in-

creases to 121 on the first day and to 163 on the second. T h e rate
drops to 143 on the third day, indicating that 163 may be a com-
pensation for the low value of 121 on the first day.
T h e average for the three days is 142, significantly below the pre-
vious rate of 201 for periodic reconditioning. We expect some
decline in this rate with time ( 2 ) but by returning to the procedure
of periodic reconditioning it can be shown that the periodic rate
has not fully been attained. Thus, on the eighth day L was omitted
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

altogether; periodically responses to SaB . . now produced, not L,

but D. T h e resulting rates on two successive days are 173 and 161
or an average of 167, which still shows a (more reasonable) decline
in rate during the 16 or 18 (chronological) days of the experiment.
T h e difference between 142 and 167 is, however, subject to a cor-
rection that will be made in the Appendix.
Another group of eight rats approximately 160 days old had
been through a very extensive series of reversed discriminations in
which the extra stimulus was a light, but in which the actual differ-
entiation had come to be based upon neither the presence nor the
absence of the light but the change from one to the other in either
direction (7). T h e average rates on the last two days of the series
were 96 and 83 respectively. T h e rate was not falling as rapidly
as this single difference would indicate. T h e procedure was then
changed so that responses in the dark periodically produced the
light, after which a response was reinforced. T h e increase in rate
was in this case practically complete on the first day, as shown in
Figure 1, Curve B. No recent periodic rates were available for
comparison, nor was the procedure changed to periodic recondition-
ing with this group.
A third group of four rats approximately 170 days old had estab-
lished a discrimination to the light without previous periodic recon-
ditioning, the data for which have been reported (7). O n the third
and fourth days the average rates were 44 and 42 respectively.
On establishing the relation of SIB to L the rate increased to 120,
133, and 139 responses per hour, as shown in Curve C, Figure 1.
A fourth group of four rats, approximately 150 days old, had estab-
lished a discrimination without previous conditioning ( 5 ) , in which
the differentiating component was the absence of a light. O n the
 268 B. F. SKINNER

fifth and sixth days of the discrimination, the average rates were 24
and 25 responses per hour respectively, the slight increase being
insignificant. Upon changing the procedure (so that periodically
a response in the presence of the light resulted in its disappearance
and the next response was reinforced) the rate increased to 61 re-
sponses per hour, dropping to 50 and 49 on the following days. T h e
unusually low values for these animals and the relatively slight in-
crease are partly due to the method of discrimination, but another
important factor is described in the Appendix.
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

T h e important point in such a series is the first day of the rela-

tion of R I V and L. T o express the matter in the vernacular: “the
rat is in the course of learning to press the lever only when L is
present; it must now learn that it is itself producing L.” T h e result
appears in the present analysis as a change from the low value of
(SA-R) obtaining under the partially completed discrimination
to that obtaining under periodic reinforcement by the differentiating
stimulus. I t is clear from our data for the first day that the change
begins very soon, but w e need to turn to the daily records them-
selves for its actual course. These records are obtained auto-
matically in the form of number-time graphs. W e are interested
in any increase shown on the day of the change over the rate that
would be expected from the continuation of the discrimination.
T h e latter could be calculated by extrapolation from the preced-
ing part of the curve, but for our present purposes we mav suppose
that the rate to be expected on the day of the change is identical
with that of the preceding day. Any error introduced by this
assumption works against the argument of the paper. If we sub-
tract this expected rate from the rate actually observed, any increase
as the effect of L will be clearly shown. I n practice, measurements
of the height of the curves were made at intervals of ten minutes
during the hour. Averages of these points on the last days of the
discrimination were subtracted from those for the first days of the
change. T h e results are given in Figure 2. Here the base line is
in each case the average summation curve for the responses of the
group on the preceding day. T h e curves show the increases above
this base line as the result of reinforcement by L.
I t is obvious that the effect is immediately felt. I n all four
groups a significant increase is evident at the end of ten minutes,
 REINFORCING EFFECT OF DIFFERENTIATING STIMULUS 269

