You are on page 1of 3


Harvard University

Under many schedules of reinforcement running.) Because of its low inertia the wheel could be
there are significant periods of time during stopped and started instantly; hence a change could
readily be made from one form of behavior to the other.
which the organism does not display the be- Responses to the lever continued to be reinforced on
havior under investigation. For example, on FI5; no reinforcement was explicitly contingent upon
a fixed-interval schedule there is commonly running.
a period of no responding after reinforcement.
The behavior which actually occurs at such RESULTS
times is apparently unrelated to any explicit The first performance when the wheel was
reinforcement, yet the schedule is at least unlocked is shown for one rat in Fig. 1. Seg-
effective in controlling the time of its occur- ments A and A' comprise a continuous record
rence. Interesting examples are found in the of the movement of the wheel. Segments B
field of psychotic behavior, where the patient and B' show the concurrent activity on the
engages in compulsive or other idiosyncratic lever. Reinforcements are marked by small
ways only when he is not executing the be- hatches. Reinforcement a in Record A is the
havior under the control of a given schedule. same event as that marked at a in Record B,
What the organism is doing when it is not and the same correspondence is maintained
showing the behavior produced by a schedule throughout the figure. It will be seen that the
of reinforcement is especially important movable wheel occupies the rat for nearly 20
when we come to set up complex behavior in min. (beginning of Record A), during which
which two or more responses are studied at the no responses were made to the lever (beginning
same time. of Record B). A first response to the lever is
METHOD reinforced at a. The rat continues to show a
Unreinforced behavior accompanying the perform- high rate of running on the wheel between
ance generated by a common schedule of reinforce- reinforcements a and c in Records A and A1,
ment was studied in the following way. A standard with corresponding low activity on the lever
rat lever and food magazine were installed alongside shown between reinforcements a and c in Rec-
the open face of an aluminum running wheel. The wheel ord B. A few responses are made to the lever,
was 13 in. in diameter, had a low moment of inertia,
and turned in only one direction against friction which however, and reinforcements are received ap-
was just overcome by a tangential force of 30 gm. proximately on schedule.
Through a gear-reduction and pulley system the rota- Before the end of the session the over-all
tion of the wheel advanced a pen across the moving rate of running has fallen off considerably
paper tape of a modified cumulative recorder. Scales
were chosen so that the commonest speed of running (end of A'), and a fairly normal fixed-interval
produced a slope in the neighborhood of 45°. Responses performance has been restored (end of B').
to the lever operated a standard cumulative recorder. Running during each interval already begins
Two rats were conditioned to respond to the opera- to show a temporal pattern, which became
tion of the magazine until they quickly seized and ate
the .1-gm. pellet of food delivered. At this stage the
consistent in later sessions. After reinforce-
wheel was immovable. After three sessions of magazine ment the rat neither runs nor presses the lever
training all responses to the lever were reinforced for for a short period of time, which is much longer
several sessions. A fixed-interval schedule was then put than the time required to eat the pellet of food.
into effect, under which a reinforcement was set up It then runs actively for distances of the order
every 5 min. (FIS) during a session of 3 hr. and 20
min. Both rats developed marked "scallops" during of 100 to 200 ft. before returning to the lever
the first session, and the fixed-interval performance to resume a fairly standard-interval perform-
was allowed to develop during several sessions until ance. After the reinforcement marked n,
the pause after reinforcement was well marked. At the for example, a short horizontal section appears
beginning of the ninth experimental session on FIS the in both records. There is then a short burst of
wheel was unlocked. The rat could now respond to the
lever or run in the wheel. (It could not do both at the running in the wheel, marked by the vertical
same time because it faced away from the lever when dotted lines in A'. The corresponding perio

2. The light was then turned off. At the beginning of each session the rat had access to the wheel for approximately J£ hr. The curve at B shows the last five segments of the fixed-interval performance prevailing at this stage of the experiment. during which 20 responses were reinforced on FI5. the following program was instituted. 1 is shown in Fig. F. lamp.. The fixed-interval segments in B are now relatively "square" and suggest FIG. 3 that responding is being postponed by the . The light was then turned on for 100 min. In order to obtain a collateral measure of the rate of unreinforced running. Running in the wheel ceases during the second half of the interval.. previously used for general illumination in the apparatus. 1 in the lever record is indicated in B'. Corresponding portions are connected by broken lines. The end of such a session for the rat in Fig. was off during this period. The segments at A and . and further running in the absence of responding to the lever was re- FIG. A 6-w. leading to the reinforce- ment at o. SKINNER AND \\.280 B. during which no re- sponses to the lever were reinforced.4' com- prise a continuous record of the performance in the wheel during the five intervals and. H. be- ginning at a. MOUSE FIG. 2 corded. and responses to the lever were quickly ex- tinguished in the absence of the light. after the light has been turned off and responding to the lever has ceased. when fairly constant responding on the lever is shown.

a reinforcement. both than in the case of the other rat.. . Since responding on the wheel falls off to The beginning of a daily session for the other zero or very nearly zero late in each interval. occasionally shown by the first rat. 2 was characteristic. at d and e. 3. For a substantial sponding positive curvature which is absent in period of time after each reinforcement. CONCURRENT ACTIVITY UNDER FIXED-INTERVAL REINFORCEMENT 281 activity in the wheel. 2 is shown in few if any adventitious reinforcements of Fig. We may therefore absence of the light. running in the wheel is greatly suppressed running wheel or to press a lever for food on a (compare the segments preceding b and c with fixed-interval schedule. the resolution of the Record /I). When pressing begins. however. When the schedule does not generate then follows. lever in a somewhat more continuous fashion Shortly after reinforcement. although Sustained running then takes place for 2 or the more rapid shift from wheel to lever seen 3 min. The rate of behaviors are absent. rat at the same stage as Fig. running falls off in a particularly smooth curve Received March 12. not all of which is be expressed in the following way. although there is little SUMMARY activity on the lever during the first two inter. This yields to responding to the a substantial rate. Note that. uniform rate is concurrent activity on the lever have a corre- assumed until reinforcement. when responses to the suppose that activity in the wheel remains lever are not reinforced. yielding to the behavior on the lever in Fig. Segment A and the first part of A' show running could occur as the result of quick the relatively fast spontaneous running in the shifts from wheel to lever. the Fig. for the rest of the interval. 1956. a fairly stable. running in the wheel is sup- to either lever or wheel. A continuous change as in Fig. running in the wheel occurs. performance appropriate to FI5 is begun (Record E). The segments of Record B showing however. When the shown in the figure. A burst of running pressed. 3 was lever is not pressed and running does not occur. For a short period of time schedule normally generates a substantial rate after reinforcement the rat does not respond of responding. A stable performance persists for competition between running and pressing can the balance of the session. When a rat is free to run in a low-inertia vals. 2. When a first response essentially unconditioned with respect to food is reinforced in the presence of the light at a.