shaping was required. The first experimental session began wi th the first lever press. The other two rats in Group A did not press the lever within the first 50 shaping trials, but did so soon after the beginning of the next training session, whereupon the first experimental session began.

The animals in Group C were trained and tested in Chamber 2 to emit the sequence trapeze puJ1 -+ lever press in darkness. After magazine training in low illumination, and before the lever was introduced, food delivery was made contingmt upon exploration of the hole in the function panel through which the lever was to project. When the animals spent most of the time at the hole, the lever and white noise were introduced, the latter gradually, and the chamber was subsequentlymaintained in darkness. Each rat pressed the lever as soon as it was introduced. The signal was then removed for increasingly longer periods, in the manner of the procedure for the other groups, except that it was not possible to avoid offsetting the signal when the animal seemed likely to press because the rats were invisible. The criterion of control used for Groups A and B was recognized during the third session of discrimination training, whereupon the trapeze was introduced at the beginning of the foJ1owing day's session. Onset of the signal now became contingent upon a rat's being successively doser to the trapeze, then touching it, then holding it, and then pulling it. The approximations to pulling the trapeze were assessed using the proximity detector. Each rat in Group C pulled the trapeze within 50 shaping trials and the Hrst experimental session began with the Hrst trapeze puB.

There were 10 experimental sessions, each of 100 trials. A regular trial consisted of aperiod without the signal, when emitting the Hrst operant produced the signal, and a period with the signal, when emitting the second operant produced food, offset the signal, and initiated the next trial. Errors, i.e., emitting one operant when the other was appropriate, had no prograrnmed effect. During a test trial the signal was not presented as a consequence of emitting the first operant in its absence, but reinforcement was still contingent upon emission of the sequence first operant -+

second operant.Sessions 1,3,6,and lOeach consisted of 80 regular trials and 20 test trials. The test trials occurred in four blocks of five, beginning with the 21st, 41st, 6lst, and 81st trials. The other six sessions consisted of 100 regular trials.

RESULTS AND DISCUSSION If a rat's behavior was not under the

control of the signal and its absence, then the behavior would be unaffected by the unavailability of the signal during test trials. On the other hand, behavior under

48

discriminative control would be affected.ln particular, because during a test trial only the occasion for emitting the first operant occurred, a rat whose behavior was under diseriminative control might have been expected to continue emitting the first operant when emission of the second operant would have been followed by food delivery. Proportions of trials during which errors involving the Hrst operant were made, during test trials and normal trials, are shown in Fig. 1. Errors foJlowing an operant by less than 0.5 sec are not represen ted. The figure shows that the three rats in Group B, and most rats in Groups A and C, were more likely to emit errors involving the first operant during the test trials of Session I than du ring the regular trials. The animals in Group B continued to exhibit this difference between regular and test trials whereas the other animals did not.

Thus the results of the earlier study were replicated in a situation in wh ich the auditory signal was relatively more prominent, and in which the auditory signal was initiaJly associated with the controlof the chained behavior in at least half of the possible animals. In addition, auditory control did not persist in darkness, indicating that behavior initially under auditory control does not necessarily come under visual control with practice.

A significant feature of the data was that, even when control was indicated by a greater proportion of first-operandum errors during test trials, these errors were often not made

on at least half 01' the test trials, suggesting other sourees of control. To some extent this was because first-operandum errors became less likely towards the end of each bloek of five, suggesting blunting of control within blocks. Such blun ting is supportcd by the distribution within blocks ()r errors involving the second operandum; these errors occurred mostly towards the end of a test block, a eonsequence of the reinforcement of the second operant in the absence of the signal.

Blun ting or not, i tremains true tha t, even considering only the first trial of each test block, the rats in Group B were not always affected by omission of thc visual signal. The visual control could have been sporadic, and shared with other sources. Alternatively, the behavior may not have been under visual control and yet may have been sometimes disrupted by the nonappearance of an event which usually occurred, namely, the onset of the visual signal. The basis of the difference between the auditory and visual signals remains obseure.

REFERENCES GILBERT R., & MOORE, M. Auditory and visuaI

controI over chained operants. Psychonomic Science, 1967, 8,267-268.

MUNN, N. L. Handbookofpsychological research on the rat. Boston: Houghton Mifflin, 1950.

NOTE I. This study was carried out at the University

of Aberdeen, Scotland, during the Iater part of 1967. It was supported in part by agrantfrom the British Medical Research Council.

Noncontlngent partial relnforcement reduces spontaneous alternation 1

!OEL ADKINS, ROBERT K. HlLLES, DA VID S. WEISBROD, 2 and MICHAEL R. EMMENS, University of Oregon, Eugene, Oreg.97403

Two groups of 12 rats were tested for spontaneous alternation in a T maze. Ss rewarded with food after each trial

altemated goal arms significantly more often (p = .01/) than Ss lor which re ward was withheld randomly on one-hallof the trials regardless of the animal's response. A possible relation between the effects of partial reinforcement and 01 brain damage in the limbic system or frontal cortex ~s discussed

Psychon. Sei., 1969, Vol. 17 (1)

It is well established that partial reinforcement increases the resistance to extinction of instrumentally conditioned responses (cf. Hall, 1966). Increased resistance to extinction also is one of a constellation of perseverative response tendendes observed in animals following damage to several specific areas of the brain, particularly parts of the limbic system (Douglas, 1967; Douglas & Pribram, 1966; Kimble, 1968; McCleary, 1966) and the frontal cortex (Brutkowski, 1964; Mishkin, 1964). A second symptom ofbrain damage in Iimbic and frontal areas is a decrease in the spontaneous alternation of left and right choices in aT maze (Kimble, 1968; Morgan & Wood, 1943). The present study wasdone to determine whether the parallel between partial reinforcement and brain damage in increasing resistance to extinction also extends 10 spontaneous alternation. If so, then parlial reinforcement should result in a decrease in alternation.