RESPONSES I N EXCESS
OF THOSE OBSERVED
ON PRECEDING DAY

W
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

MINUTES
FIGURE 2.
Data for the groups in Figure 1, showing increases on the first day of the
change over the responses observed on the last day of the discrimination
(Day 5 in Figure 1 minus Day 4 ) .

when only two reinforcements have occurred. I n Group A this

increase represents very nearly the full value : successive periods
each add about the same number of responses. I t should be noted,
however, from Figure 1 , that the average value for this group for
the first day of the change is low, and that a further increase in
rate occurs on the following day. I n the other three groups a sig-
nificant increase is apparent, but it is followed sooner or later by a
more marked acceleration. I n Group B the greatest increase is
observed between 10 and 20 minutes after the beginning of the
hour; in Group C between 40 and 50 minutes; and in Group D
between 20 and 30 minutes. These differences probably reflect the
various histories of the four groups. Group A began with some
periodic reconditioning and its only discrimination had not been
carried very far. Group B had a very thoroughly established dis-
crimination. Group C had had no previous periodic reconditioning,
which may account for its delayed acceleration. There are special
reasons for the low values in Group D, which will be noted in the
Appendix. A typical series of actual records is reproduced in Fig-
ure 3.
 270 B. F. SKINNER

T
0
-
0

i
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

v)
W
v)
Z
0
a
v)
W
a

I I
TIME IN MINUTES

FIGURE 3
A typical series of daily records for one rat in Group A, Figures 1 and 2.
In records 1, 2, 3, and 4 the rate declines during the establishment of the
discrimination. I n records 5 , 6, and 7 responses in the absence of the
differentiating stimulus periodically produce the stimulus and the rate rises.
In records 8 and 9 there is a still further increase when the procedure is
restored to that of periodic reconditioning. T h e increase is partly due to
the fact that with this r a t the differentiating stimulus was the absence of the
sound of a buzzer; consequently the buzzer w a s present almost continuously
on Days 5, 6 , and 7, and significantly depressed the rate (see Appendix).

These four experiments show, without exception, that the produc-

tion of a differentiating stimulus has a marked reinforcing effect and
that this is felt immediately. It is not possible, however, to estimate
the value of the effect very closely from the present experiment. It
is probably of the same order as the effect of the sound of the
magazine, but a mere comparison of the two periodic rates will not
yield a fair estimate for the following reasons:
 REINFORCING EFFECT OF DIFFERENTIATING STIMULUS 271
1. Since L is produced simultaneously with RIV and since R’”
follows immediately (within two seconds), SO follows RN closely
enough to have a considerable reinforcing effect upon it. Under the
conditions indicated by ‘RIV + L” it can be said that once every five
minutes a response to S A B . . is followed within two seconds by 80”
(no attention being paid to the intervening events : RN + StV- R”) .
As shown elsewhere (8), an interval of two seconds between a
response and the reinforcing stimulus reduces the effect of the rein-
forcement by not more than about 30 per cent. Consequently part
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

of the periodic rate developed under “RN + L” may be due not

to L but to D.
2. W e also do not know whether the rate remaining as part of
the discrimination must be supposed to be added to the periodic rate
under “RIV + L.” W e do not know whether “pressing the lever
in order to turn on the light” is the same reflex as “pressing the lever
in order to make the magazine sound.” If we must regard them as
discrete entities, their separate rates of elicitation must be supposed
to summate in our records. If the presentation of L has a reinforcing
effect equal to that of D, [SIV - RIV] should on its own account
reach a value nearly equal to that observed at the beginning of the
experiment. But [SIII- RIII] is by no means near zero and, on
the assumption that separate reflexes are involved, the total observed
rate should be greater than that originally observed for [ P I - R“]
alone. This is not in agreement with our result but the assumption
is not thereby ruled out. I t may be that the effect of L is even
less than we had supposed.
It is simpler, however, to regard ( S I B . . I I L R I I 1 ) and (SAB..N-
R N ) as the same reflex. W e thereby avoid the systematic question
of how to distinguish them if they are not, and we may easily account
for the present data. T h e return to a higher rate is comparable
with that observed during the abolishment of a discrimination through
reinforcement of the previously extinguished member (4, Figure 2),
where we do not expect summation. Under this interpretation we
may say that the summed effect of L upon (SN-RIV), of D upon
(aw - RIV), and of D upon ( S d B . !I1- P I ) , through induction
from ( S A B . .. -L. .II1R1I1) is approximately equal to that of D upon
(SAB.!ILRnl) directly under periodic reconditioning. But this
does not make possible a close estimate of the effect of L alone.
 272 B. F. SKINNER