METHOD The Ss were 24 male albino rats from

Rush Laboratories, Beaverton, Oregon. They were 90-99 days old at the start ofthe experimen t.

All Ss were adapted to a 22-h food-deprivation schedule for 7 days. They were then divided into two groups of 12 and tested for spontaneous alternation.

The T maze was painted a medium gray, had a wooden floor, and was of approximately the following dimensions: stern and goal alleys, 2 ft;height, 6 in.; alley width, 5 in. There were Plexiglas sliding doors between a starting box and the stern and between the choice point and the goal arms.

On each trial, S was placed in the start box for 10 sec and then all the doors were raised. When S's whole body had entered a goal arm, the door was lowered behind it. S was left in the goal box for about 20 sec and then removed by hand to the start box for the nex I trial. All Ss were given four trials a day for 15 days. Ss in one group were rewarded following each trial with eight 45-mg food pellets, regardless oftheir choice ofthe right or lefl goal arm (continuous reinforcement-Group CR). For the second group, reward also was independent of right or left choices, but the food was withheld on two randomly chosen trials of the daily four (partial reinforcement-Group PR). Ss earned daily al ternation scores of 0, I, 2, or 3.

F ollowing each testing session, all the Ss were allowed to eat in their horne cages for

Psychon. Sei., 1969, Vol. 17 (1)

Table I Number of Alternations

U nderlined Scores are for Ss in Group PR

36 31 26 23

36 30 25 23

35 29 25 23

33 28 24 22

33 26 23 21

31 26 23 21

2 h. Water was always freely available.

RESULTS The Ss in Group CR alternated 352 times

out of 540 possible alternations, for a rate of 65.2%. Ss in Group PR alternated 290 times out of 540, for a rate of 53.7%. A summed-ranks test showed that the difference between groups was statistically significant (Z = 2.29; P = .011). Table 1 shows the number of alternations (out of 45 possible) for each animal, arranged in rank order.

A pilot study done earlier and with fewer animals found a difference in the same direction and of about the same size as reported above; continuously reinforced animals alternated 78.7%, while partially reinforced animals alternated 67.4% in the pilot study.

DISCUSSION The results show that spontaneous

alternation is greater during continuous reinforcement than du ring 50% reinforcement, in the absence of any contingency between response and reward. Therefore, it appears that the similarity, mentioned in the introduction, between partial reinforcement and the effects of certain kinds of brain damage on resistance to extinction also holds for the effects of the two variables on spontaneous alternation. The results of the present study are consistent with those reported by Walker (I956), who found that thirsty rats alternated more when continuously reinforced with water than when completely unrewarded. He did not study partial reinforcement, however.

The results reported here do not prove, of course, that the decrease in spontaneous alternation is caused by changes in the functioning of neural mechanisms similar to those that follow lesions of the frontal cortex or oflimbic structures. However, the data are consistent with any of several published interpretations of Iimbic system

damage. Douglas (1967) has proposed an "attention shift" model of hippocampal functioning, for example, while Kimble (1968) has discussed the hippocampus as part of a neural system underlying Pavlovian internal inhibition. McCleary (1966) has attributed the effects of several other types of Iimbic system damage to the defective operation of opposing mechanisms of response facilitation and inhibition. These interpretations have in common at least their agreement that the deficits are due primarily to performance variables rather than to speeific changes in learning capacity , for example.

The parallel discussed in the present paper is merely suggestive. Further research is required before the relation between damage to neural systems and experimental manipulations of performance variables can be assessed. Studies of the effects of motivational variables on untrained response tendencies are particularly needed.

REFERENCES BRUTKOWSKI, S. Prefrontal cortex and drive

inhibition. In J. M. Warren and K. Akert (Eds.), The frontal granular cortex and behavior. New York: McGraw-Hill, 1964. Pp. 242-270.

OOUGLAS, R. J. The hippocampus and behavior. Psychological Bulletin, 1967,67,416-442.

DOUGLAS, R. J., & PRIBRAM, K. H. Learning and limbic lesions. Neuropsychologia, 1966,4, 197-220.

HALL, J. F. The psychology of leaming. Philadelphia: Lippincott, 1966.

KIMBLE, D. P. Hippocampus and intemal inhibition. Psychological Bulletin, 1968, 70, 285-295.

McCLEARY, R. A. Response-modulating functions of the limbic system: Initiation and suppression. In E. Stellar and J. M. Sprague (Eds.), Progress in physiological psychology. Vol. l. New York: Academic Press, 1966. Pp. 210·266.

MISHKIN, M. Perseveration of central sets after frontalIesions in monkeys. In J. M. Warren and K. Akert (Eds.), The frontal granular cortex and behavior. New York: McGraw-Hill, 1964. Pp. 219-241.

MORGAN, C. T., & WOOD, W. M. Cortical localization of symbolic processes in the rat. n. Effect of cortical legions upon delayed alternation in the rat. Journal of Neurophysiology, 1943,6,173-180.

WALKER, E. L. The duration and course of the reaction decrement and the influence of reward. Journal of Comparative & Physiological Psychology, 1956,49,167-176.

NOTE 1. This work was supported by a grant to Joel

Adkins from the University of Oregon Office of Scientific and Scholarly Research.

2. Present address: Department of Psychology, Trinity University, San Antonio, Tex. 78212.

49