T h e experiment throws some light on a technical problem con-

nected with discriminations in general. Where a differentiating
stimulus is repeatedly presented to the organism, its occasional effect
as a reinforcing stimulus must be taken into account. I n the present
method, if a light is presented every five minutes as a differentiating
stimulus, it will have an added effect upon the rate whenever its
presentation coincides with the elicitation of a response. In all ex-
periments up to the present (with the exception of Group A in this
paper) the differentiating stimulus has been set by hand, and the
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

practice has been followed of waiting until the rat is not responding
very rapidly. I n this way coincidences have generally been avoided.
However, in any close analysis of the discrimination curve, a possible
undesired result of this procedure must be considered. Only re-
sponses following a short period of no-responding are ever reinforced ;
consequently a discrimination may be developed that will have for
its effect the elimination of closely grouped responses. Although it
is probable, as will be shown in a later paper, that no effect on the
average rate will be felt, the procedure cannot be regarded as with-
out some special effect. In Group A in this paper the entire ex-
periment was conducted automatically. T h e differentiating stimu-
lus was introduced periodically by a clock, and of course without
respect to the momentary behavior of the rat. T h e discrimination
curves do not differ significantly from those obtained with other
methods, so far as the present degree of approximation is concerned,
but in a closer examination the present result shows quite definitely
that allowance must be made for an occasional coincidence.

APPENDIX
I t will be convenient to report here some experiments on a sep-
arate point which were performed with the same animals and have
some bearing upon the foregoing conclusions. A significant difference
has previously been reported between a discrimination in which L
is the presence of a stimulus and one in which it is the absence.
Where L is a light, it has been shown that its presence depresses
the rate of elicitation ( 7 ) . If the initial periodic reconditioning
occurs in the absence of the light, the introduction of the differen-
tiating stimulus will cause a sudden drop in rate where the absence
of the light is the differentiating stimulus, because during the in-
 REINFORCING EFFECT OF DIFFERENTIATING STIMULUS 273

tervening periods of responding to Saa. .L.. (now the unreinforced

member) the light is on. I n Group A in the present experiments
a control was introduced against the possibility that a discrimination
in which L was the sound of a buzzer differed from one in which
L was the absence of the sound. I n one-half of the cases the dis-
crimination was really the reverse of that described, although this
was not mentioned at the time to avoid confusion. Of the success-
ful series L was the presence of the buzzer in three cases. I n the
other four cases the buzzer was absent during the periodic recondi-
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

tioning only and was therefore present most of the time. T h e curves
of this group have been separated into two parts on this basis in
Figure 4A. T h e resulting groups are small but give some indication
of a difference. T h e horizontal solid lines are for the three rats

I I I C 13 14 1st 6 \ 7 ( 8 IS ~ l O ~ l l ~ l C ~ l 3 ~ l 4 ~ l 5 ~

PER I
D A I LY 0D S
FIGURE 4.
Curves showing the depressive effect of the stimuli used for differentiation.
Explanation in text.
 274 B. F. SKINNER

with which SIB. . L . . was the lever plus the buzzer. T h e broken
lines are for the others. There is no depressive effect on the rate
during the discrimination but the process is retarded; the decline
in rate is slower for the group in which the differentiation is the
removal of the sound of the buzzer. When the relation between
RIV and L is established, a much more significant difference is to be
seen. Where the rat is normally in silence but produces the sound
of the buzzer periodically, the effect of the reinforcement is great.
When the buzzer is sounding continuously, except when it is momen-
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

tarily silenced by a response of the rat, the effect of the reinforce-

ment is slight. Upon changing to simple periodic reconditioning
the buzzers are silenced entirely, and the depressed group responds
with a very significant increase in rate. Because of the low values
for these four rats on Days 5, 6 , or 7 it is impossible to compare the
average rate for the group with the rate on Days 8 and 9 under
periodic reconditioning.
In Group B all eight cases were as described, so that the present
problem does not arise. I n Group C the differentiating stimulus was
the presence of a light; in D, its absence. These experiments corre-
spond very well with those in Group A, Group D showing only a
slight increase when the relation of R" and L is established. Groups
C and D were subsequently tested directly for the effect of the light,
which was found to be relatively great. T h e completed series for
both are given in Figure 4. In the case of Group C! (top curve in
Figure 4) the return to periodic reconditioning (ninth day) yielded
a probably significant increase over the previous rate under reinforce-
ment from L. This group is homogeneous and constitutes the only
very clear evidence in the experiment that the chain
Srll-R1rl . . . etc.
is elicited somewhat more rapidly than
P-RN . . . etc.
During this simple periodic reinforcement the light was off. On
the 11th day it was turned on continuously beginning at 25 min-
utes. No other condition was changed. Since the light is at this
point still a differentiating stimulus, and since responses in the
presence of the light are now no longer reinforced, except periodically,
an extinction curve follows. A typical record is shown in Figure 5,
 REINFORCING EFFECT OF DIFFERENTIATING STIMULUS 275

Curve C. T h e convexity at the beginning of this curve is of the

sort that can be expected as an incidental effect under periodic
reconditioning. T h e horizontal line through the curve marks the
beginning of the extinction curve for (SIB. .L..-I?). T h e curve
falls rapidly as the depressive effect of the light emerges over its
excitatory character as a differentiating stimulus. T h e average rate
for this day, including the extinction curve, is already lower than that
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

I r 1
T I M E
FIGURE 5.
Typical daily records showing the depressive effect of a light. Curve C
shows the beginning of a record for periodic reconditioning in the absence
of the light. A t the horizontal line the light is turned on. Its first effect
is as a differentiating stimulus, yielding the extinction curve for (SAB. .L.
- R ) . T h e depressive effect emerges, however, and the final periodic slope
(now in the presence of the light) is much less than that a t the beginning.
Curve C' is for the same rat on the second day following. It begins at the
depressed slope. At the vertical bar the light is turned off, and the rate
immediately changes to a higher value.
In the case of curve D the nbsence of the light is the differentiating stim-
ulus. T h e record begins with the light on; its removal results in an ex-
tinction curve, as in C, but the subsequent rate shows an increase due to
the removal of the depressing stimulus. Record D' is for the third day
following. At the beginning the light is off; at the bar it is turned on,
and the rate immediately falla.
 276 B. F. SKINNER

of the previous day, and on the following day an unusually low value
is maintained (record omitted in Figure 5 ) . O n the next day the
light is turned off after twenty minutes. T h e rate again rises, al-
though, for a reason that is not clear, it does not quite return to its
former high value. I n Figure 5, Curve C‘, the vertical bar marks
the change to “light off.’’ T h e average for the group, plotted again
as rate rather than as number vs. time, is given in Figure 4.
A n important characteristic of the record C’ in Figure 5 is the
complete absence of any compensatory increase in rate following re-
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

moval of the depressive stimulus. T h e depression is clearly dif-

ferent from cases previously reported where any tendency to sup-
press the rate is followed by marked compensation.
I n the case of Group D (middle curve in Figure 4) the light is
on during the simple periodic reconditioning (tenth day on the
graph), and the rate shows no increase as the result of changing
from RIV-L. O n the following day it is turned off after 25 min-
utes. An extinction curve follows for the same reason as in the
case of Group C. Here, however, it leads to a greatly increased
rate, a typical example of which is given in Figure 5, curve D. After
two days of this rate, (omitted in Figure 5 ) the light is again
turned on and the rate falls to its previous low level. I n the particu-
lar case of curve D’ (Figure 5 ) the effect is especially great imme-
diately after the light is presented (at the vertical bar). Some
slight recovery is evident toward the end of the hour.

SUMMARY
A discrimination is set up in which a response to a lever is elicited
practically only when a differentiating stimulus ( L ) is present. An
occasional response in the absence of L is then arranged to produce
the appearance of L periodically. T h i s has the effect of recondition-
ing the response in the absence of L. T h e rate of change is ex-
amined through its effect upon the rate of elicitation observed under
periodic reconditioning. T h e change is found to begin immediately.
Some later acceleration is usually observed.
A related experiment demonstrates that a differentiating stimulus
such as a buzz or light may have a marked depressive effect upon the
rate of elicitation during periodic reconditioning.
 REINFORCING EFFECT OF DIFFERENTIATING STIMULUS 277

E N cE s
REFER
1. SKINNER, B. F. On the rate of formation of a conditioned reflex. J. Gen.
Psychol., 1932, 7 , 274-286.
.
2. - T h e rate of establishment of a discrimination. J. Gen.
Psychol., 1933, 9, 302-350.
3. “Resistance to extinction” in the process of conditioning.
J . Gen. ‘Psychol., 1933, 9, 420-429.
4. - . T h e abolishment of a discrimination. Proc. Nut. Arud.
Sci., 1933, 19, 825-828.
5. . A discrimination without previous conditioning. Proc.
Nut. Acad. Sci., 1934, 20, 532-536.
Downloaded by [University of Sussex Library] at 10:29 05 February 2015

6. .T h e generic nature of the concepts of stimulus and response.

1. Gen. Psychol., 1935, 12, 40-65.
.
7. - A discrimination based upon a change in the properties of
a stimulus. J. Gen. Psychol., 1935, 12, 313-336.
8. -. T h e effect on the amount of conditioning of an interval of
time before reinforcement. J . Gen. PsyclroI., 1936, 14, 279-295.
Harvard Uniwersity
Cambridge, Massachusetts

L’EFFET RENFORCANT D’UN STIMULUS DIFFERENCIANT

(RCsumC)
Une discrimination est Ctablie oh une riponse ?I un levier ne se montre
presque que quand un stimulus difftrenciant ( L ) est prCsent. Une rCponse
occasionnelle en I’absence de L est ensuite faite pour produire l’apparence
de L. Ceci a I’effet de rendre la riponse de nouveau conditionnelle en
l’absence de L. On examine la vitesse du changement dans son effet sur la
vitesse de la mise en action observCe dans le rtconditionnement pCriodique.
On constate que le changement commence immtdiatement. On note or-
dinairement quelque accClCration plus tard.
Une expCrience ayant rapport 4. celle-ci montre qu’un stimulus diffkren-
ciant tel qu’une sonnette bourdonnante ou une lumitre peut avoir un effet
dtpressif sur la vitesse de la mise en action pendant le reconditionnement
pkriodique.
SKINNER

D I E VERSTARKUNGSWIRKUNG DES DIFFERENZIRENDEN

REIZES
(Referat)
Eine Unterscheidungsaufgabe wurde gegeben, in der eine Antwort mit
einern Hebel nur dann gemacht wurde, wenn ein differenzierender Reiz
(L) vorhanden sei. Eine gelegentliche Antwort bei der Abwesenheit von L
wird dann so gemacht, dass er L erzeugt. Dies hat als Folge die Wieder-
 278 B. F. SKINNER

bedingung der Antwort in der Abwesenheit von L. Die Schnelligkeit der

Veranderung wird unter periodischer Wiederbedingung beobachtet. Es
wurde festgestellt, dass die Veranderung sofort einsetzt. Einige spatere
Beschleunigung wird gewohnlich bemerkt.
Ein anhliches Experiment beweist, dass ein differenzierender Reiz, wie
ein Summen oder ein Licht eine ausgesprochene depressive Wirkung auf
die Schnelligkeit der Hervorrufung wahrend der periodischer Wiederbe-
dingung haben mag.
SKINNER
Downloaded by [University of Sussex Library] at 10:29 05 February 2015